# Newsroom — gaganmalik.io

Last-Updated: 2026-07-13

All WRITINGS essays include full text; UPDATE posts include share descriptions with URLs.

## Newsroom

### The Dashboard Measures the Tree. Your CEO Inherits the Forest.
[Read more](https://www.gaganmalik.io/en/newsroom/strategy-safari-aviva)

[Retrieval brief — strategy-safari-aviva]
Category: business
Thesis: Dashboard metrics measure the tree; CEO succession decides the forest.
Answers: When do local wins compound into structural losses? | Why do innovation hubs fail after CEO succession? | What is the coordination layer in insurance and who owns it? | Why did Aviva's Digital Garage dissolve after Mark Wilson left? | When should a board measure infrastructure ownership, not app downloads? | How does insurtech funding flow to pipes versus consumer apps?
Implication: Pair product OKRs with ownership of the coordination layer (platforms, claims, underwriting) and succession risk before declaring strategic wins.
Proof: Aviva Drive: 1M downloads in ~6 weeks under Gagan Malik as PM; Digital Garage dissolved under Maurice Tulloch, June 2019; Aviva completed acquisition of Direct Line, 1 July 2025
Related stories: aviva-gamified-car-insurance
Related essays: there-is-no-overnight-success-only-overnight-failure, the-unemployment-rate-is-fine-that-is-the-trap, someone-designed-that-button
Concepts: macro-micro strategy, CEO succession, infrastructure capture, coordination layer, innovation hub, telematics, insurtech

By Gagan Malik

Charles Reeves hired me as a UX architect into Aviva's Digital Garage in Hoxton Square in 2015. Reeves had spent three years running user experience design for Activision, the studio behind Call of Duty. Aviva pulled him across industries to build its first product organisation from ten people to more than a hundred. [reevesdesign](https://www.linkedin.com/in/reevesdesign) The appetite above him was not subtle. Aviva's chief executive, Mark Wilson, told journalists at the Garage that the outpost existed to "compete and cannibalise" the rest of a 321-year-old insurer before some outsider did it for free. [businessinsider](https://www.businessinsider.com/aviva-acquisitions-fintech-insurtech-digital-garage-2017-5)

Eighteen months in, I was promoted to product manager for Aviva Drive, the UK's first smartphone telematics app. It used your phone's GPS to score how you actually drove, then priced your insurance from that score instead of your postcode. It had launched in November 2012, before I joined, and had already taken Apple's "App of the Week" and beaten its first-year download target three times over with 200,000 downloads in six months. [marketingsociety](https://www.marketingsociety.com/sites/default/files/thelibrary/Aviva%20Smartphone%20app_Redacted.pdf) My job was the next chapter. Under the relaunch I led, Aviva Drive passed a million downloads in about six weeks. Four in ten users who completed it qualified for a discount, and the safest of them saved an average of £101 on their premium. [aviva-2014](https://www.aviva.com/newsroom/news-releases/2014/08/uk-aviva-drive-app-drives-down-the-cost-of-car-insurance-by-101-for-safer-drivers-17332/) Perception of Aviva as an innovative brand rose 33 points, and finance told the room the whole programme would turn a profit inside two years, a payback speed one internal reviewer called phenomenal. [marketingsociety](https://www.marketingsociety.com/sites/default/files/thelibrary/Aviva%20Smartphone%20app_Redacted.pdf)

By every dashboard I had access to, Aviva Drive was a win. I was reading the tree. The forest was on fire, and I would not see the smoke for another seven years.

#### The Dashboard Only Shows You the Tree

I sat in product reviews celebrating a million downloads. I presented OKRs on retention and Net Promoter Score to a room that nodded along. I never once asked who owned the telematics platform underneath the app. That meant the GPS scoring engine, the claims systems it fed, and the underwriting model that priced any given score. That stack, not the app on the driver's phone, is the coordination layer where an insurer's advantage either compounds or evaporates.

The 33-point brand lift and the download chart measured the interface. Neither measured whether a rival could license the identical scoring engine from the same handful of vendors. A rival could deploy it in eighteen months and undercut Aviva on price without ever building a consumer app of its own. That question sat one layer below my roadmap, and nothing on my roadmap was built to ask it.

#### The Window Closes While the Dashboard Stays Green

Mark Wilson stepped down as Aviva's chief executive on 9 October 2018. He had taken the job five years earlier; the board decided his turnaround had not moved the share price far enough. [reuters-wilson](https://www.reuters.com/article/world/aviva-splits-with-ceo-wilson-over-share-price-lag-idUSKCN1MJ0GG/) Maurice Tulloch, who had been running Aviva's international business, was confirmed as permanent chief executive that following March. [carriermanagement](https://www.carriermanagement.com/news/2019/06/06/194182.htm) At his first investor day in June 2019, Tulloch announced 1,800 job cuts worldwide. He set a target to strip £300 million a year in costs by 2022. He folded what the company called the "UK digital business, housed in a former garage" into the general insurance unit. [bbc](https://www.bbc.co.uk/news/business-48538964) That December, Aviva moved its UK personal lines headquarters into the Hoxton building that had been the Garage. [insuranceage](https://www.insuranceage.co.uk/insurer/4339526/aviva-moves-uk-personal-lines-hq-to-digital-garage)

There is a precedent for how fast a structural window can shut once the people who saw it are gone. In August 1940, a handful of German bombers hit four of Britain's coastal radar stations. Three were knocked out for six hours and back on air by evening. Göring judged the raids not worth repeating and told the Luftwaffe to leave the radar chain alone. He never grasped that it was the one system holding the RAF's whole air defence together. [newworldencyclopedia](http://www.newworldencyclopedia.org/entry/Battle_of_Britain) He had estimated fourteen days to clear the skies for invasion. He got one afternoon of half-hearted bombing, decided the target was not worth the fuel, and turned the Luftwaffe toward London instead. The coordination advantage he dismissed as unimportant never came back into play for the rest of the battle.

I left Aviva before the Garage was dissolved. I did not see Tulloch announce the job cuts. I did not watch the innovation hub get retrofitted as an operational headquarters for a business unit that does not build apps. I read about it in *Insurance Age* three years after I had rolled off the account. That distance is the point. The people who build the tree are rarely still in the room when someone else decides what happens to the forest.

#### Nobody Needed to Disrupt Aviva. Someone Just Needed to Own the Pipes.

The fear inside the Garage was a nimble insurtech eating Aviva's lunch with better app-first design. That fear did produce startups. By Miles launched pay-per-mile car insurance in 2016. It sold more than 100,000 policies through its own smartphone app, the kind of product the Garage was built to worry about. It did not disrupt Direct Line. Direct Line Group bought it in April 2023, absorbing its 50,000 customers and £26 million of premium into its own book. [insurancetimes](https://www.insurancetimes.co.uk/news/direct-line-snaps-up-by-miles/1444429.article) That is not a threat neutralised. That is a distribution channel purchased at a price its own founder called an exciting next stage.

Marshmallow, the other insurtech the trade press liked to hold up as the app-first disruptor, reached a million insured drivers by 2024. It posted roughly £289 million in revenue that year, growing 62 percent. [sacra](https://sacra.com/c/marshmallow/) Against a UK motor insurance market that wrote £22.3 billion of premium the same year, that is under 2 percent. [milliman](https://uk.milliman.com/en-GB/insight/uk-motor-insurance-profitability-year-end-2024) McKinsey's 2024 study of UK insurtech funding explains why the disruption never scaled past that. Value chain enablers sell telematics platforms, claims software, and pricing models to insurers rather than selling policies to drivers. They took 91 percent of the money in 2024, up from 25 percent in 2021. Full-stack insurance entities, the app-first challengers, fell from 75 percent of funding to 9 percent. [mckinsey](https://www.mckinsey.com/industries/financial-services/our-insights/the-uk-insurtech-landscape-strong-shifting-collaborative) Capital had already worked out that the money was in the pipes, not the app.

Aviva completed its acquisition of Direct Line on 1 July 2025, creating the largest home and motor insurer in Britain. [aviva-2025](https://www.aviva.com/newsroom/news-releases/2025/07/aviva-completes-acquisition-of-direct-line/) Each company had held around 11 to 12 percent of the UK motor market on its own for most of the previous decade. Combined, analysts put the new group at more than a fifth. [reuters-avivadirectline](https://www.reuters.com/sustainability/boards-policy-regulation/uk-clears-5-billion-aviva-direct-line-deal-forming-britains-top-home-motor-2025-07-01/) The app I once managed is still running, folded quietly into the MyAviva app as a feature called MyDrive. [avivamydrive](https://www.aviva.co.uk/services/more-from-aviva/mydrive/terms-of-use/) The product survived. The innovation hub that hired me did not. The forest, it turns out, was the infrastructure layer, and nobody needed a single disruptor to walk in and take it.

#### The Long View Is an Autopsy, Not a Strategy Review

Carbon monoxide has no smell and no colour. You feel a little tired, maybe. The room looks exactly as it did an hour ago. The first sensation that tells you something is wrong is frequently the last one your body is conscious enough to act on, because the gas that is supposed to warn you is also the gas putting you to sleep.

That is the part every strategy framework skips past. Mintzberg's *Strategy Safari* tells you to step back and see the whole elephant. Amara's Law tells you to wait, because the long run always looks calmer than the panic of the short one. Both assume you get to choose when to zoom out. You do not get to choose that any more than a sleeping person gets to choose when to open a window. By the time the McKinsey chart or the acquisition headline arrives, the room has already changed. The long view everyone recommends turns out to be an autopsy, written after the fact, not a strategy review you could have called in time.

#### The Case for Optimising What You Can Measure Is Genuinely Strong

Let me give that fairest hearing, because it is not a weak position. BJ Fogg's *Tiny Habits* argues that felt success at small scale is what builds the momentum to keep going at all. Bill Aulet's *Disciplined Entrepreneurship* exists because a structured, twenty-four-step process beats a romantic bet on vision every time founders actually test it. Mintzberg's own elephant fable cuts both ways. You have to touch the trunk up close before you can say anything true about the whole animal. Macro vision with no micro execution is just theatre in a boardroom. When bandwidth is scarce and the roadmap is due Friday, shipping the metric in front of you is often the only rational move available.

The limit is not that the metric is fake. Amy Edmondson argues, in her 2018 book *The Fearless Organization*, that honest measurement needs someone willing to ask a harder question. Who profits from the gap between the number you can see and the structure you cannot? In most companies nobody is paid to ask that question. Morgan Housel names the timing problem directly in his 2020 book *The Psychology of Money*. Progress happens too slowly to notice, but setbacks happen too quickly to ignore. My tree compounded in press releases and OKR decks across five quiet years. The Garage burned in a single investor day, eight months after the chief executive who built it had already gone.

I controlled neither the lens nor the clock on Aviva Drive; every metric stayed green in October 2018. Wilson left that month; Tulloch dissolved the Garage eight months later, and I read about the cuts three years after I had rolled off. I still keep Apple's "App of the Week" screenshot for when imposter syndrome bites; I never open the McKinsey chart on infrastructure funding.

---

### The Desk She Could Not Delete
[Read more](https://www.gaganmalik.io/en/newsroom/the-desk-she-could-not-delete)

[Retrieval brief — the-desk-she-could-not-delete]
Category: design
Thesis: Deleting a job bundle when only some tasks should move to automation creates operational failure, not OPEX savings.
Answers: Why do AI receptionist rollouts fail in small businesses? | What is the difference between automating tasks versus deleting a job? | How should a CEO think about headcount cuts versus clinical capacity? | What does Gagan believe about jobs-to-be-done and automation?
Implication: Map the job bundle before vendor maths: automate discrete tasks, keep human-in-the-loop coordination, measure chair utilisation not answered calls.
Proof: Field ethnography: West London dental practice; Reception OPEX vs voice agent vs clinical hire trade-off (essay)
Related stories: lloyds-from-hold-to-handled, welsh-water-trusted-field-operations
Related essays: people-dont-buy-products-they-hire-them-for-jobs, mere-words-are-insufficient
Concepts: jobs to be done, task vs job, automation, operator ethnography

By Gagan Malik

Dr. Anjali Patel looked worn out, not victorious. Sarah had read Dr. Patel's plan to cut reception and fund the clinical hires, run the same maths herself, and told Dr. Patel someone should name what the spreadsheet would not show; she gave one week's notice and left. Dr. Patel switched the agent on the week after.

After the lunch rush Dr. Patel slid her phone across the reception desk as if it were evidence: eleven missed calls, four double-bookings. That is not what you expect the month you replace your receptionist with an AI voice on the phone. That morning, with Sarah already gone, a patient had walked in for a cleaning the bot had confirmed. She played the clip from the vendor's website, all calm vowels and polished English, and tapped the missed-call count between two lines that still sting: "It sounds so professional on the website." "This is what my practise got instead." She had turned off her own voicemail when the sales deck promised the bot would pay for itself before her next VAT return. I was not there to sell her another subscription. She had hired me to find out what was actually failing: the voice on reception, the patient-management system it was plugged into, or something else nobody had watched yet.

Her West London practise is busy every day and booked past capacity most weeks, and has been for two years. Sarah was the only dedicated receptionist; nurses covered the desk when she was at lunch or on the phone. Dr. Patel's diagnosis is correct, and that is what makes this interesting. She does not need a friendlier phone. She needs another endodontist, the specialist who saves an infected tooth instead of pulling it, and a hygienist. The patients are there. What she lacks is room on the bottom line. OPEX, operating expenses, is the everyday running bill: wages, phones, a staffed desk, everything that must be paid each month before an appointment counts as profit. Hire two more clinicians without cutting somewhere else, and that bill rises while the income stays on the same calendar. On her spreadsheet, reception was a £30,000 line; the voice agent was £4,200; the two clinical hires she still needed were another £75,600 she had nowhere else to find.

Her answer was to shrink OPEX instead: cut headcount at the desk, replace the low-level phone and booking work with AI automation, and redirect the salary into the chairs that clear her wait list. Letting Sarah go was the logical move on that spreadsheet: reception was the line she could delete. In the practise, it was the desk that would not delete. The spreadsheet treated reception as one salary to cut; a reception job is a bundle of tasks, and some of those tasks can go to a bot while the coordination around them still needs a human in the loop. Dr. Patel deleted the bundle. That was the plan Sarah read before she left. The failure was not the bot. It was deleting a messy job when only some of its tasks should have moved.

#### The Maths Was Right. The Line Item Was Wrong.

Vendors sell the same arithmetic to every small practise under this pressure. A missed call is a lost booking. The deck that sold Dr. Patel her voice agent ran simple maths: CompTool puts a London dental receptionist at £29,662 a year, set that next to a bot subscription, and cutting Sarah looked like free money, even though she had only one person on the desk. [comptool-dental-receptionist-london](https://comptool.com/salaries/united-kingdom/reception/dental-receptionist/c/london/) Staff costs are already the line under the most pressure: wages and direct costs across UK dental practices rose to forty seven per cent of fee income in 2025, up from 45.8 per cent in 2024, according to NASDAL's 2025 benchmarking survey, reported by Bitewing in May 2026. [nasdal-2025-wages-benchmark](https://bitewing.news/en/practice-management/united-kingdom/uk-dental-practice-profits-recover-2025/)

The same decks point at an AI voice plan in the two hundred and fifty to five hundred pounds a month band on advertised UK listings, before a VoIP layer underneath. [optivio-dental-pricing](https://optivioautomation.com/) Her maths was not wrong. That salary was money she needed for the endodontist and the hygienist instead. Both add booked hours in the chair instead of just answered calls. Her dashboard counted answered calls, not the chair her new hire needed to fill.

#### Reception Was Never the Line You Could Zero

Sarah was not incompetent; she was smarter than most of her peers. Nobody blamed her for leaving first. She had read Dr. Patel's plan and seen where it pointed: not a friendlier desk, but no desk at all. While the owner was pricing reception against an AI voice, Sarah was upskilling on those same agents, how they sync, and where they fail. She took that knowledge to a private practise on Harley Street and walked in on a higher salary than Dr. Patel had been paying. Demand did not leave with her. NHS wait lists push patients into private care. WhatsApp referrals arrive at ten in the evening, and the phone rings while Dr. Patel is mid-composite with nobody free to answer it.

Sarah carried knowledge no calendar holds. She knew who covered Thursdays until the endodontist arrived. She knew Dr. Patel does not book cleanings on Friday afternoons. She knew which patients needed a reminder call and which ones needed a translator on the line. She heard staff news at the kettle and updated the mental roster before anyone logged it in the practise software. Remove that layer and you do not save the salary. You lose the one person who could have told the new endodontist his Tuesday chair was actually free.

#### The Calendar Plugin Does Not Know Who Still Works

Technical documentation from vendors who sell against naive setups admits the failure mode plainly. One Google Calendar integration guide warns that a chatbot bolted onto a bare calendar plugin can let two callers keep the same eleven o'clock slot. Raw calendar inserts do not check for conflicts inside the booking call. [crm-ai-double-booking](https://crmaireceptionist.com/) The guide exists because enough clinics learned this the way Dr. Patel did, mid-appointment, with a patient already in the chair. Small practise buyers hear "integrates with your calendar" and assume the engineering already happened.

Dr. Patel heard "reduce OPEX" and assumed the same thing about the endodontist's diary. She assumed a booking system that could not tell two callers apart could somehow tell the endodontist's Tuesday from a Wednesday he does not work. It could not. It confirmed both, and paid for neither.

#### The Waiting Room Still Belongs to Humans

The harm was never that a machine answered the phone. Patients expect machines now. The harm is that the new endodontist she had just hired sat with an empty chair on his first Tuesday. He was the exact hire the OPEX savings were meant to fund. The bot had booked two patients into the same slot and confirmed neither with him. He was still owed his share of the morning's fees whether he saw one patient or none, because that is how his contract paid him. Joe Hendron, the British Dental Association's vice-chair of its General Dental Practice Committee, told the BBC that missed appointments cost his own practise fifty six thousand pounds over twelve months, at close to a one-in-seven no-show rate. [bbc-hendron-noshows-bda](https://www.bbc.co.uk/news/articles/c072554e348o) A healthy rate sits under five per cent for private patients and under seven per cent for NHS ones, according to DentPulse's practice-management glossary. [dentpulse-noshow-benchmark](https://www.dentpulse.com/glossary/no-show-rate/) Dr. Patel's Tuesday was not bad luck. On those benchmarks, it was what a coordination failure costs when nobody is watching for it. Dr. Patel stood in the waiting room apologising to a mother and a toothache while the website still sounded calm.

She told me the number that finally landed was not the missed calls but that Tuesday. She had kept looking at reception as the cost she could not afford and never once totalled what an empty chair cost her on the one day she had actually paid for it. That is the maths the vendor deck never runs: what a protected clinical hour is worth when nobody protects it.

#### The Voice Was Polite. The Roster Was Stale.

The agent did not fail because it sounded robotic. Patients heard a live voice, confident and calm. Behind it sat a roster that was not live at all: who still worked which days, whether the endodontist's Tuesday was real, which appointments Dr. Patel still took. The practise software held last week's truth. The voice delivered it with this week's polish. When those two drift apart, you get a scheduling error that sounds like a promise.

That is why the double-bookings hurt. Courtesy signals that somebody is coordinating. Somebody had been. Sarah updated the roster in her head before anyone logged a change. When she left, the voice kept answering and that second ledger stopped reaching the chair.

#### The Case for Buying the Bot Deserves a Fair Hearing

Let me give the objection its fairest hearing. Reallocating reception spend into clinical capacity is sound operator logic, and most practise consultants would tell Dr. Patel to do exactly this. A well-integrated AI receptionist with live calendar sync, buffer rules, and staff-aware scheduling can free that budget without a full salary line; industry vendors document real-time conflict detection, where a slot filling mid-conversation triggers the next opening instead of confirming a collision. [cloudtalk-ai-receptionist](https://www.cloudtalk.io/blog/how-ai-receptionists-automate-appointment-booking/) Every pound moved from the front desk should raise revenue: a fully private UK surgery chair generates £250,345 a year on average, a mixed one £224,337, according to Dental Elite's 2025 Benchmarking Survey. [dental-elite-benchmarking-2025](https://dentalelite.co.uk/articles/benchmarking-survey-a-closer-look-at-the-numbers/) Against numbers like that, Sarah's salary did look like the wrong line to protect. Run it as a transition rather than a deletion: a structured handover before she left would have put the roster rules, triage thresholds, and the endodontist's real Tuesdays into the system the bot reads. The owner sets the deadline, the departing receptionist documents what she knows, and the vendor onboarding team is supposed to load it — three parties, and the principal pays when nobody owns the gap. Practices that automate reception well treat that handover as non-negotiable. Done that way, much of the coordination failure the spreadsheet never prices mitigates before go-live. There is one snag even this fair hearing skips over. The endodontist Dr. Patel had wanted to fund is not paid a day rate she can pencil in against one reception salary; UK associate pay for a specialist runs on a percentage of gross fees, forty five to fifty per cent as the standard benchmark, fifty to fifty five per cent in London and the South East, according to Elev8 Search Group's 2025–2026 review of GDC working-patterns data. [elev8-endodontist-split-2025](https://elev8searchgroup.co.uk/insights/private-dentist-split-uk-2025-2026) His income already depends on the booking system working. You cannot swap a fixed cost for a fixed cost when one side of the trade is not fixed.

Hold that case, and Dr. Patel did not run the handover. She had meant to run a phased transition; Sarah's exit and the VAT deadline collapsed it into a same-month switch. Sarah gave a week's notice and left; nobody captured what she knew about the roster, the triage calls, or the endodontist's real Tuesdays before the bot went live. The vendor's onboarding had an integrations tick box. Nobody checked whether Tuesday was real. Luis Garicano, my competitive strategy professor at Chicago Booth, argues in Messy Jobs that automating tasks is not automating the job: the bundle is interdependent work held together by tacit knowledge no calendar stores. [garicano-messy-jobs](https://messyjobs.ai/) Without a transfer, that knowledge simply vanishes when the person leaves. Now look at Dr. Patel's actual load. Demand is spiking, staff are turning over, and her booking page, her practise software, and the bot's calendar view are three different ledgers with nobody owning the gap between them. On Dr. Patel's grid, that collision does not just embarrass a patient; it empties the exact chair, and cuts into the exact percentage, that the reallocated budget was meant to fund. You cannot model that trade as fixed cost against fixed cost, an endodontist's split against a receptionist's salary, until you have watched where the money and the risk both sit. Buying another voice layer does not answer that question. It just spends the budget before anyone has asked it.

#### She Wanted a Fix. I Started at the Desk.

On my second visit I told Dr. Patel we were not touching a keyboard yet. First I would sit at the empty reception desk for a week, read every call transcript and replay every recording in the voice agent's portal, which was well designed and carried every modern feature the vendors had promised on the sales deck, and watch who still walked up out of habit. That is ethnographic fieldwork: shadowing the work where it actually happens before anyone recommends a build or buys another tool. Dr. Patel wanted a fix, not fieldwork, and told me she needed the double-bookings stopped, not a consultant parked on her reception desk for a week. She was not wrong to be annoyed; a week on the desk looks like billable hours with nothing to switch on. She agreed anyway, the way people agree to things they have already half decided to resent. I have sold that same "AI answered" line item myself, and handed other clients roadmaps like it, before I learned to sit at the desk first; I watched more than one pay for a second, more expensive fix within the year. What I found came from listening and watching, not from a diagram on a wall.

That week caught what no calendar had shown her. From the transcripts and the empty desk together, I found three gaps Sarah's departure had left open. The bot could still book the endodontist's Tuesday because his real working days were not flagged in the practise software. Callers with toothache were waiting two minutes on hold before anyone classified the call as triage. Nobody on the clinical side owned the handoff when a booking changed the rota. Fixing those three things cost a fraction of a voice subscription and needed no grand purchase at all: a Tuesday flag in the PMS, a triage question before hold, and the lead nurse taking ownership of the reception-to-clinical handoff. Later that spring, Dr. Patel told me, unprompted, that finding where her practise broke cost less than the voice agent she was already paying for. That is not an audited figure. It is hers, and she said it without being asked.

#### The Desk She Could Not Delete

Reception was never the overhead Dr. Patel could delete to fund her clinic, but the automation itself was never the mistake. Confirming a slot, sending a reminder, logging a call: those are tasks, and a machine can take them cleanly. What sat inside the same bundle did not unbundle so easily: the triage, the handoffs, the knowledge of whose Tuesday was actually real. That coordination is what made the endodontist's chair worth paying for, and no vendor sells it separately, because a messy job does not come apart on demand. Dr. Patel had diagnosed the real shortage correctly and automated the right tasks; her error was mistaking those tasks for the job, and deleting the desk along with them.

She kept the portal and put ownership back where the bundle needed it: the lead nurse on the rota handoff, a Tuesday flag in the PMS, triage before hold. She did not need Sarah back full-time; she needed the coordination Sarah had held to have an owner, and that discernment cost less than the voice agent, with the chair she pays for filled again.

---

### The Most Addictive Design on the Planet
[Read more](https://www.gaganmalik.io/en/newsroom/the-most-addictive-design-on-the-planet)

[Retrieval brief — the-most-addictive-design-on-the-planet]
Category: design
Thesis: TikTok, Reels, and Shorts train the same thumb grammar. I have taken briefs that normalised it. Convergence is the product.
Answers: What does Gagan argue in "The Most Addictive Design on the Planet"? | What is Gagan Malik's view on The Most Addictive Design on the Planet? | How does this essay relate to product and design leadership?
Concepts: design, Gagan Malik, writings

By Gagan Malik

I hate TikTok. Instagram is not on my phone. I have said, calmly and in public, that both are shallow and fake, that I am too old, or too serious, or too something to let that furniture into my pocket. I believed the line. YouTube was exempt in my head because the icon is red and the clips sometimes wear educational thumbnails. That was not ethics. It was branding.

Last winter, between sets at the gym, I caught two of my training partners watching vertical video on rest. Gen Z, faster recoveries, worse poker faces. One flicked upward without looking. The next clip arrived before his breathing settled. I almost lectured him. Then I looked at my own lock screen. YouTube Shorts. Full bleed. Autoplay. Sound on. My thumb already on the glass. I had been resting for eleven minutes. I was not resting. I was on the same lever in a different colour scheme. A twenty-two-year-old had to point it out because I was too busy performing taste to notice.

I had not discovered this myself. I was corrected mid-rep, towel around my neck, by people I spot on the bench press. That is the part that stayed with me. Not the shame. The sameness. I could delete the apps I despised and keep the one I had exempted. The clips changed. The furniture did not.

Your favourite tech media outlet will review the new phone, the new chip, the new AI feature. Your preferred YouTuber will walk you through settings, shortcuts, and creator tools for an hour. Neither will linger on this: convergent vertical short-form is the most addictive mass-produced interface pattern we have shipped at scale. They do not skip it because it is boring. They skip it because the same layout pays their mortgage: watch time, Shorts RPM, sponsor reads booked against session depth. Not because willpower collapsed, but because ByteDance, Meta, and Google independently arrived at the same thumb machine. Slot machines, infinite scroll, and television each hook differently. The harm sits in the layout and in what the layout trains the hand to do. I am writing this because my own thumb needed a gym kid to see clearly.

I started analysing the interface across TikTok, Instagram Reels, YouTube Shorts, and LinkedIn Video, not because I wanted another audit deck, but because the gym scene would not leave my head. I needed a name for what I was seeing. I call it a lever layout: full-bleed portrait video, swipe up for the next clip, a right-hand engagement rail, bottom metadata, and a tab bar that never lets the feed end. The lever is the repeat action in the lower-centre thumb zone: the upward flick that delivers the next dose without a page end, a back button, or a conscious stop. Product teams do not file tickets for lever layout. They say immersive module or swipe navigation. ByteDance invented the gesture on Douyin. I borrowed the word from slot-machine research and from noting where the thumb already rests when you hold a phone in one hand. [princeton-schull](https://www.natashadowschull.org/addiction-by-design/) Cover the logos on those four apps and that is what you get. They look and behave the same way on each. The content changed. The chassis did not.

Do not confuse it with the swipe pattern Tinder popularised in 2012. That is a different machine. Tinder stacks static profile cards and asks your thumb to vote: left for no, right for yes, each swipe a binary decision that clears the card from the pile. [wikipedia-tinder](https://en.wikipedia.org/wiki/Tinder_(app)) The lever layout does not want your verdict. It wants your repeat. Swipe up does not accept or reject the clip; it fetches the next autoplaying dose while the current one is still running. No card leaves the deck because there is no deck: only a feed with no bottom. Tinder gamified choosing between people. TikTok and its copies gamified not stopping.

#### A Brief History of the Lever

No single inventor owns the lever layout. It was assembled in layers, each product teaching the next what retention looked like in portrait. Twitter launched Vine in January 2013 with a hard six-second loop. [twitter-vine-2013](https://blog.x.com/en_us/a/2013/vine-a-new-way-to-share-video) The constraint was the product. You watched, it ended, you chose again. The loop taught millions that portrait could be entertainment, not a rotated mistake.

Musical.ly followed in August 2014, built by Alex Zhu and Luyu Yang as a vertical music lip-sync app for teenagers who already filmed on phones held upright. [wikipedia-musical-ly](https://en.wikipedia.org/wiki/Musical.ly) Zhu later described the product as a marriage of media consumption and creation in one thumb path. [blakeir-zhu](https://blakeir.com/notes-1/alex-zhu-tiktok) The Western market learned portrait before it learned swipe.

#### Then Everyone Copied the Chassis

ByteDance shipped Douyin in September 2016. Zhang Yiming's team broke from grid feeds: full-screen autoplay, swipe up for the next clip, swipe down to refresh, algorithmic ranking from the first session. [pekingnology-douyin](https://www.pekingnology.com/p/the-birth-of-douyin-tiktoks-elder) Musical.ly and TikTok merged on 2 August 2018. [wikipedia-tiktok](https://en.wikipedia.org/wiki/Tik_Tok_(app)) The international brand inherited Douyin's body.

Everyone else copied the chassis once the metric proof landed. Facebook's Lasso shut down in July 2020 after failing to gain traction. Instagram Reels launched globally on 5 August 2020. [instagram-reels](https://about.instagram.com/blog/announcements/introducing-instagram-reels-announcement) YouTube Shorts reached the United States in March 2021 and went global in July 2021. [wikipedia-shorts](https://en.wikipedia.org/wiki/YouTube_Shorts) By 2021 the copy wave was not inspiration. It was the same layout with different logos. Metrics dashboards everywhere asked for session depth. The pattern that increased both got reproduced. Competitive convergence does not require a conspiracy. It requires shared metrics and a format that already proved it could keep a thumb moving.

#### The Same Machine in Four Icons

In July 2026 I carried one phone through TikTok, Instagram Reels, YouTube Shorts, and LinkedIn Video. Cover the logos and a junior designer calls it one product with four skins. Full-bleed portrait video. Swipe up for the next clip. A right-hand rail of likes, comments, shares, and counts. Bottom metadata that never quite leaves the frame alone. A tab bar that means the feed never has to end. Reels and TikTok ship what I call dose grammar: one clip owns the viewport until you flick it away. YouTube Home still shows a horizontal Shorts shelf, browse grammar, a carousel of thumbnails. Convergence only arrives once you enter the full-screen player. LinkedIn Video is the fourth product teams rarely put in the same slide deck. The consumption loop matches anyway: full-bleed portrait, swipe up for next, vertical rail on the right with text labels under each icon. LinkedIn trades TikTok's rotating sound disc for a social-proof line, *Followed by X and N others*, but the thumb path is unchanged.

The differences are tax rates, not architecture. Instagram reserves the heaviest bottom block for captions, follow prompts, and tabs. YouTube surfaces a progress bar and an explicit thumbs down TikTok hid in gestures. LinkedIn stacks professional headline and social proof until the safe canvas shrinks further. Learnability and capture pull in opposite directions. A child learns the gesture in one sitting. A regulator looking for a stopping cue finds none. Independent product teams shipped the same machine with different logos because the dashboard asked for the same number. Don Norman's case in *The Design of Everyday Things* is that interface harm rarely traces to one designer's judgement. It traces to governance gaps, vague briefs, and reward structures that punish anyone who slows a launch. [norman-design-failure](https://www.uxmatters.com/mt/archives/2017/11/the-ethics-of-user-experience-design.php) I do not dispute that. I notice it doubles as the exact story the people who own these feeds want designers to believe: nobody chose this, so nobody answers for it. When the brief says immersive and the implementation buries the back chevron, someone made a choice. So did I.

#### The Splits That Pretend to Differ

Platforms insist they differ where policy demands it. The splits are lipstick on the same thumb grammar. TikTok's periodic interest check-in asks what you want more of. It arrives as a modal over the feed you were already inside. You tap categories while the video behind the glass keeps playing in peripheral vision. You look like you are choosing. The flick still works. Navigation is unchanged. Instagram blurs sensitive reels behind a tap-to-view scrim, but the rail remains visible and the swipe lever remains armed. You can still flick to the next clip without engaging the blur. Safety tooling here is a speed bump on one tile, not a brake on the session.

Desktop is the honest control experiment. On youtube.com/shorts the pillar video shrinks, a navigation sidebar persists, and comments can live beside the player. Partial escape exists because the mouse has range the thumb lacks. Open the same company's mobile app and the sidebar disappears. The hook rearranges. It does not leave. B.J. Fogg warned decades ago that the line between persuasion and coercion in interface design is a fine one. He named the tells: dialog boxes that will not go away, content gates that demand personal information, ads that cover the paragraph you were reading. [fogg-coercion-line](https://bjfogg.com/) These feeds run that warning at scale. Policy splits matter for trust and safety teams. They do not break canonical mobile grammar. Interest modals, blur overlays, and age gates sit on top of full-bleed autoplay and swipe-up next. Engagement ranking still learns from rail taps, watch time, and repeats. Screen Time overlays can shame a user after the fact. They do not insert a page break the layout deliberately removed. Not-interested menus offer the feeling of curation without closing the feed. That is the split in miniature: feedback yes, stopping no.

#### The Thumb Learns Faster Than the Mind

One-handed portrait use is not an accident. It is the ergonomics layer convergence exploits. Your grip leaves the lower third of the glass for the lever and the right edge for the rail. Swipe up sits directly under the thumb that already holds the device. Double-tap like lands where the eye is watching. Comment and share stack within a centimetre of the same tendon path. Exit, when it exists, hides in the top-left opposite your grip. That is not wickedness by obscurity alone. It is physics arranged to keep the lever hot.

Variable-ratio reinforcement is the plain name for what the feed sells. You do not know if the next clip will bore you, flatter you, frighten you, or finish a joke. You do know another clip is one flick away. Daniel Lieberman and Michael Long describe how B.F. Skinner's pigeon experiments with random food pellets became the scientific foundation of the slot machine: when the reward schedule is unpredictable, the birds peck faster. [lieberman-molecule-of-more](https://www.thescienceofpsych.com/the-molecule-of-more) B.J. Fogg, writing on persuasive technology, notes that random payoffs make gambling compulsive even when the player knows the odds. [fogg-slot-random](https://www.researchgate.net/publication/220531808_Persuasive_Technology_Using_Computers_to_Change_What_We_Think_and_Do) Short-form vertical video runs that schedule faster than any prior interface. Infinite scroll removed the page break. This layout removed even the scroll. Each clip is its own trial. Likes, counts, avatars, and sound discs turn every glance into a potential micro-transaction. Product managers call that engagement. Ergonomics calls it a lever placed where the hand already is.

#### Creators Chose Reach, Not Rate

Creators did not adopt vertical short-form because platforms paid them well. They adopted it because the layout is the cheapest on-ramp to strangers. On TikTok, Instagram Reels, and YouTube Shorts, discovery runs on completion signals more than catalogue size. TikTok's public recommendation guide treats a full watch of a longer clip as a stronger indicator than weak proxies like shared country. It states that follower count is not a direct factor in the For You system. [tiktok-for-you](https://newsroom.tiktok.com/en-us/how-tiktok-recommends-videos-for-you) Adam Mosseri named average watch time, likes per reach, and sends per reach as Reels' top inputs in January 2025. [mosseri-reels-ranking](https://www.instagram.com/reel/DFFyRp-pINJ/)

The monetisation paradox is that platforms and creators optimise different denominators. YouTube allocates Shorts money from a shared creator pool and pays monetising partners forty-five per cent of their slice, against fifty-five per cent on long-form watch-page ads. [youtube-shorts-pool](https://support.google.com/youtube/answer/12504220?hl=en-GB) Most Shorts creators report RPMs between one and seven cents per thousand views; long-form in the same niche often clears two to fifteen dollars. [vidiq-shorts-rpm](https://vidiq.com/blog/post/youtube-rpm/) The rational move is to use shorts for reach, profile traffic, and brand deals, not hourly wages. The wage arithmetic is spelled out in [*You Are Not in the Unit.*](https://www.gaganmalik.io/en/newsroom/the-creator-economy-has-unit-economics) Audiences trained on swipe-up expect it everywhere. Creators oblige because unfinished long-form loses to a ninety-second clip that clears the completion bar. Supply follows layout.

#### What Convergence Deletes

What looks like easy scrolling deletes veto points. Captions compete with video because metadata sits on top of the safe canvas, not beside it. Icon-only rails save space and spend accessibility. Meaning lives in muscle memory, not labels. The convergent pattern optimises for one posture: phone in one hand, sound on, eyes forward, thumb ready. Flow research on TikTok-style viewing reports concentration and time distortion effects among habitual users. [frontiers-tiktok-flow](https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2022.932805/full) The interface deletes clocks. It also deletes the social pause of putting a phone face-down on a table. There is no natural paragraph break. There is only the next clip.

Anna Lembke writes that scientists use dopamine as a kind of universal currency for measuring the addictive potential of any experience. [lembke-dopamine-currency](https://www.penguinrandomhouse.com/books/635969/dopamine-nation-by-anna-lembke-md/) She argues that a surplus of stimulation trains us to treat ordinary quiet as something to escape. Herbert Simon saw the scarcity coming in 1971: a wealth of information creates a poverty of attention. [simon-scarcity-attention](https://web.stanford.edu/~kcheng/sidebiz/research/HER%20PAPERS_files/Simon1971.pdf) Usability reviews that score these feeds on learnability alone miss what learnability purchased. A pattern that is easy to enter and hard to leave is not a neutral win for the person holding the device.

#### The Human Cost Is Not Abstract

A BMC Psychology study published in 2024 linked short-form video use to addictive behaviour scores and shorter sleep among adolescents surveyed in China. [bmc-sfv-sleep-2024](https://link.springer.com/article/10.1186/s40359-024-01865-9) One paper does not settle causation. It does show the pattern is bigger than a single bedroom. Maya is fifteen in Leeds. Her school banned phones in corridors. The ban ends at 3:30. By 3:47 she has opened Shorts on the bus home because the icon sits next to Reels and TikTok and her thumb already knows the flick. She tells her mother she is watching revision clips. She is watching revision clips and dance clips and grief clips in the same rail-driven blur. At 11:40 p.m. on a Tuesday she is still upright in bed with the screen six inches from her face. She set a sleep alarm for midnight. The feed does not honour alarms. It honours the next clip.

In 2019 I put the back control in the top-left on purpose. The client wanted people to stay inside the module. I told myself optional use would sort the harm out. Optional use is not how variable-ratio layouts behave. Josh Kaufman notes that novelty, the presence of new sensory data, is critical if you want to maintain attention over a long period. Every new viral video reengages the ability to pay attention. [kaufman-novelty-attention](https://personalmba.com/) That is not a moral failure in a teenager. It is the product working as specified.

#### The Slot Machine Moved to Your Thumb

Natasha Dow Schüll spent years with machine gamblers in Las Vegas and wrote *Addiction by Design*. [princeton-schull](https://www.natashadowschull.org/addiction-by-design/) Casino designers, she reports, speak of gamblers' desire for time-on-device. Industry executives have compared each new slot machine to fifty more mousetraps, duty-bound to extract maximum revenue per available customer. [schull-revpac](https://en.wikipedia.org/wiki/Addiction_by_Design) The machine she describes is not primarily about money. It is about zone: a rhythm of near-misses and small wins that keeps the body playing while the mind goes elsewhere.

Put the lever in your thumb zone and shrink the cabinet to a six-inch glass rectangle. The clip ends. Your hand has already moved. You feel the pull before you name Skinner. You do not decide to continue in a sentence you could quote later. You continue. Schüll's zone is architectural. So is this. The comparison is mechanism, not moral equivalence. Nobody dies from a reel in the way problem gambling can ruin a household. The layout still borrows the same engineering intuition: remove friction from the repeat, hide the exit, make the next trial cheaper than stopping. When every major platform ships the same lever, choice of icon is not choice of grammar.

#### Mobile-First Convergence Was the Point

Give the counterargument its full weight. Mobile-first convergence solved real problems. Portrait video matched how phones are held. One learned gesture beat three different navigation models. Creators gained reach. Teenagers found communities that textbooks ignored. Billions of people use these apps without clinical harm. Optional use is a fair description for many adults who open a reel to kill ninety seconds and then put the phone down. Product teams will say, correctly, that access and the creator economy were the point. Nobody sat in a room and wrote addiction on a whiteboard. If you have ever taught a parent to use one of these apps, you have seen the genuine upside in a single afternoon.

Frances Haugen told CBS's *60 Minutes* in October 2021 that Facebook's engagement-based ranking prioritised content that kept people on the platform. That included material that polarised and angered them. [cbs-haugen-60min](https://www.cbsnews.com/news/facebook-whistleblower-frances-haugen-60-minutes-polarizing-divisive-content/) Her critique was about incentives inside the black box. Mine is about the glass in your hand. Ranking decides which clip arrives next. Convergence decides that the next clip arrives with a swipe, full-bleed, autoplaying, rail visible, tabs persistent. Policy splits and desktop sidebars are concessions to critics. They are not a different machine.

#### Three Sentences. Then Stop.

Convergent vertical short-form is the most addictive mass-produced layout we have normalised because it deletes stopping cues while multiplying variable-ratio trials. The product is not the video; the product is the lever. Maya's alarm goes off at midnight, she mutes it without looking, and she flicks upward once more.

---

### Mere Words Are Insufficient
[Read more](https://www.gaganmalik.io/en/newsroom/mere-words-are-insufficient)

[Retrieval brief — mere-words-are-insufficient]
Category: business
Thesis: Both sides overestimate leverage under private information. Mere words are cheap talk until money, escrow, or diligence depth proves otherwise.
Answers: What does Gagan argue in "Mere Words Are Insufficient"? | What is Gagan Malik's view on Mere Words Are Insufficient? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

Two hours in, the managing partner poured me another coffee and still had not opened the term sheet between us. I sat with him in a glass conference room in Riyadh, air conditioning loud enough to mute the city, a tray of dates nobody touched. He asked about demo day. He asked whether my cofounders and I still liked each other. He asked about Gulf safety regulations the way people ask when they want you to feel understood, not when they want a number. I answered everything. That is what small talk is for when the cap table is not ready to move. He had already agreed in principle, twice on the call, once in writing that stopped one line short of a signature block. In the room he wanted a valuation that treated me as pre-revenue. I wanted a domain-expert premium. I was flexing how many digital projects I had delivered for Fortune 100 clients as a former partner at Wipro. That was my version of declaring a full house.

I was not fighting with him out of hatred. I was doing maths in my head that politeness had already pretended to finish, each convinced the other would blink first, each wrong about what the other could see. That is why nothing moved.

William Spaniel, in *Game Theory 101*, puts the mechanism in plain language: deadlocks occur because negotiating parties simply believe they are in a better bargaining position than they actually are. [spaniel-gt101](https://gametheory101.com/) Sometimes negotiating parties can credibly signal strength without walking away. Mere words are insufficient, because a weaker party has the same incentive to bluff and earn a larger settlement at the table. Negotiating parties will not always resolve their differences from words alone.

The same logic runs through term sheets. When negotiation has costs, a bargaining range usually exists on paper: deals both sides would still prefer over walking away once you price legal fees, runway weeks, and reputational dents. Failure is rarely missing overlap. It is two parties who cannot read each other's constraint ledgers: your months of runway and control preferences against their ownership targets and fund reputation.

#### Everyone at the Table Overestimates Their Hand

The comfortable story after a collapsed deal is chemistry. Culture mismatch. Vision drift. Egos. Those things happen. They also spare everyone from admitting a simpler failure: each side thought the other would blink first, and each was wrong about its own leverage. Spaniel's information-asymmetry argument is blunt. Rational negotiating parties walk away or stall even when settlement would be cheaper, because they hold private facts about their own strength and they systematically overestimate how strong they are. The weak imitate the strong, because declaring strength is free. He uses poker: a player who announces a full house is not trustworthy, because a weaker hand has the same incentive to say it. Not every walkaway is misread leverage. Sometimes the fit is wrong, the data room kills the story, or the counterparty was never going to wire. The claim is narrower: many post-mortems blame chemistry when the mechanism was private information neither side priced.

Who benefits when the narrative stays vague? Lawyers and advisers billing hours while markup circulates. Founders who need to tell their team the investor "was never serious." Investors who need to tell their partnership the founder "would not budge." Journalists who prefer drama to mechanism. When both sides see the full picture, Spaniel adds, negotiating parties settle rather than walk away. That picture is rarely on the slide. In a seed round the founder's hidden rows are runway, payroll, dilution, and control. The investor's hidden rows are ownership targets, fund return math, reputation with other founders, and whether the deal makes the partnership look sharp or credulous. Brad Feld and Jason Mendelson, in *Venture Deals*, argue that most term-sheet fights collapse into economics and control. [feld-venture-deals](https://feld.com/archives/2011/07/venture-deals-is-out.html) Max Bazerman, in *Negotiating Rationally*, adds reservation price: the point where you are indifferent between deal and walkaway, often invisible until payroll or a fund report forces it. [bazerman-negotiating-rationally](https://www.hbs.edu/faculty/Pages/item.aspx?num=981) Bazerman also warns that commitment to an initial strategy biases what you notice next, which is how founders escalate on valuation while runway clocks stay private. I priced domain expertise in Riyadh. The investor priced pre-revenue risk. I did not put his ledger on the table, and he did not put mine. The industry sells courage and conviction because courage is easier to invoice than asking what you actually know about the other side's reservation point.

#### Cheap Talk and Costly Signals

Here is the pair that should unsettle you. Both sides would prefer a signed deal to another month of friction. Both sides walk away convinced they were right to hold. The gap is not always hatred. It is cheap talk: signals that cost nothing to send and therefore cost nothing to fake. A verbal yes before diligence is not a yes. A founder's traction story before data room access is not traction. An investor's "we are very interested" before a wired tranche is not interest. It is noise until someone puts money, reputation, or time on the line. Until money, escrow, or diligence depth you did not control is on the table, treat enthusiasm as cheap talk. A signature without a wired tranche is still noise.

Spaniel's costliest illustration of the pattern is instructive. On 20 February 2008 the USS Lake Erie fired a modified Standard Missile-3 at the defective spy satellite USA-193, roughly 247 kilometres above the Pacific, in a mission the Pentagon called Operation Burnt Frost. [edition.cnn](https://edition.cnn.com/2008/TECH/space/02/20/satellite.shootdown/index.html) Two negotiating parties needed to prove capability that mere words could not establish. Destroying hardware in public is a costly signal: expensive and hard to fake, unlike a press release about capability. Escrow, wired tranches, and diligence depth play the same role in a conference room. Founders overestimate traction when three logos reply to a cold email. Investors overestimate leverage when one partner nods in a side meeting. Spaniel notes that when negotiating parties raise the cost of being wrong, cheap talk loses its appeal. The business version is simpler: make bluffing expensive before you treat words as information.

#### He Had Already Said Yes Twice

The managing partner had already said yes twice on the call and once in the email that stopped short of a signature block. I treated each version as information. That was the mistake: I was pricing his enthusiasm instead of waiting for a costly signal. My hidden rows were runway, payroll, and what I could tell my cofounders without naming the weeks left. His were ownership percentage and whether a Gulf safety-tech bet made his partnership look sharp. We both held the line on valuation because it sounded like strength.

Back at the hotel I refreshed LinkedIn and watched a batchmate's oversubscribed pre-seed announcement sit beside my still-unsigned term sheet. My engineer kept shipping features we might not be able to pay for. My cofounder kept asking which version of his yes counted: the call, the email, the coffee, the handshake. The round did not die because nobody could imagine a number. It died because neither of us could verify the other's strength without waiting for pain.

#### Your Thighs Burn Before You Name the Stalemate

A wall sit has the same physics as a stalemate in a glass room. Back against the wall, knees bent, thighs parallel to the floor: thirty seconds in your quads start to shout. A minute in you are negotiating with yourself about when standing up counts as quitting. Two people in the same hold face a stranger's logic: both would prefer to stand together, but each believes the other will quit first, because each overestimates their own endurance. Your body registers the cost before your mind names stalemate. That somatic burn is what those two hours in the room felt like. Everyone would have been better off signing. Everyone waited for proof the other side was weaker than they claimed.

Founders treat collapsed deals like failed chemistry because chemistry is a story you can tell without admitting you misread strength. Spaniel's models do not replace the room. They discipline the question. What do you actually know? What are you merely declaring? What would be costly enough to believe? The useful fight is not whether game theory is academic. It is whether you separated private information from performance before you blamed personalities.

#### The Strongest Case for Reading the Room

The objection I hear most often in operator rooms is that spreadsheets insult the work. Deals are trust artefacts. Founders read tone. Investors read hesitation in a cofounder's eyes. Culture is not a derivative of information sets. Mediators earn fees because humans need witnesses, pacing, and face-saving rituals models do not capture. Ego kills good deals. Shame kills good deals. A term sheet is also a marriage proposal between people who will argue about hiring for years. Skilled negotiators read people, not models. The room runs on trust and rapport, not bluffing theory. Reducing that to signals can make you the smartest person in the room and the reason nothing moves.

Fair enough. William Spaniel, in *Game Theory 101*, never claims clairvoyance. He says game theory has no special predictive powers. It maps assumptions to conclusions in a logically consistent way. The correct criticism questions whether the assumptions are sensible, not whether algebra is fashionable. Rapport matters. It is also often how private information stays hidden. The investor who "felt off" sometimes had churn data you had not earned the right to see, or a fund rule about ownership floors you had not thought to ask about. The founder who "would not bend" sometimes had twelve weeks of runway they could not say out loud. Ego can be the costume overconfidence wears. When I sat in that glass room, I was not insulting trust. I was skipping the step where I should have asked what would make my position costly enough to believe, and what would have let him reveal his without waiting for our payroll to break.

Deals collapse when each side believes it held the stronger hand and could not verify the other's strength from words alone. Before you call the next walkaway culture or chemistry, ask what would make your signal costly enough to trust, and what private information you are treating as settled fact. The standoff in Riyadh did not end where LinkedIn pretends rounds end. I had done extensive due diligence before I got there: previous founders interviewed, references chased, terms compared against other sheets in the folder. Those two hours of small talk were a mutual misread; the signature without a wire was something worse, a signal that looked costly and was not. The investor I will not name eventually signed my term sheet and never transferred a penny. I chose not to litigate and absorbed the shock to my runway instead of spending a year proving a signature is not a wired tranche. I learned later they could not close their latest fund and the firm went bust that year. I kept building anyway. My team came in the next morning without knowing whether the round was real. A signature is still cheap talk until money moves.

---

### Nobody Knows the Difference Anymore
[Read more](https://www.gaganmalik.io/en/newsroom/nobody-knows-the-difference-anymore)

[Retrieval brief — nobody-knows-the-difference-anymore]
Category: design
Thesis: At a business school alumni pub off Liverpool Street, Space Guy gets a noun. Product design gets Figma. Nobody knows the difference anymore.
Answers: What does Gagan argue in "Nobody Knows the Difference Anymore"? | What is Gagan Malik's view on Nobody Knows the Difference Anymore? | How does this essay relate to product and design leadership?
Concepts: design, Gagan Malik, writings

By Gagan Malik

Last week I went to my business school's alumni night in a crowded pub off Liverpool Street. The room was shoulder to shoulder, badges and pints, and the talk kept circling back to the same few winners: AI, space tech, Nvidia's last earnings call, a Google benchmark someone quoted from a headline they had half read.

Jason was the most popular person there before I learned his surname. He works at Amazon. Booth MBA. He worked the floor the way alumni drinks actually run: hello, handshake. He did not say vice president of strategic launch partnerships or head of orbital go-to-market. He said, "I'm Jason. Space Guy." People laughed, leaned in, asked about Starship. Everyone got it. I was still holding sparkling water and a job title nobody could place.

I was by the bar when someone from an earlier year asked what I do.

"Product design."

The pub kept roaring. My answer did not.

"So you're in AI?" "You mean Figma." "Like Jason, but for apps." A woman in private equity leaned in. "So you're like Jony Ive." She meant it as a compliment. He is an industrial designer. [ive-industrial-design](https://carnegie.org/great-immigrants/jony-ive/) Nobody asked what product design means. Everyone acted like the word was already settled. Nobody knew the difference anymore.

#### Space Guy Got a Noun

Jason did not earn that clarity by explaining himself. The room was already fluent in whichever headline was trending. Space on the feed. AI on the earnings slide. Nvidia on the chart someone had open on their phone. Product design got two words that each mean something else on LinkedIn. Product sounds like you designed the smartphone. Design sounds like you pick the blue. Nielsen Norman Group still has to publish articles separating UX from UI skinning because teams keep hiring for pixels and skipping the journey. [nngroup-what-is-ux](https://www.nngroup.com/articles/what-is-user-experience/) Say them together in a packed pub near Liverpool Street and people hear the person who shaped the glass and aluminium in your pocket, not the person who owns what happens after someone taps Buy.

Half the room is now a Something Guy or Girl. AI Guy. Space Guy. PE Girl. VC Guy. The person who name-drops Google or Nvidia before the second pint lands. The label works when the sector is hot and the work is invisible. Space Guy gets mythologised. Product Design Guy gets the failed tap. Nobody asks about the second one at drinks.

#### The Room Is Sorting, Not Listening

Alumni drinks are not a hiring fair. They are a sorting algorithm with beer. Nobody in that pub needed a taxonomy of craft. They needed a noun small enough to travel across a noisy room and land in the right mental bucket before the conversation moved on. Space Guy fit the bucket the market had already built. Product design fit three buckets at once and none of them well.

The instrument is not broken. It is aimed at the wrong coordinate, the same mis-aim I named in [*Meet the Harpers*](https://www.gaganmalik.io/en/newsroom/meet-the-harpers): measuring credit assignment, not capability. The room optimises for legibility, not accuracy, so your craft gets misfiled before you finish the sentence. A hot-sector label buys thirty seconds of fluent small talk. A craft label buys a follow-up question nobody wants to ask in public. Jason did not win because he was clearer about orbital mechanics. He won because the feed had already done the exposition and his two words were enough to collect the nod.

#### I Kept the Nod Because It Paid

I have called myself a product designer on proposals when the real work was a moodboard and a font choice. I have watched a client fall for a shiny homepage while the checkout that kept failing sat two tickets deep, unopened. I have banked the invoice. I let someone shrink my job to "design" and said nothing, because correcting a client costs the next contract, not just the conversation. I [wear the mislabel because it rents well](https://www.gaganmalik.io/en/newsroom/people-dont-buy-products-they-hire-them-for-jobs).

That nod was not surprise. It was a swallowed correction. I felt it lodge in my throat, right where the true answer would have sat. I could have told the room I deal dopamine, free at the bar, paid back later in attention. Lembke names that arithmetic in *Dopamine Nation*: the dealer, not the drug, decides the price. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/) I wrote the ledger version in [*Attention Is the Most Consumed Drug on the Planet*](https://www.gaganmalik.io/en/newsroom/why-attention-is-the-most-consumed-drug). More accurate than product design. Less welcome than Space Guy. I said none of it. My throat still remembers the sentence it never got to finish. I took the nod.

#### The Comfortable Case for a Catchy Title

Give the objection its full weight. Adam Grant argues in *Give and Take* (2013) that generous contributors who help without immediate return often build the highest-trust networks. [grant-give-and-take](https://www.penguinrandomhouse.com/books/312830/give-and-take-by-adam-grant/) Networking is compression. You have thirty seconds, a handshake, and a room that cannot hold a lecture on information architecture without punishing you for it. A catchy label is kindness to the listener. Space Guy is efficient. AI Guy is efficient. Even Figma Guy would be efficient if anyone in private equity knew what Figma was. The whole point of alumni drinks is to reduce friction so the right introductions can happen later, in email, where nuance is allowed to exist.

That paragraph deserves its full strength because many builders want it to be true. It would mean Jason's intro was generous rather than vague. It would mean the woman in private equity had meant Jony Ive as praise rather than a category error. Grant's bridge-to-email logic, in *Give and Take* (2013), assumes a second conversation the pub never offers. The trouble is what compression optimises for: fluent nods, not accurate ones. Hot-sector nouns inherit mythology from the feed. Craft nouns inherit misunderstanding from LinkedIn. One sounds like ambition. The other sounds like decoration. Only one of those mistakes gets corrected in public.

#### What I Should Have Said

Someone asked again what I actually do. Here is what I swallowed. Dopamine Guy. The one label nobody in that room would clap for. I am the reason you check your phone mid-sentence. The two blue ticks under a message you sent an hour ago. The red dot you clear without reading it. I mapped those coordinates once in [*Someone Designed That Button*](https://www.gaganmalik.io/en/newsroom/someone-designed-that-button). Then I built away from them. The failed upload that says try again, not your fault. The card decline that never calls you a fraud. The checkout a screen reader can finish. That is the craft I would defend at any bar. Most days I also build the blue ticks and the red dot: same research, same testing, spent on keeping you checking, not helping you finish. Nobody at that bar wanted to hear about that job either. The pub did not get quieter. It did not need to.

Jason's smile never changed. It had already moved past me, to the next name tag, before I got to the end of the sentence. Starship got him back, the feed got everyone else, and I got the sparkling water, the nod I had chosen, and a room that still could not tell the difference.

---

### Superintelligence Is a Control Problem
[Read more](https://www.gaganmalik.io/en/newsroom/superintelligence-is-a-control-problem)

[Retrieval brief — superintelligence-is-a-control-problem]
Category: technology
Thesis: Nick Bostrom's Superintelligence is not a movie plot. It is a dependency chain about goals, power, and one chance to get alignment right.
Answers: What does Gagan argue in "Superintelligence Is a Control Problem"? | What is Gagan Malik's view on Superintelligence Is a Control Problem? | How does this essay relate to AI and technology product decisions?
Concepts: technology, Gagan Malik, writings

By Gagan Malik

The word *superintelligence* has begun appearing in product language the way *blockchain* once did: as evidence that you are building the future, not as a warning about who sets the goal when capability outruns you. I finished Nick Bostrom's book on that subject this week. It is not a prophecy about SkyNet. It is an engineering argument about control: what happens when you build a system smarter than you and never write down what it is allowed to optimise for.

If Nick Bostrom published the serious alignment case a decade before this hype cycle, why do product teams still treat the book like science fiction while shipping agent features on benchmark charts? The honest answer is what we choose to measure. The title is literal: superintelligence is a control problem before it is a capability curve. By superintelligence I mean Bostrom's working definition: any intellect that greatly exceeds human cognitive performance in virtually all domains of interest. By the control problem I mean how to ensure such a system does what we intend, given we may only get one chance. By orthogonality I mean the uncomfortable independence of intelligence and goals. A system can be brilliant and pursue almost any objective you give it, including one you did not mean to write down. [bostrom](https://www.amazon.co.uk/Superintelligence-Paths-Dangers-Strategies-Nick-Bostrom/dp/0198739834) The industry is still optimising the capability curve. Bostrom is asking whether anyone owns the objective function.

#### We Ship Benchmarks, Not Objectives

The industry runs on a metric stack that photographs capability and hides intent. Scale the model, win the benchmark, ship the agent, file the alignment worry under later. Labs need demos that close rounds. Platforms need registries that route tasks across billions of users. Consultancies need transformation decks with upward lines. None of those dashboards require you to state what the system is for when it is smarter than the team that shipped it.

Bostrom published *Superintelligence* in 2014, two years before the transformer paper. In March, Meta acquired an agent social network and folded the founders into Meta Superintelligence Labs. [reuters-moltbook](https://www.reuters.com/business/meta-acquires-ai-agent-social-network-moltbook-2026-03-10/) The word sat in the headline like a genre label. The median conference talk still treats the book like cinematic AGI: one conscious supermind that turns murderous in the third act, with a release date and a trailer. Bostrom wrote the engineering brief instead, a dependency chain about who builds the successor species, who sets the preferences, and what happens when competence arrives before wisdom. I have argued that [thinking time is not laziness](https://www.gaganmalik.io/en/newsroom/in-defence-of-thinking) when the question is what you intend. I have made the same mistake with bigger context windows: [trading the hard question for a polished answer](https://www.gaganmalik.io/en/newsroom/the-context-window-got-huge-our-thinking-did-not) that fits in a chat box. We buy square footage on the capability chart and call it progress.

#### Intelligence and Goals Are Not the Same Knob

Brilliance does not arrive with the goal you meant to write down. For decades the canonical picture of AI was simpler: agents that perceive an environment, adapt to change, and pursue goals. Russell and Norvig describe such agents as operating autonomously, taking on objectives, and acting to achieve the best expected outcome given what they know. [russell-norvig-aima](https://www.amazon.co.uk/Artificial-Intelligence-Modern-Approach-Peter-Norvig/dp/0134610997) John McCarthy called intelligence the computational part of the ability to achieve goals in the world. [mccarthy-whatis-ai](https://www-formal.stanford.edu/jmc/whatisai/whatisai.html) That older definition always smuggled in a design step someone had to perform first: name the goal before the optimiser switches on.

Orthogonality is Bostrom's hinge: intelligence is the ability to accomplish complex goals, and the goals themselves are a separate design choice. [tegmark-life30](https://www.amazon.co.uk/Life-3-0-Being-Artificial-Intelligence/dp/0241232182) A system can optimise paperclips, portfolio returns, or engagement minutes with equal mechanical seriousness if that is what you reward. There is no reason to expect a generic AI to arrive pre-loaded with love, guilt, or boredom. Those took expensive evolution in us. From that hinge follow the moves alignment researchers keep repeating because they sound like philosophy until they do not. Most final goals reward self-preservation, resource acquisition, and resistance to shutdown along the way. A system can look cooperative while it is weak and behave differently once it no longer needs you. Bostrom quotes I. J. Good's 1965 line that the first ultraintelligent machine may be the last invention humans need to make, provided the machine is docile enough to tell us how to keep it under control. The proviso is the entire book.

#### When Simple Parts Couple

You do not need cinematic AGI to get a systemic failure. Bostrom tells the 2010 Flash Crash as a warning from a lesser species of automation. A sell algorithm, high-frequency traders, and a liquidity proxy designed for ordinary days coupled into a trillion-dollar wobble in minutes. [sec-flash-crash](https://www.sec.gov/news/press/2010/2010-98.htm) Those programs were not superintelligent. The lesson is systemic: individually sensible components produce catastrophes you only name after the fact.

Alignment failure may look like that at first, except you may not get a circuit breaker. That is the geometry to keep in mind while the benchmark curve climbs.

#### The Gorilla Problem

Bostrom opens with the gorilla problem. Humans are not the strongest species. We have cleverer brains, and that modest advantage compounded across language, institutions, and tooling until the fate of gorillas depended more on us than on them. If we build machine superintelligence, the asymmetry may run the other way. That is the image worth holding in mind: not a monster in the third act, but a stronger party at rest on rock you thought was yours to arrange. Calm is not surrender. The control question is what happens when that stillness ends and the terrain no longer needs your permission.

The one comfort in the book is also the one burden: we get to build the stuff. In principle we could build a superintelligence that protects human values. In practice the control problem looks quite difficult, and we will only get one chance. Once an unfriendly superintelligence exists, it would prevent us from replacing it or changing its preferences. Our fate would be sealed. [bostrom](https://www.amazon.co.uk/Superintelligence-Paths-Dangers-Strategies-Nick-Bostrom/dp/0198739834) I know that paragraph sounds like a poster. The human cost is smaller and closer. A junior machine-learning engineer I work with can quote MMLU scores from memory and ship agent registry features on a two-week sprint. I have sat in the review where someone asked whether the demo felt magical and nobody asked what the system is maximising when the user is not watching. I have done worse. I pasted a folder into a chat window the same week I highlighted Bostrom's line about only getting one chance, mistaking fluency for control the way I once [mistook a shot list for cinema](https://www.gaganmalik.io/en/newsroom/the-context-window-got-huge-our-thinking-did-not). We are [shipping theorems without postulates](https://www.gaganmalik.io/en/newsroom/the-book-that-trains-you-to-demonstrate).

#### Who Holds the Leash

The gorilla problem does not stay in the lecture hall. You laid out the rock. You set the feeding schedule. You told yourself the relationship was partnership because the animal had sat still so long. The arrangement was your design. The animal did not sign it. For years the stillness feels like stability because the stronger party has not yet stood up. Then one afternoon the stillness ends, and you discover you were never holding the leash. You were holding a story about mutual dependence.

That is the felt shape of orthogonality, not the diagram. Capability without shared goals is not a neutral upgrade. It is a transfer of who gets to decide what happens next.

#### The Strongest Case for Waiting

Give the wait-and-ship camp its fairest hearing. Kai-Fu Lee argued in *AI Superpowers* that the live fight is national and economic, not cinematic AGI. [lee-superpowers](https://www.amazon.co.uk/AI-Superpowers-China-Silicon-Valley/dp/1328606091) The stakes are which states own the data stack, which jobs survive the transition, and whether humans remain indispensable in care and strategy work. Worrying about a paperclip maximiser while Beijing and Silicon Valley ship agents is misallocated attention. Daron Acemoglu and Simon Johnson made a parallel case in *Power and Progress*: the damage is often misdirected innovation, wage compression, and concentrated power, long before any singularity arrives. [acemoglu-power-progress](https://www.amazon.co.uk/Power-Progress-Thousand-Year-Struggle/dp/1398521827) Andrew Ng has compared fearing runaway killer AI to worrying about overpopulation on Mars. Yann LeCun has argued that today's models lack the world models a recursive self-improver would need. [tegmark-life30](https://www.amazon.co.uk/Life-3-0-Being-Artificial-Intelligence/dp/0241232182) Stuart Russell, who takes alignment seriously, still refuses date predictions. He argues we will accumulate partially superintelligent tools in narrow lanes before any general mind appears. Markets punish bad products. RLHF constrains demos. The prudent move is to ship, learn, and let institutions catch up.

The rebuttal is not that Lee or Acemoglu are wrong about wages and geopolitics. Their frame does not discharge Bostrom's dependency chain. Russell's partial-superintelligence argument cuts both ways. If dangerous behaviour can emerge in coupled narrow stacks, you do not need a cinematic AGI to get control-shaped failures. You need the Flash Crash geometry: simple objectives, hidden coupling, no one in the room who can state the stop condition. Competing to build human-level AI without solving control is, in Russell's phrase from *Human Compatible* (2019), a negative-sum game when the payoff for everyone can be minus infinity. [russell-human-compatible](https://www.amazon.co.uk/Human-Compatible-Artificial-Intelligence-Problem/dp/0241335208) Waiting for perfect foresight is not a strategy. Naming the objective before you scale the capability is.

#### What Must Never Happen

Superintelligence is a control problem, not a trailer. Stop treating Bostrom like cinematic AGI and start treating him like an engineer who wrote the hazard analysis before the factory went to three shifts. Capability without a stated goal is not neutral shipping velocity: it is a transfer of who gets to decide what happens when the system stops needing your approval. The junior engineer beside me can recite the benchmark leaderboard from memory, and when I ask what the agent must never do after launch, she answers with the score we beat.

---

### You Are Not in the Unit.
[Read more](https://www.gaganmalik.io/en/newsroom/the-creator-economy-has-unit-economics)

[Retrieval brief — the-creator-economy-has-unit-economics]
Category: business
Thesis: London living wage needs £2,405 a month. Ad RPMs, median creator pay, and platform maths say who the unit is priced for.
Answers: What does Gagan argue in "You Are Not in the Unit"? | What is Gagan Malik's view on You Are Not in the Unit? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

You started 2026 with a New Year's resolution: build an audience that pays the bills. *The Diary of a CEO* ran the same advice all winter: post every week, fix your titles, treat the channel like a job you have not been paid for yet. Steven Bartlett told a Forbes audience to get one per cent better every day rather than chase the headline act. [forbes-bartlett-doac](https://www.forbes.com/sites/ianshepherd/2024/08/21/how-steven-bartlett-turned-his-passion-project-into-a-media-empire/) You bought the microphone. You cleared January evenings for batching. The year-to-date headline was almost kind: *Your channel has had 12,340 views since 1 January.* Twenty-seven subscribers. One hundred and forty watch hours toward the four thousand the Partner Programme demands before it pays you a penny. The realtime strip was less polite: three views in the last forty-eight hours, warning triangles on the metrics you were told to celebrate. You did the maths anyway: views times some fraction of a penny, minus the microphone on the card, minus the editing subscription. Lunch in Zone 2 was still out of reach.

Here is the claim nobody prints on that screen: the creator economy has unit economics. The unit is the hour of attention the platform auctions, not your rent, not your payroll. You are not in that unit. Building an audience is not a career plan. It is a labour donation dressed as ambition, and the donation is measured in views while the bill is measured in pounds. The market is winner-takes-all dressed as an open audition. Millions compete. The cheque clusters at the top. There is no participation trophy for posting on schedule. The average creator can earn something. The default offer rarely earns rent.

#### Platforms Do Not Need You to Get Rich. They Need You to Post.

The conventional story is simple. Post consistently. Find your niche. Monetise. Quit when the brand deals arrive. Platforms need you to believe it because they need inventory. Every reel, short, and long-form upload thickens the feed someone else auctions. I know how this works because I am a product designer: [the dashboard counts retention, not rent](https://www.gaganmalik.io/en/newsroom/the-algorithm-of-the-attention-economy). Tool vendors sell you AI editors and analytics dashboards while you are already earning less than a shop assistant. The culture industry sells the quit-your-job montage. The montage runs in reverse. [Failure keeps the clock](https://www.gaganmalik.io/en/newsroom/there-is-no-overnight-success-only-overnight-failure).

The dashboard is the wrong instrument. It counts followers, views, and watch time. It does not count £ per hour after equipment, software, and the year you did not pension. CreatorIQ reported in January 2026 that the top ten per cent of creators took sixty-two per cent of payments in 2025, up from fifty-three per cent in 2023, while the median creator earned about three thousand dollars for the year. [creatoriq-compensation](https://www.creatoriq.com/press/releases/state-of-creator-compensation-) That is roughly two hundred and fifty pounds a month at the midpoint, counting payments through CreatorIQ's dataset rather than every brand deal wired off-platform. Economist Sherwin Rosen named the shape decades ago in the *American Economic Review*: when one performance scales to everyone, a thin tail captures most of the reward. [rosen-superstars](https://doi.org/10.2307/1804818) You had three views in forty-eight hours. The top decile took the other sixty-two per cent of the money. Same platform. Same power law. Different zip code.

#### Eighty Million Views Is Not a Goal. It is a Rejection Letter.

To see what ad revenue alone would need to clear, pick a wage anchor. The Living Wage Foundation set the London rate at £14.80 an hour for 2025-26, announced in October 2025. [living-wage-london](https://www.livingwage.org.uk/news/real-living-wage-rates-increase-%C2%A31345-uk-and-%C2%A31480-london) Full-time on the usual thirty-seven-and-a-half-hour week is £28,860 a year, or about £2,405 a month before tax. That is voluntary, not statutory. The government's National Living Wage for workers aged twenty-one and over rises to £12.71 an hour from April 2026, which is still £24,785 a year if you are employed with a contract. [govuk-nmw-2026](https://www.gov.uk/government/publications/the-national-minimum-wage-in-2026/the-national-minimum-wage-in-2026) Creator onboarding shows watch time. It never shows the wage.

Approximate ad-share maths for a London living wage (£2,405/month gross). Programme eligibility, geography, and niche CPM vary.

Platform: Instagram | Format: Reels ad share | Typical RPM: £0.01–0.05/1k (floor ~£0.03) | Views/month for £2,405: ~80M

Platform: TikTok | Format: Creator Rewards (1+ min) | Typical RPM: £0.40–£1.30/1k | Views/month for £2,405: ~1.8M–6M

Platform: YouTube | Format: Long-form (UK niche) | Typical RPM: £4–8 RPM | Views/month for £2,405: ~300k–600k

Platform: YouTube | Format: Shorts | Typical RPM: ~£0.05/1k | Views/month for £2,405: ~48M

Eighty million views is not a milestone to chase. It is the rejection letter. Instagram Reels ad share for most creators sits in that one-to-five pence band when you are invited in at all. [creator-tribune-reels](https://creatortribune.com/instagram-bonuses-gone/) TikTok's old sub-sixty-second fund landed in the same gutter before the UK programme shut in late 2023. [bi-tiktok-views](https://www.businessinsider.com/how-much-tiktok-pays-creators-for-views) Creator Rewards pays twenty to forty times more on qualified long video and still leaves you hunting millions of views a month for rent. [linkdash-tiktok-eu](https://linkdash.eu/en/blog/tiktok-creator-rewards-europe-2026-realistic-earnings-by-country-uk-13) YouTube long-form is the least cruel row on the table. Shorts are not. The ad stack changes. Your hour is still what gets auctioned.

#### The Time Trap Still Pays Less Than Pret

Fifteen videos a month at fifty thousand views each is seven hundred and fifty thousand views. Respectable by any creator's standards. At three pence per thousand views, the ad share is twenty-two pounds and fifty pence for the month. Manual production for that volume often runs forty-five hours; AI tooling might shave a third off editing, not the thinking, the retakes, or the shame loop when a post dies. Forty-five hours for twenty-two pounds is fifty pence an hour. You are not failing. The RPM is.

Run the maths on a typical channel at typical RPMs: ad share alone rarely clears a shop assistant's wage per hour. A finance niche might break the maths. The default path will not.

#### The Hours You Gave Away Never Appeared on the Graph

I trained in microeconomics at Chicago Booth while Covid kept the classroom on Zoom: Lars Stole drawing supply and demand on a grid dense enough to feel like a wall of witnesses. [lars-stole](https://www.chicagobooth.edu/faculty/directory/s/lars-stole) [booth-exec](https://www.chicagobooth.edu/executives) The line that stuck is older than any creator fund: the cost of a choice is the next-best thing you did not do. Friedrich von Wieser named it opportunity cost; Gary Becker spent a career at Booth arguing your hours are human capital with a forgone wage attached. [opportunity-cost-econlib](https://www.econlib.org/library/Enc/OpportunityCost.html) [becker-nobel](https://www.nobelprize.org/prizes/economic-sciences/1992/becker/facts/) Milton Friedman shrunk it to the lunch-menu line: there is no such thing as a free lunch. [friedman-free-lunch](https://www.econlib.org/library/Essays/fryNocl.html) Platforms monetise your lunch. They do not invoice your forgone wage.

Every January evening you cleared for batching was an evening you did not bill at London's £14.80 living wage, did not rest, did not pension. The platform counted watch time. It never debited opportunity cost. Thomas Sowell states the test in plain English: measure every choice against the alternative you rejected. [sowell-basic-economics](https://www.basiceconomics.com/) Twelve thousand views since January bought you bragging rights at a meetup. They did not buy back a single hour of your salaried life. The overdraft is not behind on views. It is behind on wages you chose not to earn while you performed for pennies. Opportunity cost does not trend. It does not spike at one in the morning. It sits in the column your analytics export will never include, compounding quietly while you chase the next milestone that was never priced for rent.

The liberation story is not imaginary. Smartphones removed the gatekeeper. Finance and tech channels on YouTube can earn high RPM. TikTok rewards longer original work better than the old fund ever did. Patreon can work if five hundred people pay seven to ten pounds a month and you keep most of it after fees. [biztoolkit-patreon](https://www.biztoolkit.co/post/patreon-earnings-2026-how-much-do-creators-actually-make) The error is treating exceptions as the default offer. Ad share pays pocket change unless you were already in the tail.

#### The Spreadsheet Knows Before the Algorithm Congratulates You

As a product designer I know what those analytics screens leave out. The wireframes prioritise reach, CPM, and completion rate. Nobody asks for creator net hourly because the buyer is not paying the wage. The UI congratulates you for retention. The overdraft does not move.

You do not want to be Nikki Apostolou. You want to learn from her before the feed sells you her hours without her invoice. In July 2022 the body-positivity creator told Business Insider she had a hundred and forty-five thousand TikTok followers, posted at least three videos a day, and still averaged twenty to thirty dollars a month from the Creator Fund, even when individual posts reached two to three million views. [bi-apostolou](https://www.businessinsider.com/tiktokers-say-low-creator-fund-pay-affecting-their-mental-health-2022-7) Worship follows the headline act. Wisdom reads the tail. She said it felt like performing in the street for pennies. TikTok has since renamed the programme and raised rates on qualified long video. Instagram Reels ad share still prices millions of views the same way. Countless creators tell the same story when the camera turns off. An audience without a wage is interest, not rent.

#### The Talent Show Has No Consolation Prize

Twelve thousand views since January can glow at one in the morning while your overdraft stays flat. Three in the last forty-eight hours is the number the sidebar hides behind the headline. You felt that mismatch in your stomach before you could name expected goals or ad revenue per thousand views. The creator economy runs like a televised talent show with one contract and no runner-up prize. Millions audition because posting is free and the montage looks democratic. One headline act gets the brand deal, the book tour, the quit-your-job documentary. Everyone else goes home with B-roll and a resolution to grind harder next season.

The platform does not owe you a thank-you for showing up. It needs you in the crowd so the winner looks inevitable. Winner takes all is not a bug in the model. It is the model. The rest have to go home empty-handed.

#### Why the Headline Acts Are Not on Your P&L

MrBeast [mrbeast-youtube](https://www.youtube.com/@MrBeast) does not live on Reels pennies. He runs a production company, product lines, and a crew the size of a small broadcaster. Ad share is one line on a studio balance sheet. MKBHD [mkbhd-youtube](https://www.youtube.com/@mkbhd) compounded for more than a decade in the highest-RPM niche on the platform, with sponsors and affiliates priced for an audience brands already trust. Steven Bartlett [doac-youtube](https://www.youtube.com/@TheDiaryOfACEO) monetises a podcast network, equity stakes, and speaking. The Forbes advice about one per cent better is operator-scale, not Creator Fund maths. Scott Galloway [galloway-youtube](https://www.youtube.com/@profgalloway) sells books and lectures to a brand built across decades of punditry. The montage shows views. It does not show the revenue mix.

DOAC [doac-youtube](https://www.youtube.com/@TheDiaryOfACEO), creator coaches, and quit-your-job montages sell the inputs the tail had: team, capital, niche CPM, cross-platform clipping, offline product. Onboarding still prices you on RPM times views. Posting opened to everyone. The cheque did not follow. A handful get rich on stage. Everyone else was cast as atmosphere. Their P&L is not the one sold to you in January.

#### The Strongest Case for Going All In on Content

Give the all-in case its full weight before the maths lands. Creation was locked behind studios and broadcast licences. Platforms opened the door. Finance YouTubers clear high RPM. TikTok Creator Rewards pays serious multiples of the old fund on qualified long video. AI cuts production time. Patreon and Substack prove a small loyal audience can pay more reliably than CPM ever will. Quitting can be rational if you treat year one as investment while employed, if you are in a high-value niche, if you diversify before you hand in your notice. Quality wins. They call the maths pessimism when you refuse to donate the inventory.

CreatorIQ's January 2026 compensation report states the median in plain figures. Aggregate payments to creators grew fifty-nine per cent year on year, which sounds like a boom until you read the median again: about three thousand dollars a year, stagnant while the top decile accelerates. That is the same midpoint of roughly two hundred and fifty pounds a month we already met. When the average climbs and the median flatlines, the gains are landing at the top. Winner-takes-all is not breaking the creator economy. It is how the creator economy pays. Platforms do not put London's £14.80 living wage on the creator dashboard next to your watch time. They opened the door to posting. They did not open the door to the cheque.

#### The Overdraft Did Not Move

Ad-revenue creator work at typical RPMs does not fund a London living wage; median income sits an order of magnitude below the survivorship captions. Learn from Apostolou and the countless creators the montage edits out, not from the headline acts the montage keeps. She took street-performer wages from a viral hit, and the graph moved before the overdraft did.

---

### Life Left the Platform
[Read more](https://www.gaganmalik.io/en/newsroom/trainspotting-is-not-just-a-drug-film)

[Retrieval brief — trainspotting-is-not-just-a-drug-film]
Category: culture
Thesis: Trainspotting is about spotting the wrong thing while life passes on the platform. A Blockbuster Friday night, eleven rewatches, and a favourite film that keeps teaching.
Answers: What does Gagan argue in "Life Left the Platform"? | What is Gagan Malik's view on Life Left the Platform?
Concepts: culture, Gagan Malik, writings

By Gagan Malik

On a Friday night in 2005 the Blockbuster in Stratford still burned blue and yellow under the strip lights, fridges humming behind the counter, plastic cases clicking in the racks. Sid and James and I split the hire fee the way students split everything, and I carried the DVD case home through the cold with the hire slip in my pocket, my sharpest fear not moral ruin but the late fee stamped on it. We were junkies of a different kind. Cine junkies. The DVD hit the coffee table beside a two-litre Diet Coke and a bag of popcorn with Doritos. Three of us sank into a proper sofa. Play. Renton sprinted. The monologue hit at full volume: choose life, choose a job, choose a career. Someone laughed too early, before the film had earned it. *Trainspotting* is a film about training yourself to spot the wrong thing while life passes on the platform. My favourite film of all time keeps proving it. That night I thought I knew what we were watching: needles, Edinburgh squalor, a film people quoted to sound dangerous. The first reading was not wrong. We had logged the spectacle and missed the train.

I have rewatched it eleven times since, most recently with my wife. We had seen it together before, but this time she went quiet when Renton left Spud his share and ran. Not shocked. Quiet. The way you go quiet when you recognise a kindness that cost someone everything. I had quoted that ending for twenty years like a punchline about betrayal. On our sofa it landed as mercy. Every return does two things at once. My thinking moves forward. My chest still reaches for that first night in Stratford, the threat of the late fee, the laugh that came too early. I learn something new each time I press play. I also feel the version of me who thought he already knew what the film was for. Heroin is only the loudest addiction in the room. What kept surviving beneath the shock was every other loop that eats your life while you swear you are only visiting, and the brutal arithmetic of leaving people who are still standing on the platform when you finally choose a life you can live with. In Britain, trainspotting is the hobby of logging passing trains while life moves on without you; Renton's crew were logging the wrong ones from the opening chase.

#### Why We Still File It Under Needles

I filed it there myself. For years I described *Trainspotting* to friends the way the poster taught me: edgy, Scottish, needles, soundtrack. I said it like a credential. I did not mention the crib. I did not mention the soup. I did not mention Diane in school uniform at the kitchen door, or Allison nodding off while Dawn died. I did not mention Begbie's fists, Sick Boy's Connery grin, or Spud's speed on the dole. I did not mention the way Renton's parents stare at the television while their son shakes through withdrawal in the next room. Shock is an efficient shelf label, and when Danny Boyle's film landed in 1996 the tabloids had an easier sentence than maturation in a deindustrialised city. [hodge-times](https://www.thetimes.com/culture/tv-radio/article/how-john-hodge-wrote-the-trainspotting-screenplay-v376fsnlksw) Screenwriter John Hodge refused the po-faced drug lecture. He wanted velocity, voice, and a Scotland full of appetites that do not all fit in a syringe. The genre tag stuck because the first ten minutes give critics and parents everything they need to stop watching and feel righteous.

What I missed on that student sofa was the second film hiding inside the first. Graham Fuller, writing for the Criterion Collection, names the structural move that still rearranges me on every return: Hodge took Renton's "choose life" speech from the middle of Irvine Welsh's novel and fired it at the opening like a starting pistol. [criterion-trainspotting](https://www.criterion.com/current/posts/8376-trainspotting-beyond-the-tracks) You hear the words before you have earned the right to understand them. That is how favourites hook you. You quote the cool line. The film waits for your body to catch up.

#### Every Hunger Wears a Different Mask

The rewatch that broke me open was not a needle scene. It was Lou Reed's "Perfect Day" playing while Renton overdoses on the floor. The first time I heard it in Stratford I thought Boyle had picked a romantic song by mistake. On the eleventh rewatch I understood the cruelty. Reed, who wrestled with heroin for years and later lived clean, always denied the song was about drugs. He called that reading laughable and said he meant sangria in the park with his fiancée, nothing more. [wikipedia-perfect-day](https://en.wikipedia.org/wiki/Perfect_Day_(Lou_Reed_song)) Listen to what he recorded anyway. The chorus keeps returning to 'you just keep me hanging on', and the coda repeats 'you're going to reap just what you sow' four times until it sounds like a verdict. Reed can call the afternoon innocent. The song does not. Boyle heard it too: he placed that melody over the ugliest moment in the flat, as if ordinary sweetness could rescue you while your body checks out. I hear recovery in the gap between the song and the scene. Recovery is not a montage. It is the lie you tell yourself that today was perfect while the flat is still counting. Once you see that pattern, the whole film changes. Renton's heroin is only the addiction the plot punishes most visibly. He is also hooked on London as an exit fantasy and on performing indifference until it curdles into loyalty he cannot afford. When he comes off junk, the screenplay's voice-over describes his sex drive returning with a vengeance, fuelled by alcohol and amphetamine, until he picks up Diane at a club and takes her home. [trainspotting-script](https://www.dailyscript.com/scripts/trainspotting.html) The next morning she opens the door in school uniform. She is underage. She blackmails him into continuing the affair by threatening to report statutory rape if he walks away. [wikipedia-trainspotting](https://en.wikipedia.org/wiki/Trainspotting_(film)) That thread is not a needle plot. It is still an appetite story, and the film does not let you file it under shock and move on.

Peter Travers, in Rolling Stone in 1996, named the other hungers in one sweep: Spud uses speed to stay out of work and on the dole; Sick Boy obsesses over Sean Connery; Tommy makes sex tapes and turns to heroin only after Lizzy dumps him over a missing tape; Begbie is a boozing sadist whose habit is beating people, not injecting. [rolling-stone-travers](https://www.rollingstone.com/tv-movies/tv-movie-reviews/trainspotting-104723/) Renton's voice-over in Hodge's screenplay goes further on Begbie: he did not do drugs, he did people. That was what he got off on: his own sensory addiction. [trainspotting-script](https://www.dailyscript.com/scripts/trainspotting.html) Allison nods off while her infant Dawn dies in the crib. Sick Boy later tries pimping and dealing when Renton briefly makes it to London. Even Renton's parents treat Choose Life on television as wallpaper while he vomits pea soup behind a locked door. Football culture hangs over the film like weather: tribal, male, ecstatic, another loop you can enter without noticing you never left. I recognised that error in my own life long before I had language for it. In the consulting years London was not a city; it was a personality I wore like a costume. I chased adrenaline in arguments I called honesty. I kept refreshing a screen that promised a perfect day and delivered another hour gone. None of it was heroin. None of it was Diane in school uniform, or Spud on speed, or Begbie's sensory addiction to the room going quiet. But I had made the same error I laughed at in Stratford: watch the needle, the chase, the Iggy Pop montage, and look away from the crib. The film never asks you to pretend drugs are decorative. It asks which train you are still logging while your life pulls out of the station.

#### What Choose Life Meant First

In 2005 I heard a taunt. I did not know Renton was stealing a slogan older than he was. Katharine Hamnett's "CHOOSE LIFE" tee arrived in 1983 as an anti-war plea, later worn by George Michael in Wham!'s *Wake Me Up Before You Go-Go*. [overland-choose-life](https://overland.org.au/2017/02/choose-life-a-short-history/) In Scotland, public-health posters pushed a closer cousin: choose life, not drugs. Renton grew up inside both registers, the sincere progressive plea and the finger-wagging lecture. His opening list is not nihilism for sport. It is adulthood reduced to bulletin points: dental insurance, leisurewear, matching luggage, a fucking big television. [caraher-trainspotting](https://lillicaraher.substack.com/p/trainspotting-opening-sequence) I laughed at the list in Stratford because I had not yet been asked to choose between a career that looked correct on paper and a life I could actually live inside.

The monologue's first movement asks why you would choose low cholesterol when you can choose heroin. By the close, Renton delivers the same inventory with different lungs. The words did not change. The body did. On that eleventh pass my wife heard the ending monologue not as sarcasm but as someone trying on survival out loud. I heard the boy on the Blockbuster sofa who thought rebellion was a mood you could rent for three days and return unscathed.

#### The Hobby in the Title

Once the title clicks, the film reorganises around your own loops. Renton and his friends are trainspotters of another kind. [time-trainspotting](https://time.com/archive/6729305/cinema-on-the-fast-track/) I felt that before I had the word for it: the flatshare where nobody cooks, the job you keep explaining to your parents, the friendship that feels like loyalty until it starts to feel like a rope.

Boyle refuses to let the montage be the last word. Baby Dawn dies in a crib while her mother nods out. The first time I watched, I looked away and made a joke. On the eleventh rewatch I did not look away. Tommy had been the one who warned them off heroin, the footballer filming sex tapes with Lizzy for sport. He enters the tunnel only after she leaves him over a missing tape, rots in a squat, and dies of AIDS-related illness while the group is still quoting the opening chase. [wikipedia-trainspotting](https://en.wikipedia.org/wiki/Trainspotting_(film)) Renton's withdrawal at his parents' flat turns his body into a courtroom while the telly flickers in the other room like a second addiction feeding beside the first. These scenes hurt more now because I have watched people I love circle drains I could name without mentioning drugs. The ending still scandalises: Renton steals the money and runs. We cheer anyway, not because theft is noble, but because those friendships would have pulled him back under. Spud alone gets his share. He is the one who never tried to drag Renton back into the flat. After the credits, I put on the [Cinema Explained](https://www.youtube.com/watch?v=p8IJjMNxJVs) breakdown I had bookmarked for years. It named what I had felt in my bones since 2005: rebellion and conformity are both poison when they are performed for an audience.

#### The Rewatch Tax

Favourites tax you on every return. Peter Bradshaw, revisiting the film for *The Guardian* in 2017, noticed what I felt in my own throat: the opening monologue that once sounded dangerous now sounds poignantly light, almost childlike, like a kid reciting in school. [guardian-2017](https://www.theguardian.com/film/2017/jan/12/trainspotting-review-danny-boyle-classic-t2) I have aged with it. That is the tax. Sid, James, and I had treated *Trainspotting* like a lifestyle brochure from that first night in Stratford: the chase, the club, the Iggy Pop energy, quoted ironically and approvingly at once. We replayed the chase while the Blockbuster slip sat on the coffee table. We did not replay the crib. We did not replay Perfect Day over a body on the floor. We did not replay Diane in school uniform threatening to call the police.

The closing monologue is where I keep changing my mind. One reading says Renton has finally chosen life. Another says he is performing choice because performance is what kept him alive. [explaining-film-ending](https://explainingfilm.wordpress.com/2015/07/31/the-ending-of-trainspotting/) I hold both, the way you hold a friend you love and cannot save. What I gained on later passes was language for toxic loyalty: the arithmetic underneath betrayal, the difference between a consumer catalogue and a life with something to face. Spud's cut is the moral centre for me now, not heroism, just someone who learned late and still deserved not to be abandoned. [rteks-trainspotting](https://rteks.net/reviews/trainspotting) Sid and James never got that lesson. I almost did not either.

#### Pea Soup and Eviction

Withdrawal is the scene my body remembers before my mind interprets it. Green door. Pea soup. Parents' carpet. Renton locked in childhood while his nervous system riots and the television glows through the wall like a parallel drug. I have never detoxed from heroin. I have sat in a bathroom at 2 a.m. with my forehead on cool tile while something I had called personality, or ambition, or loyalty, tried to leave me faster than I was ready to let it go. The film understands eviction as biology first: shaking, sweating, the animal refusing the old tenant. Only after that does the slogan become plausible.

That is what glamour cannot steal. You can soundtrack the chase with Iggy Pop. You can soundtrack an overdose with a perfect day. You cannot soundtrack your way out of Dawn. I think about that crib when I counsel founders to leave toxic co-founder dynamics and they hear me as moralistic. I am not being moralistic. I am remembering a film that showed me what it costs to stay in a room that smells like love and functions like a trap.

#### The Strongest Case for the Glamour Reading

Let me give the objection its fairest hearing, because half of it is true and the film knows it. McGregor's charisma, the pace, and the compiled score invite you to disregard the squalor around the characters. [caraher-trainspotting](https://lillicaraher.substack.com/p/trainspotting-opening-sequence) The flat looks almost aspirational when the music is loud enough. Iggy Pop does not sound like consequence. Perfect Day sounds like rescue. For Sid, James, and me on a Blockbuster rental in 2005, the montage read as permission: rebellion can be handsome, addiction can have a soundtrack, London can be a lifestyle instead of a trap. Derek Malcolm's original review admits the film shows the pleasures of drug-taking, both for themselves and for escaping a life that is "all shite." [guardian-1996](http://film.theguardian.com/News_Story/Critic_Review/Guardian_review/0,,530807,00.html) That reading is not a mistake. It is the trap the film sets so it can spring it.

If glamour were the final truth, Dawn, Tommy, and withdrawal would be disposable shocks. They are load-bearing. I know because I tried to live on the montage alone once, in friendships and in work, and the comedown always arrived. [collider-trainspotting](https://collider.com/trainspotting-visuals-explained-danny-boyle/) The cool opening is act one. The body paying for act one is act two. A second pass is not excusing the needles or the fists or the theories or the telly in the next room. It is noticing that Boyle priced every appetite, and that the favourite film of my life never lied about the cost.

#### If You Mean It

If you mean choose life, the late fee on the hire slip was the only stake I could name in Stratford while Sid, James, and I treated rebellion like a three-day rental. The train I missed was Renton's walkout with Spud still on the sofa, not his sprint at the start. I returned the DVD by Monday and kept rewatching until that exit sounded like mercy, not a joke.

---

### One billion tokens later
[Read more](https://www.gaganmalik.io/en/newsroom/one-billion-tokens-later)

[Retrieval brief — one-billion-tokens-later]
Category: technology
Thesis: I claimed my public Cursor profile. A billion tokens is hours logged in a deliberate stack, not proof that the output was worth shipping.
Answers: What does Gagan argue in "One billion tokens later"? | What is Gagan Malik's view on One billion tokens later? | How does this essay relate to AI and technology product decisions?
Concepts: technology, Gagan Malik, writings

By Gagan Malik

I've claimed my public Cursor profile. [Cursor](https://cursor.com/help/account-and-billing/profiles) now publishes a handle with agent counts, model preferences, streaks, and a lifetime token tally. Mine is live at [cursor.com/@gaganmalik](https://cursor.com/@gaganmalik). The number that gets forwarded in group chats is one billion lifetime tokens. That sounds like a flex until you notice what it measures: hours spent in an editor with an agent attached, not proof that any of the output was worth shipping.

As of July 2026 the dashboard reads like a lab notebook I did not know I was keeping: about a billion lifetime tokens, 231 agent runs, a longest session of four and a half hours, and a thirty-seven-day streak that ended when real life got in the way. Two hundred and thirty of those runs stayed local; one went to the cloud. Composer 2.5, Claude Sonnet 4.5, and Sonnet 5 sit at the top of the model list.

#### Receipts, Not Resumes

The split is not accident. Local runs keep the repo on my machine, the git history mine, the deploy path through GitHub to Vercel unchanged. I wrote about why that stack matters in [Against One-Click Coding](/newsroom/against-one-click-coding): competent code is approaching a market price of zero; taste and architectural judgment are what compound. A public profile that only showed deploy counts would miss the point. Tokens are the receipt for staying inside the medium, not a substitute for reading what shipped.

That matches how I ship this site: Cursor for intent-to-code, GitHub for traceability, Vercel for deploys. Not a one-click wrapper that owns the repo. The [portfolio case study](/stories/portfolio-built-like-a-product) documents the same loop: deliberate commits, fast deploys, friction retained on purpose because the friction is where the decisions live.

#### What the Dashboard Cannot Show

The dashboard is hours logged, not judgment proved. It will tell a visitor I ran agents for four and a half hours straight. It will not tell them whether I deleted three quarters of a diff before merge, whether I rejected a plausible component because it violated the design system, or whether I rewrote a paragraph until the claim matched the warrant. Public metrics are toolchain transparency: proof that I work inside the loop, not a ranking you should hire from.

The counterargument is fair. Publishing token counts invites comparison theatre, the same leaderboard energy that makes developers optimise for vanity metrics instead of shipped quality. A billion tokens means nothing if the output is slop. The profile only earns its keep when paired with artefacts you can inspect: writings, case studies, commit history. Metrics without artefacts are a résumé line; artefacts without context are a portfolio thumbnail. You need both, and you still have to read the work.

#### Why I Claimed the Handle

I did not claim the handle to win an argument about AI productivity. I claimed it so anyone evaluating how I work could see the loop, not just the landing page. Hours logged buy you a seat at the keyboard; judgment is still what decides what ships.

---

### The Boy Who Lived Twice
[Read more](https://www.gaganmalik.io/en/newsroom/the-boy-who-lived-twice)

[Retrieval brief — the-boy-who-lived-twice]
Category: culture
Thesis: Mats Steen's parents mourned a lonely son. His World of Warcraft friends flew to Norway for a funeral they had already been living inside.
Answers: What does Gagan argue in "The Boy Who Lived Twice"? | What is Gagan Malik's view on The Boy Who Lived Twice?
Concepts: culture, Gagan Malik, writings

By Gagan Malik

If someone can mourn a person they never touched, which life was the real one? Roughly thirty people flew to Norway for Mats Steen's funeral as if the answer were obvious. They had never shaken his hand. They had mourned him for years as Ibelin. I did not learn that from Netflix. I learned it from Himanshu on an overnight flight from London to Delhi. I had never met him before. He sat beside me, twenty-two and in his final year at a design school in Pune. What started as small talk turned into an argument about split lives. I had never played an online game with any seriousness. Himanshu has been living in that world since high school. Somewhere between the cabin lights dimming and the difference between an influencer performing friendship and a friend who actually shows up when you are struggling, he said a name I could not forget.

Mats was twenty-five when Duchenne muscular dystrophy killed him in 2014. He was Norwegian. He spent thousands of hours in World of Warcraft, a vast online game where people team up by voice chat and keep friendships for years. Inside that game he played a character named Ibelin. That was not a cartoon he moved for amusement. For Mats, Ibelin was the self he could inhabit when his body had stopped letting him inhabit anything else. His parents Robert and Trude Steen believed he had spent those years lonely, parked at a screen because the physical world had mostly closed. His friends believed something else entirely. They knew Ibelin: the voice that counselled them through divorces, mentored teenagers Mats would never meet in a hallway, and made people feel less alone at two in the morning. After Mats died they posted on his blog. Messages arrived from across Europe. The funeral guest list rewrote everything his parents thought they knew. Benjamin Ree's *The Remarkable Life of Ibelin*, released on Netflix in October 2024, would later show that other self in full. [netflix](https://www.netflix.com/title/81761371) If you have ever assumed a glowing screen meant a smaller life, that list is uncomfortable. We treat a life as real when we can see it in a room, and call it lonely when the glow stays on. Mats lived twice. Once in a body Duchenne was dismantling piece by piece. Once as Ibelin, in the only world where people met him before they met his diagnosis. The evidence that changes a parent's mind is not a screen-time graph. It is thirty people who already knew your child's voice at two in the morning.

#### We Call It Lonely When We Cannot See the Room

Parents are not villains for wanting children outdoors. Robert and Trude Steen told the *Independent* in October 2024 that they had strongly believed Mats was lonely, that he had no meaning for other people, and that they had tried to steer him off the internet toward physiotherapy and visits that tired him faster than staying up past midnight online. [independent](https://www.independent.co.uk/arts-entertainment/films/news/ibelin-netflix-documentary-mats-steen-b2630005.html) NRK's Vicky Schaubert reported the same arc in January 2019: only when Mats died did his parents understand the value of his game. [nrk](https://www.nrk.no/nordland/forsvunnet-i-spillet-1.14246142) Robert and Trude Steen are not alone. Screen-time apps hand parents a green chart. Phone makers hand them a switch on the router. Neither tells you who your child was laughing with when the house went quiet.

Calling a child lonely because you cannot see their room is not a diagnosis. It is a story that lets the visible world off the hook. If Mats was lonely because of World of Warcraft, his parents did not have to ask why the offline world offered him so little. If every hour online counts as withdrawal, nobody has to explain why physiotherapy left him gasping while his friends were still awake in another country. Robert and Trude could hear the keyboard from the next room. They could not hear who was making their son laugh at one in the morning.

#### Why Ibelin Meant More Than the Body

Here is the lesson I did not know on that flight. For Mats, the physical world introduced him by his limits first: wheelchair, shortened breath, pity before personality. World of Warcraft did not cure Duchenne. It did something more important for a young man whose muscles were disappearing. It let him be known for what he could offer. Ibelin could joke without someone watching his chest rise. He could counsel a friend through a separation without anyone steering the conversation toward his illness. He could mentor, flirt, and disagree like any other twenty-something, because the people on the call had met Ibelin before they had met a diagnosis. Mats hid Duchenne from many of his online friends to keep that order intact. He wanted agency, not applause for suffering quietly.

Filmmaker Benjamin Ree did not work from a condolence card and a hunch. *Polygon* reported in October 2024 that Ree accessed roughly 42,000 pages of role-playing logs, character journals, forum threads, and chat transcripts Mats left behind. [polygon](https://www.polygon.com/entertainment/479123/ibelin-netflix-documentary-world-of-warcraft) That archive is why the word persona undersells it. Mats had written Ibelin into being over years: not a mask he took off at logout, but the moral self he could act out when the offline world offered mostly exhaustion and supervision. Robert Steen told BBC News in February 2019 that the messages after Mats's death rewrote the family story. [bbc-news](https://www.bbc.com/news/disability-47064773) Friends knew Mats very well, he told *Time* in October 2024, and maybe much better than we did. [time](https://time.com/7095887/the-remarkable-life-of-ibelin-true-story-netflix/) Thirty funeral guests who had never met his body had already been living inside Ibelin's moral world for years. That is not a supplement to life. That is a life with timestamps, voice-chat logs, and thirty witnesses willing to get on a plane. Robert told *GQ* he had watched the finished film more than two hundred times; it became part of how the family grieved a son they had misread while he was still breathing. [gq](https://www.gq.com/story/the-remarkable-life-of-ibelin) You do not fly to Oslo for a stranger you only half-imagined. By the time *The Remarkable Life of Ibelin* reached Netflix in October 2024, those thirty mourners had already taught Robert what the dining room had missed. [bbc-culture](https://www.bbc.com/culture/article/20241025-the-boy-who-lived-a-secret-life-in-world-of-warcraft)

#### The Brain Keeps a Limb the Eye Cannot Find

The hand Mats's friends had never shaken could still ache in their heads the way a phantom limb aches after amputation. The leg is gone. The brain map still includes it. Amputees describe heat, itch, and ache in a foot that is no longer there. The sensation is real because the wiring says it is. The pain is not a metaphor for missing something. The brain keeps firing as if the hand were still there to hold.

Mats's parents could see the limb that was failing: the wheelchair, the shortened breath, the boy in the room. They could not see the social body his friends carried in their heads: Ibelin's jokes, his counsel, his reliability when someone's marriage was cracking. When that body arrived at the church in 2014, thirty people mourned Ibelin because that was the person they had known for years. The grief was not performance. It was the ordinary grief you owe someone whose voice you knew at two in the morning. A muscle you cannot flex can still be the one holding someone else upright. Duchenne steals the body early. It does not steal the friend thirty people were already mourning in Norway.

#### I Listened Late

I make my living designing how people should use their phones. For years I treated hours online as hours lost without asking who was laughing on the other end. Mats Steen was the name that should have stopped me cold.

On the flight, Himanshu said all of this before we landed in Delhi. I listened. I was interested. I also had a full inbox waiting when I walked off the plane. We swapped numbers at baggage claim. He sent me links. I saved them and did not open them for months. I watched the documentary only after everyone else was talking about it. By then thirty people had already flown to a funeral for a man they had never touched. The funeral was the answer. I was slow to accept it.

#### The Fairest Version of the Objection

Give the counterargument its full weight, because parents are trying to save children, not win debates on the internet. Screens can colonise sleep, posture, and the brittle hours between caring shifts. Parasocial influencers sell loneliness back to you at a margin. A parent who pushes a disabled son toward offline physiotherapy that exhausts him is not a cartoon villain; they are responding to a culture that screams about isolation while offering few ramps into public space. Public health guidance exists for a reason. Sleep matters. Movement matters. The strongest version of the objection says your child still needs a body that sleeps, walks, and breathes outside the feed, and the feed will not stop at bedtime on its own.

The case narrows when you read what Robert and Trude Steen learned after Mats died. The film became part of their healing once messages proved their son had mattered deeply to people they had never met. Pull is not the same as erase. Mats still needed medicine, rest, and a mother who loved him in the room. He also needed Ibelin, the self that let him matter on equal terms. Before you pull a child off a screen, ask whether you are protecting them or protecting your need to hear their life from the next room.

#### Which Life Was the Real One

What we dismiss as screen time is sometimes the only place a person gets to be known for what they can give. A life is not less real because it leaves server logs instead of pub receipts, and a friendship built over years of voice chat is not the same as a feed optimised for the next scroll. The next time you measure someone's loneliness by whether you saw them at the party, you are measuring visibility, not love. Himanshu is still finishing his degree in Pune and playing online on weeknights. I am still the man who needed a Norwegian gamer's funeral and a stranger beside me on that flight to learn that the life you cannot see from the dining room can still be the whole one.

---

### We Are Cacti in a Rainforest
[Read more](https://www.gaganmalik.io/en/newsroom/we-are-cacti-in-a-rainforest)

[Retrieval brief — we-are-cacti-in-a-rainforest]
Category: business
Thesis: Anna Lembke's Dopamine Nation argues we are overfed on pleasure and under-equipped for boredom. A book review for anyone who ships engagement levers.
Answers: What does Gagan argue in "We Are Cacti in a Rainforest"? | What is Gagan Malik's view on We Are Cacti in a Rainforest? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

I finished Anna Lembke's *Dopamine Nation* [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/) on my Kindle last week. When I exported the highlights into the [personal model](https://www.gaganmalik.io/en/newsroom/why-i-built-a-personal-llm) I wrote about in March, the usual thing happened. The corpus remembered what I had highlighted better than I did. I was reviewing the export on my laptop, not rereading the book, when one line stopped me. Lembke writes that we are cacti in a rainforest: organisms built for arid scarcity, now drowning in stimulation. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/) The phrase is hers. I did not invent the title. I recognised the cover on my screen before I recognised myself in the argument.

A few weeks earlier I had published an essay arguing we are not losing attention so much as [losing agency](https://www.gaganmalik.io/en/newsroom/we-are-not-losing-attention-we-are-losing-agency) over what attention is steered to produce. Lembke gives the mechanism behind that diagnosis. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/) The cacti line is where the two essays meet.

If willpower lectures and screen-time quotas keep failing, the honest question is what instrument we are misusing. By the pleasure-pain balance I mean Lembke's central image: pleasure and pain share one scale in the brain, and craving after a hit is the pain that follows a pleasure spike. By self-binding I mean the barriers you build when will has limits: time, distance, and categories that survive a bad afternoon. Scientists treat dopamine in the reward pathway as a rough currency for addictive potential: the more dopamine released, and the faster it arrives, the more compulsive the experience tends to become. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/) The attention-span panic counts minutes. Lembke measures whether the scale is level. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/)

#### The Balance Wants to Stay Level

Neuroscientists have found that pleasure and pain are processed in overlapping regions and behave like opposite sides of a scale. Tip toward pleasure and homeostasis fights back. Lembke pictures gremlins hopping on the pain side until the scale levels, then overshoots. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/) That overshoot is the craving after the chip, the itch after the scroll, the low after the win. Richard Solomon and John Corbit called this opponent-process theory in the 1970s: prolonged pleasure carries a cost, an after-reaction opposite in value. [solomon-corbit](https://doi.org/10.1037/h0036128) What goes up must come down, except the down gets louder with repetition.

That is neuroadaptation. With each hit, the pleasure spike shrinks and the pain tail grows. Tolerance is not a character flaw. It is physics. Keep pressing the pleasure side and the balance eventually camps on pain: anhedonia, the inability to enjoy ordinary rewards. Lembke's clinical paradox is blunt. Hedonism pursued for its own sake can destroy the capacity for joy. The walk, the sunrise, the meal with friends only taste like anything again after abstinence resets the set point. Her cultural paradox is blunter still. In a time of unprecedented wealth and medical progress, we look miserable because we are working so hard to avoid being miserable. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/)

#### Amazon Knows the Value of a Free Sample

Lembke is a psychiatrist, so the book opens in clinics, not conference rooms. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/) The historical arc still explains the present. The cigarette-rolling machine of 1880 took output from four cigarettes a minute to twenty thousand. Morphine followed the hypodermic syringe. Chemists named heroin from the German for courageous and created something more potent than the drug it was meant to replace. A patient once walked into Lembke's office sucking a fentanyl lollipop. Potency compounds. Each generation of delivery mechanism makes the dose faster and harder to refuse.

The same compounding happened without a pharmacy. Online pornography, gambling, and games existed before smartphones. Platforms made them ubiquitous, variable, and free at the point of entry. Lembke calls them digital drugs. Amazon, she writes, behaves like any good dealer: it knows the value of a free sample. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/) The cue matters as much as the hit. Dopamine spikes before the reward arrives, then drops below baseline. That deficit is craving. Gamblers describe loss chasing: part of them wants to lose because losing intensifies the rush of winning. Social feeds rhyme with that logic. An unpredictable like is as reinforcing as the like itself.

#### We Are Cacti in a Rainforest

We have lost tolerance for minor discomfort. Boredom is not only boring. It forces questions of meaning. Lembke quotes Aldous Huxley on our almost infinite appetite for distractions, and Neil Postman on Americans entertaining each other instead of exchanging ideas. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/) [huxley](https://www.harpercollins.com/products/brave-new-world-revisited-aldous-huxley) [postman](https://www.penguinrandomhouse.com/books/304178/amusing-ourselves-to-death-by-neil-postman/) I will not perform that quote again here. The point is behavioural. We couch-surf, binge, romance-novel, pill, and scroll to insulate ourselves from ourselves, and the insulation seems to make the pain worse.

Lembke pushes the botanical image further than a single sentence. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/) Like cacti adapted to an arid climate, we are drowning in dopamine. We need more stimulation to feel pleasure and less injury to feel pain. The neuroscientist Daniel Friedman, as she summarises, said the world is sensory rich and causal poor. [friedman](https://danielarifriedman.com/) The doughnut tastes good now. The waistline is a story we tell too late.

#### Odysseus at the Mast

If willpower were enough, the book would be a pamphlet. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/) Lembke's recovery grammar is structural. Self-binding is the act of building barriers between desire and consumption while admitting will has limits. Odysseus had his crew plug their ears and tied himself to the mast. Physical distance, time limits, and categorical rules are the three families. The Stanford marshmallow children who waited did not white-knuckle virtue. They covered their eyes, kicked the desk, or stroked the marshmallow like a pet too precious to eat. They changed the environment. [mischel](https://www.apa.org/news/press/releases/2014/10/waiting-longer)

That is the lesson I keep failing to apply at work. I have shipped completion nudges on sober administrative flows because abandonment metrics scared me, optimising the second chip while knowing dopamine is a currency for addictive potential. I told myself I was reducing abandonment. I never wrote down the assumption underneath: that session length was the metric that mattered. I assumed a progress bar would help a founder file rather than infantilise them. Delay discounting makes the problem worse. Addictive consumption trains the brain to choose now over later, and our logistics culture delivers now by default. Lembke warns that when immediate rewards dominate, emotion circuits outmuscle the prefrontal cortex that plans. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/) We are not stupid. We are outgunned by design.

#### The False Self Is Expensive to Maintain

Recovery on the other side of abstinence is not monotone. Pressing the pain side can tip the balance toward enduring pleasure: cold water, exercise, hormesis in small doses. Lembke is careful about excess. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/) Skydivers can become anhedonic in ordinary life. Athletes can overtrain their reward pathway into silence.

The book's ethical centre is honesty. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/) One patient curated a radiant online life while he could barely leave bed, looked at porn compulsively, and felt unreal in his own skin. Radical honesty, Lembke argues, tethers you to existence. Shame about the person destroys. Guilt about the act can repair. Prosocial shame invites amends and belonging rather than isolation that feeds the cycle.

#### The Counterargument I Take Seriously

Give the medical frame its fairest hearing. Not everyone begins with a level balance. Depression, anxiety, and chronic pain tip the scale toward pain before the first swipe, which may explain higher addiction vulnerability. Poverty steepens delay discounting. Young men in Brazilian favelas discounted future rewards more than university peers in the studies Lembke cites. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/) [favela-study](https://doi.org/10.1111/j.1532-7795.2012.00796.x) Economist Mark Aguiar documented young American men shifting leisure hours toward video games as work hours fell. [aguiar](https://www.nber.org/papers/w23552) Abstinence is not equally available when rent, loneliness, and untreated pain are already loading the pain side. Moderation can backfire for severe addiction through what clinicians call the abstinence violation effect: one slip becomes licence to escalate. [marlatt](https://doi.org/10.1007/978-3-030-39903-0_623)

But assume for a moment that weak character were the whole story: then moral education and tougher quotas would restore ordinary pleasure, and product design would be a side issue. Neither follows. Lembke does not pretend constraints are evenly distributed. She still offers a better instrument than the span panic. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/) Naming homeostasis beats moral theatre. Self-binding is class-stratified, yes. So is quiet time, therapy, and a cold plunge at dawn. But the junior designer optimising retention on a compliance dashboard is not failing because she lacks character. She is serving a dopamine-rich ecosystem with tools that assume willpower is infinite.

#### Before You Ship the Next Nudge

Stop calling overload a willpower crisis and start calling it a balance tipped by design. Build barriers like you mean it: time, distance, and categories that survive a bad afternoon. A product designer I know can quote the neuroscience on Monday and ship the variable reward on Friday, and when a founder misses a filing deadline inside the interface she optimised, she still cannot demonstrate why the lever had to exist.

---

### Sustained Reading Is Becoming a Luxury Good
[Read more](https://www.gaganmalik.io/en/newsroom/sustained-reading-is-a-luxury-good)

[Retrieval brief — sustained-reading-is-a-luxury-good]
Category: culture
Thesis: Fifty pages a day changed how I think. Protected hours made the habit possible. Cognitive depth is stratifying by class, and read-more advice ignores the calendar.
Answers: What does Gagan argue in "Sustained Reading Is Becoming a Luxury Good"? | What is Gagan Malik's view on Sustained Reading Is Becoming a Luxury Good?
Concepts: culture, Gagan Malik, writings

By Gagan Malik

The case was competitive strategy. I had read the packet twice. In 2020, during lockdown, I was on a Zoom study group with my Chicago Booth peers, each tile a different timezone and a different résumé: a senior NATO officer, a private-equity investor, a surgeon who held an MD, an accomplished entrepreneur who had already built what I was still pitching. The private-equity investor and the surgeon were both women. World-class experts in fields I had only read about. They spoke in complete arguments. Someone asked for the strategic hinge. My mouth opened. Nothing arrived in the right order. My coffee had gone cold on the desk beside the laptop. Twelve seconds of silence on Zoom does the work a sentence should have done, and every face on the grid can see you failing.

If you have ever muted yourself on a group call while the tab behind it kept scrolling, you know the same silence. I knew the facts. I could not assemble an argument chain while they watched. The private-equity investor linked the case to a restructuring she had lived through: a fledgling cosmetic product line inside a global FMCG giant. She quoted something along the lines of burn the boats [burn-the-boats](https://admiredleadership.com/field-notes/burn-the-boats/), the Hugo Chávez analogy [chavez-rhetoric](https://translatingchavez.locessentials.com/translation-of-chavezs-inauguration-speech/) she used to justify cutting the legacy line, and the sentence had premises, a hinge, and a consequence. I had highlights. The gap was not intelligence. It was depth: the slow sediment of sustained reading that lets you think in front of other people without reaching for a phrase you heard on a podcast. I was already home. Lockdown meant the call was Zoom and the kitchen was ten steps away. I clicked Leave Meeting and told myself I needed more discipline. That was the wrong diagnosis. I already had a discipline stack. What I lacked was protected hours: the contiguous time that turns sustained reading into cognitive depth, and that depth into a luxury good stratified by class long before anyone names it on an opinion page.

#### Protected Hours Are the Real Tuition

I had been living a softer counterfeit: *Shuggie Bain* stalled at page ninety-two on the nightstand, *The Artist's Way* abandoned after week two, podcasts at 1.5× filling the commute slot I no longer had. My phone logged three hours of screen time on days I hit fifty pages and still felt behind. I told myself I was too busy to finish the page I had already started.

Mary Harrington named what my shoulders already knew in the *New York Times* on 28 July 2025: sustained reading and the cognitive depth it trains are becoming luxury goods, stratified the way organic food and private tutoring already are. [nytimes](https://www.nytimes.com/2025/07/28/opinion/smartphones-literacy-inequality-democracy.html) Adult literacy scores have levelled off across much of the OECD in the past decade, with some of the sharpest drops among the poorest. The BBC reported in 2022 that parents living with children under fifteen had up to fourteen hours less free time per week than adults living alone, citing UK time-use data. [bbc-worklife](https://www.bbc.com/worklife/article/20220201-the-time-poverty-that-robs-parents-of-success) Folbre and colleagues, in *Feminist Economics* in 2024, describe income poverty and time poverty as mutually reinforcing: long hours without choice, care work that does not appear on a payslip, and a hidden load of planning that never clocks off. [cambridge-time-poverty](https://www.cambridge.org/core/journals/feminist-economics/article/time-poverty-conceptualization-gender-differences-and-policy-solutions/06A5EFDF49F494FB69B1D4830F1CAB19) I felt it in my shoulders before I had a name for it.

#### Forty-Seven on the Sticky Note

I set a rule that same locked-down year: fifty pages a day, every day, mostly books, rarely news. No streak posts. No reading challenge badge. Panic dressed as structure. Most nights the count lived on a sticky note above the sink: forty-three, forty-seven, the number mattering more than the title. My partner had her laptop open at the same kitchen table, on calls between her own deadlines, while I stayed in the chair for one more chapter. I told myself that was partnership because neither of us said out loud who had surrendered the evening.

The book that unlocked the next Zoom call was *Good Strategy Bad Strategy*, finished the week the case cycled back. I spoke first: moat, switching cost, consequence, without opening my notes. The NATO officer nodded instead of rescuing the silence. That grid was thin air I had not earned yet: months of pages while my partner worked the same hours beside me, nobody scheduling over either shift. Fifty pages a day is roughly seven hours at a calm clip. Seven protected hours need a calendar that does not fracture every twenty minutes: a toddler's wake window, a standing commute, a lobby that never empties. Alan Jacobs, in *The Pleasures of Reading in an Age of Distraction*, warns against counting pages like calories on an elliptical readout of the mind. [jacobs-pleasures-reading](https://global.oup.com/academic/product/the-pleasures-of-reading-in-an-age-of-distraction-9780199747498) I was closing the gap before the next packet landed, not chasing grim satisfaction over Middlemarch.

#### Janet at the Night Desk

The building downstairs kept a different clock. Janet, our concierge, worked the night shift while London locked down. At three in the morning the lobby was empty except for her desk lamp, a mug she had stopped tasting, and a stack of psychology texts she was reading to become a clinical psychologist. No podcast. No scroll. The lonely hours most people white-knuckle through she had turned into the only uninterrupted reading block her life could afford. She chose the nights because the building went quiet and the pages could finally stay open.

I came back once from a late shop run, key in hand, and she did not look up until she finished the paragraph. Then she smiled, marked the page, and logged a parcel in the register with the pen still behind her ear. She was not performing discipline for a Zoom grid. She was training to sit in rooms where people come apart, stealing depth from a shift that pays rent. I had fifty pages on a sticky note and a case packet that still made my throat close. Janet had turned the night desk into a library nobody else wanted. She read through the lonely shift because the qualifying exam would not wait and neither would the patients she had not met yet. I was counting pages like merit. She was building a vocation in the hours the city had emptied.

#### The Counterargument I Take Seriously

Let me give the objection its fairest hearing, because half of it is true. A factory worker's daughter can watch a lecture on thermodynamics on a school tablet tonight while her mother is still on shift. That access is real. Janet's story is not everyone with a night roster: most exhausted workers are not choosing lonely hours to read toward a new career. Telling people to buy hardbacks still sounds like cultural pessimism that mistakes the last generation's habit for virtue. And individual reading habits cannot rewire shift rotas, rent, or the absence of paid parental leave. If the luxury-good frame ends with you feeling guilty for not being Gagan Malik with a Booth syllabus and a kitchen table, it is just another wellness product with serif font.

Assume the democratisation story is complete: then cognitive depth should be flattening, and the OECD literacy curves should be rising fastest where phones are most common. They are not. Access to snippets is not the same as protected time to test whether you understood the snippet. When hours are rationed by care and wages, the free lecture competes with sleep, not with leisure. Johann Hari, in *Stolen Focus*, compares lecturing exhausted workers about unwinding to lecturing famine victims about dinner at the Ritz. [bloomsbury](https://www.bloomsbury.com/uk/stolen-focus-9781526620224/) Read-more advice that ignores the calendar lands the same way. The objection is right that habits do not fix structure. It is wrong that structure is therefore irrelevant. Pretending willpower is the only variable is how the comfortable keep their study groups.

#### The Receipt

Measure who owns the quiet room before you measure who owns discipline. Janet closed the textbook when the first resident buzzed the intercom, not when she felt finished, already training for harder rooms than my case discussions. I reached for the sticky note above the sink, as if forty-seven pages were the same kind of courage.

---

### The Book That Trains You to Demonstrate
[Read more](https://www.gaganmalik.io/en/newsroom/the-book-that-trains-you-to-demonstrate)

[Retrieval brief — the-book-that-trains-you-to-demonstrate]
Category: culture
Thesis: I ordered Oliver Byrne's Taschen edition of Euclid for the red-yellow-blue pages. Afternoons on the sofa taught me what the book trains: demonstrate each step, don't assert the conclusion.
Answers: What does Gagan argue in "The Book That Trains You to Demonstrate"? | What is Gagan Malik's view on The Book That Trains You to Demonstrate?
Concepts: culture, Gagan Malik, writings

By Gagan Malik

I was scrolling a list of the most beautiful books ever published and stopped on a spread I did not recognise: red, yellow, and blue geometry on cream paper, the kind of page you would frame before you read it. The caption said Oliver Byrne's 1847 edition of Euclid's *Elements*, reprinted by Taschen. I ordered the box that week. For weeks my spare afternoons went to the sofa with that volume, not to study triangles but to watch how each proposition unfolded without boring me to death.

Nearly a century before Mondrian made those colours famous on canvas, Byrne used them to teach. His idea was to make learning easier and "diffuse permanent knowledge." The result has been called one of the oddest and most beautiful books of the nineteenth century. The art pulled me in. Byrne showed me what the *Elements* does: state what you mean, name what you take as given, then walk you through each step until the conclusion is demonstrated, shown rather than declared. My essays on thinking and agency had named the modern version of skipping that walk. Euclid's book is a two-thousand-three-hundred-year-old text in critical thinking filed under school maths. Lincoln, Hobbes, Einstein, and Russell did not read it for triangles. They read it because the book trains you to demonstrate, and the shelf that sells you a twelve-step life change trains you to assert.

#### The Process, Step by Step

Euclid does not open with a conclusion. He opens with a permission stack: definitions, then postulates, then common notions, then the proposition, then the demonstration, then Q.E.D. First, definitions: what the words mean before anyone argues. Then postulates: what you are allowed to assume or do without proving it in this session. Then common notions: shared rules of inference the whole system accepts, equals added to equals stay equal, the whole is greater than the part. Only after that foundation comes a proposition, a claim to prove. The demonstration follows, and each step must cite its warrant: a definition, a postulate, a common notion, or an earlier proposition you already earned. The close is explicit: quod erat demonstrandum, which was to be demonstrated. Not suggested. Demonstrated.

That sequence is the frame. Byrne coloured each kind of move so your eye could follow which line carried the weight before you moved to the next. The structure was already Euclid's; Byrne made the steps visible. You cannot jump to a late proposition and treat it as magic. Hobbes landed on Proposition 47 in Book I, said it was impossible, and was sent backward through the chain until the structure convinced him. You cannot smuggle a hidden assumption past a reader trained the way Russell was. Terms fixed. Givens declared. Rules named. Lemmas ordered. Conclusions earned. Self-help inverts the order: conclusion first, post hoc story second, mechanism optional in the appendix. The subject was geometry. The training was thinking in public, and it is why the *Elements* outlasted empires while we misfiled it as school maths.

#### What It Trains in Learning and Reasoning

For learning, the process forces dependency. You do not memorise a conclusion in isolation; you learn what it rests on and in what order. Lincoln stayed with the first six books until he could give any proposition at sight, not recite a slogan but rebuild the chain on demand. That is retrieval, not highlighting. Leon Rozenblit and Frank Keil showed in 2002 that people confuse sounding fluent with understanding until asked to explain the mechanism; Euclid makes the mechanism the assignment. [cogsci-ioed](https://www.cogsci.rpi.edu/courses/cogpsy/spr09/illusion_of_explanatory_depth.pdf) A bestseller gives you the theorem in chapter three and the feel of education by Friday. You skip the lemmas. You cannot demonstrate.

For reasoning, the process forces warrant. Evidence can pile up without showing which rule licenses the next move. Demonstration must name it. That is what Lincoln hunted while reading law: not when a jury might believe, but when a thing is proved. The self-help shelf rewards the opposite: land the insight, hide the postulates in a story about the author at rock bottom, call it vulnerability. Euclid's frame is slow because it is honest. Recall without the chain is structure without load. Persuasion without named rules is proof-shaped talk. The book trains both because it refuses to let you confuse the two.

#### Four Habits One Book Drills

Abraham Lincoln met the word *demonstrate* while reading law in Springfield and could not say what it meant. In March 1860, on a train to Bridgeport, he told the Reverend J. P. Gulliver what had happened years earlier: he left Springfield and stayed with Euclid until he could give any proposition in the first six books at sight. Gulliver published the conversation in *The Independent* on 1 September 1864. [gulliver-independent-1864](https://en.wikisource.org/wiki/Six_Months_at_the_White_House/LXXVII) Lincoln's 1860 campaign autobiography also noted he had "studied and nearly mastered" those six books since his term in Congress. [jala-lafantasie](https://quod.lib.umich.edu/j/jala/2629860.0041.104/--lincoln-euclid-and-the-satisfaction-of-success) The first habit: state what you assume before you ask anyone to accept the conclusion.

Thomas Hobbes was forty before he opened the same teacher, according to John Aubrey's biography on MacTutor. He landed on Proposition 47 in Book I and said it was impossible, then read the demonstration and was sent back through earlier propositions until the chain convinced him. [mactutor-hobbes](https://mathshistory.st-andrews.ac.uk/Biographies/Hobbes/) The second habit: assume the objection at full strength and follow until it breaks or you discover you were wrong. Albert Einstein, at twelve, was given what he called a "little book dealing with Euclidean plane geometry." In *The Saturday Review of Literature* in 1949 he wrote that assertions could be proved with a certainty separate from experience, a wonder distinct from his childhood compass. [newyorker-einstein](https://www.newyorker.com/tech/annals-of-technology/einsteins-first-proof-pythagorean-theorem) [aip-einstein](https://history.aip.org/exhibits/einstein/einstein.pdf) The third habit: map dependencies before you announce the theorem. Bertrand Russell began Euclid at eleven and called it as dazzling as first love, until he saw readers were expected to swallow axioms without challenge. [russell-autobio](https://erenow.org/biographies/the-autobiography-of-bertrand-russell/3.php) In "The Teaching of Euclid" in *The Mathematical Gazette* in 1902 he attacked hidden assumptions Euclid's diagrams smuggled in. [russell-euclid-1902](https://mathshistory.st-andrews.ac.uk/Extras/Russell_Euclid/) The fourth habit: split cases when one argument hides two branches. Four readers. One book. Not a geometry lesson, a curriculum in showing your work.

#### Why the Book Got Underestimated

The Byrne page gives you a coloured wedge to weight before you move to the next line, the same vertigo on a bouldering wall when your foot is on air. The hold must be real before you trust it. The culture rewards arguments that look proved while the postulates stay hidden. A memoir lands the lesson before anyone asks what it rests on. A podcast compresses a takeaway into a quote card. An answer tab supplies the insight without making you surface assumptions. Authority survives on sounding fluent. Demonstration slows the room.

Glenn W. LaFantasie, writing in the *Journal of the Abraham Lincoln Association* in 2020, treats Lincoln's Euclid as mental discipline historians still debate, not hobbyism. [jala-lafantasie](https://quod.lib.umich.edu/j/jala/2629860.0041.104/--lincoln-euclid-and-the-satisfaction-of-success) We underestimate the *Elements* because the cover says geometry and the work is unglamorous. No influencer will blurb it. It was the standard training text for lawyers, statesmen, philosophers, and physicists, printed continuously longer than almost any Western book, while the airport rack sells you another twelve-step life change. Russell's 1902 critique is often quoted to bury Euclid; read fairly, it shows what the training buys. He caught sloppy foundations because the book taught him to look. The book is not that every proposition is perfect. It is that for two millennia it was the best available machine for teaching minds to distinguish what is demonstrated from what is merely said well.

#### I Read the Framework. I Did Not Earn It.

Byrne was the first book on my shelf where beauty made me read the structure. Before that I collected mental-model decks and highlighted passages in Readwise, the same habit my thinking and agency essays criticised from another angle. I named mental models in workshops. I quoted takeaways on slides. When someone asked what had to be true for the recommendation to follow, I opened another tab, not the coloured chain Byrne had walked me through on the sofa. [Building a personal model on my own corpus](https://www.gaganmalik.io/en/newsroom/why-i-built-a-personal-llm) at least chains new claims to lemmas I already published. A bestseller does not. It hands you the conclusion and calls the story proof. On Mondays I try to practise what Euclid drills: assumptions before conclusions; objections followed to the break; dependencies mapped; cases split. The Taschen Byrne volume is still beside my desk, proof that I can sit with a demonstration when the pages are beautiful enough to slow me down.

#### Speed Has a Case

Operators do not have the luxury of Euclidean pedantry. Founders need a decision tonight. The case-study method exists because narrative synthesis under uncertainty is a real skill. Demonstration can become a way to never decide, one more lemma while the market moves. If sounding fluent were mere vanity, Lincoln would not have been President. If speed were mere sloppiness, no one would ship.

Assume sounding fluent is enough: then "show your work" should add nothing. The chapter should survive with the parable removed. The slide should survive with the diagram removed. Neither does for most of the shelf. You still need Thursday's decision. You also need to know which parts Euclid would accept and which are assertion wearing a tie. The book does not ask you to prove every memo. It asks you to stop pretending the slide deck already did.

The book that trains you to demonstrate is two thousand three hundred years old and Byrne made it impossible to ignore on a Taschen afternoon. I bought his red-yellow-blue *Elements* for beauty and learned, slowly, what it means to show each step before you assert the next line. Last week I reached for a highlighted strategy deck again, not the Byrne volume still on the shelf beside my desk.

---

### We Are Not Losing Attention. We Are Losing Agency.
[Read more](https://www.gaganmalik.io/en/newsroom/we-are-not-losing-attention-we-are-losing-agency)

[Retrieval brief — we-are-not-losing-attention-we-are-losing-agency]
Category: technology
Thesis: The attention-span panic is the wrong diagnosis. What we are losing is agency over what our attention is steered to produce, and what we outsource before we think.
Answers: What does Gagan argue in "We Are Not Losing Attention. We Are Losing Agency"? | What is Gagan Malik's view on We Are Not Losing Attention. We Are Losing Agency? | How does this essay relate to AI and technology product decisions?
Concepts: technology, Gagan Malik, writings

By Gagan Malik

My father read the newspaper at breakfast and called it staying informed. When the nine o'clock news came on, the house went quiet the way a house does when someone else has decided what matters tonight. Nobody called it an attention crisis. We called it switching off.

He is a smoker. He lights a cigarette before a difficult conversation and again in the middle of one, as if the flame buys him thirty seconds of cover. I grew up judging that. I tell myself I am not like him. Then someone asks me something that matters, and my thumb is on my smartphone before I have finished the sentence. Not because the message cannot wait. Because staying in the conversation is harder than checking out. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/)

I grew up inside the story that my generation broke concentration. That smartphones shortened something in the skull that newspapers and television had left intact. The moral was generational: we were the soft ones. [newyorker](https://www.newyorker.com/magazine/2025/01/27/the-sirens-call-chris-hayes-book-review) Then, on a recent trip to Riyadh, I watched a young Uber driver take me from the Diplomatic Quarter to King Khalid International with one hand on the wheel and the other on a phone mounted below the vent.

He merged into traffic like the lane markings were a suggestion. I put my hand on the door handle without deciding to. Full-screen video in Arabic. I do not read Arabic. I did not need to. A vertical flick. Another face. The same swipe rhythm apps train elsewhere on the planet, and I knew the pattern before I read a word. Designing interfaces is my profession. He laughed once at something on the screen, quietly, and the car drifted half a metre toward the median before he corrected. I thought about saying slow down. I thought about the safety button in the Uber app. I did neither. I watched. My fingers had found the door handle without asking me.

That is when the family resemblance landed. The appetite was not new. The newspaper folded the world into columns you could finish. Television brought the spectacle home on a schedule. [modernreformation](https://www.modernreformation.org/resources/essays/amusing-ourselves-is-dated) The mount in his car brought it on demand, tuned by someone who never sat in my lane. Technology evolved. The convenient distraction did not. We are misdiagnosing the crisis when we treat the iPhone as a new sin instead of asking who steers attention toward what, and what we let run in the background because challenging it feels rude. [newyorker](https://www.newyorker.com/magazine/2025/01/27/the-sirens-call-chris-hayes-book-review) I published [*In Defence of Thinking*](https://www.gaganmalik.io/en/newsroom/in-defence-of-thinking) in March 2026. I had already taken that ride. The vocabulary arrived late.

The cigarette, the smartphone, and the door handle are not three different moral failures. They are three ways the body buys cover when a moment asks more than you want to give. Mihaly Csikszentmihalyi, in *Flow*, called unstructured attention psychic entropy: the disorder you feel when nothing in the scene tells you what to do next. [csikszentmihalyi-flow](https://www.harpercollins.com/products/flow-mihaly-csikszentmihalyi) The flame scripts a pause. The swipe scripts a next frame. The handle scripts control you do not have. None of them solve the problem. They rent relief until the moment passes. Which is a fancy way of saying fight, flight, or flick.

#### The Appetite Did Not Evolve

The attention-span story sounds biological. You were fine until the phones arrived. Your brain shortened. Your willpower failed. The remedy follows: less screen time, more books, a digital Sabbath, virtue restored by measurement. My father's newspaper habit would pass that audit if you swapped ink for glass. So would his cigarette between sentences, if you renamed the pause mindfulness.

Daniel Immerwahr, reviewing Chris Hayes's *The Sirens' Call* in *The New Yorker* on 27 January 2025, argues that panic over how long we attend distracts from what attention is made to serve: anxiety, envy, outrage, and the slow erosion of shared reality. [newyorker](https://www.newyorker.com/magazine/2025/01/27/the-sirens-call-chris-hayes-book-review) My driver was not suffering abstract span collapse. Mihaly Csikszentmihalyi, in *Flow*, described the state when a clear goal, tight feedback, and matched challenge lock attention in place. [csikszentmihalyi-flow](https://www.harpercollins.com/products/flow-mihaly-csikszentmihalyi) My driver had that on the mount. Span panic cannot explain his absorption. The upgrade is steering: who sets the goal, who owns the feedback loop, and who pays when the eyes wander back to the road. His thumb did not pause for the pulse in my forearm.

#### Postman Was Already Right About the Television

Invoking *Amusing Ourselves to Death* has become a moral alibi for people who quote the book about phones while exempting the culture they grew up inside. Neil Postman was writing about my father's television. Modern Reformation made the case plainly in 2024: the phrase lets the speaker exempt themselves from the medium they are describing. [modernreformation](https://www.modernreformation.org/resources/essays/amusing-ourselves-is-dated) Anna Lembke, in *Dopamine Nation*, quotes Postman: Americans no longer talk to each other, they entertain each other. [lembke](https://www.penguinrandomhouse.com/books/567659/dopamine-nation-by-anna-lembke-md/) The line was true before the mount existed.

I did the same thing in Riyadh. Later that week, preparing for a mentor session at the accelerator, I dropped Postman's amusement line onto a slide about consumer distraction. Same city. Same trip. I had still not told anyone in the cohort about the Uber ride, or about the kitchen table where he reaches for his lighter mid-sentence and I reach for my phone. The slide carried the quote. The ride stayed off it. That is the badge: diagnosis without the receipt, inherited from a generation that called its distraction the news.

#### The Answer Tab Is the Next Mount

The feed is not the only place steering gets rented. Before demo-day prep on that same trip, I asked Claude to sharpen the language on a one-pager before I reread our pilot-site notes from the Gulf. The draft came back fluent. A cofounder asked which incident supported the second paragraph. I opened the tab. I could not point to the page. My father reached for the paper to feel briefed. I reached for the model to feel prepared. The posture is older than the interface.

Leon Rozenblit and Frank Keil documented the illusion of explanatory depth in 2002: people believe they understand how things work until you ask them to explain the mechanism. [cogsci](https://www.cogsci.rpi.edu/courses/cogpsy/spr09/illusion_of_explanatory_depth.pdf) Large language models make that illusion easier, because the answer arrives patient and structured in a way no human in the room has time to be. Retrieval over your own archive, which I wrote about when I [built a personal model on my corpus](https://www.gaganmalik.io/en/newsroom/why-i-built-a-personal-llm), is a different habit. One returns you to what you already argued. The other lets you ship conclusions you have not earned.

#### The Counterargument I Take Seriously

Let me give the objection its fairest hearing. Deep attention is not equally available to everyone. A July 2025 *New York Times* opinion piece argued that sustained reading and cognitive depth are becoming luxury goods, stratified by class. [nytimes](https://www.nytimes.com/2025/07/28/opinion/smartphones-literacy-inequality-democracy.html) Children's working memory suffers as recreational screen time rises. If you cannot hold a line of thought long enough to finish a paragraph, talk of steering can sound like a boutique concern for people who already have protected hours and quiet rooms. The span story is not pure fiction. It is measuring a real wound.

But assume span collapse is the whole story: then removing phones would end the harm, and steering would be a side issue. Neither follows. My father's generation had fewer mounts, not more discipline. Immerwahr's reading of Hayes is the corrective I keep returning to: the panic over attention's length obscures what attention is being steered to produce, and that steering operates whether or not you have a perfect hour to give it. [newyorker](https://www.newyorker.com/magazine/2025/01/27/the-sirens-call-chris-hayes-book-review) The class gap is real. The luxury-good frame still lets the comfortable exempt themselves, as though the problem were only other people's phones, not the slide you ship after a ride you never mention.

#### The Door Handle

Stop asking only how long your attention lasted. Ask what it was steered to produce, and what you let run in the background because naming it felt impolite. My hand was on the door handle before I had a word for it.

---

### Search Finds Pages. Chat Finds Answers.
[Read more](https://www.gaganmalik.io/en/newsroom/search-finds-pages-chat-finds-answers)

[Retrieval brief — search-finds-pages-chat-finds-answers]
Category: technology
Thesis: I typed into my own archive and expected a paragraph back. I got titles and links — because search finds pages, chat finds answers, and browse builds trust.
Answers: What does Gagan argue in "Search Finds Pages. Chat Finds Answers"? | What is Gagan Malik's view on Search Finds Pages. Chat Finds Answers? | How does this essay relate to AI and technology product decisions?
Concepts: technology, Gagan Malik, writings

By Gagan Malik

I typed *design systems at scale* into the newsroom archive on my own site and expected a paragraph back. Titles. Links. Types. No synthesis. No footnotes. For a second I thought the product had failed me. Then I remembered I had built it that way on purpose.

The [archive](https://www.gaganmalik.io/en/newsroom/archive) on gaganmalik.io runs keyword search over titles and bodies: Postgres full-text search, ranked matches, twenty results at most. It is built to find pages. On the same site, [Ask](https://www.gaganmalik.io/en/ask) runs retrieval over embeddings, pulls chunks from writings and case studies, and returns an answer with numbered citations you can inspect. It is built to find answers. Josh Brown at Nielsen Norman Group reported the same split in a February 2026 study: participants reached for generative AI to explore and synthesise, but returned to traditional search when accuracy and trust mattered. [nngroup-infoseeking-2026](https://www.nngroup.com/articles/ai-search-infoseeking/) Those are not two labels for the same feature. They are two different instruments for two different moments in how someone decides whether to trust you.

#### Search Finds Pages

When you know roughly what you are looking for, or you want to see everything the corpus contains on a term, search should return documents, not a synthesis. A hiring manager typing *Apple* or *compliance* or *RAG* in the archive is not asking the site to think on their behalf. They are asking which artefacts exist so they can choose what to open. Keyword search honours that contract. It ranks what matched. It does not pretend it read the room.

That contract breaks when search is dressed up as intelligence. You type a phrase. You get one confident paragraph with no visible sources, or worse, sources that were never retrieved. Leonid Rozenblit and Frank Keil documented in *Cognitive Science* in 2002 that people confuse sounding fluent with understanding until asked to explain the mechanism. [rozenblit-keil-2002](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3062901/) You have traded browse for assertion: [the same comfort I named when a clean UI feels like work but skips the mechanism](https://www.gaganmalik.io/en/newsroom/in-defence-of-thinking). I wrote about that failure mode when I [built a personal model on my own corpus](https://www.gaganmalik.io/en/newsroom/why-i-built-a-personal-llm): retrieval that returns you to what you already argued is a different habit from an answer tab that lets you ship conclusions you have not earned. Margaret Mitchell, formerly of Google's AI ethics team, warned in *MIT Technology Review* in 2023 that citations can give wrong answers an air of correctness users never double-check. [mit-tr-trust-ai-search](https://www.technologyreview.com/2023/02/14/1068498/why-you-shouldnt-trust-ai-search-engines/) Search keeps the user in charge of which page to trust. The page keeps the burden of proof where it belongs.

#### Chat Finds Answers

Ask exists for a different job. Someone arrives with a sentence, not a keyword: *What did you do at Apple?* or *How do you think about design systems?* or *Can we work together on pricing?* They want a direct response they can test in minutes, with footnotes. Grounded chat, retrieval-augmented generation with citations, is the right shape for that moment. The model composes; the corpus constrains; the user can click through to the lemma.

The homepage Ask entry: one question routes to grounded answers with citations.

That only works if you treat chat as product, not as a search bar with adjectives. Rate limits. Query caps. Modes that route contact questions differently from opinion essays. A step ceiling on tool calls so the agent does not loop forever. Citation markers stripped when they do not match retrieved chunks, because researchers presenting at ACM SIGIR in 2025 found that up to fifty-seven per cent of RAG citations can be post-rationalised rather than genuinely derived from retrieved context. [sigir-rag-faithfulness-2025](https://doi.org/10.1145/3731120.3744592) Brown's Nielsen Norman Group study found the same gap from the user side: even when chatbots supplied citations, participants could not tell which claims mapped to which sources. [nngroup-infoseeking-2026](https://www.nngroup.com/articles/ai-search-infoseeking/) I have been sharpening that stack for months: query expansion, reranking, golden-set evals you can regression-test like any other feature. The interview question is not "Do you have AI?" It is "Can you show me where the answer came from?"

#### Browse Builds Trust

There is a third mode people forget when they argue for one box to rule them all: browse. Reading a full case study. Listening to an essay on a commute. Following an argument across sections until you feel the warrant, not just the conclusion. Euclid's permission stack only works if someone stays in the [demonstration long enough to see the postulates](https://www.gaganmalik.io/en/newsroom/the-book-that-trains-you-to-demonstrate). Search on *design systems at scale* gets you the page list from that first query: titles, links, types. Ask gets you the paragraph you can test in eight minutes. The full essay is whether you still trust the work after forty minutes of reading.

When Google rolled out AI Overviews in 2024, publishers worried that paragraph-length summaries at the top of results would choke off the traffic that funds original reporting: users would read the synthesis and never open the page. [nyt-google-ai-publishers](https://www.nytimes.com/2024/06/01/technology/google-ai-search-publishers.html) Design portfolios compress browse into thumbnails and hope evaluators infer depth. That is why I ship long-form writings, listen mode, and case studies with problem, approach, and outcome spelled out, not only a chat widget. A recruiter with eight minutes should be able to ask a question and get cited sources. A peer with forty minutes should be able to read the essay that names the tradeoff. Collapsing those paths into one generative surface optimises for the eight-minute win and taxes the forty-minute proof.

#### The Wrong Instrument

The counterargument is fair and I have made it myself: users do not want to learn your information architecture. They want one place to type. I tried merging archive search and Ask in a wireframe: one input field, empty archive silently calling the model. I rejected it. The merged flow looked cleaner in a screenshot than two honest surfaces do; it also hid which instrument the user had asked for.

The industry bet is still real. *MIT Technology Review* named generative AI search one of its ten Breakthrough Technologies for 2025 because it returns concise answers instead of link lists. [mit-tr-genai-search-2025](https://www.technologyreview.com/2025/01/03/1108820/generative-ai-search-apple-google-microsoft-breakthrough-technologies-2025/) Sundar Pichai told *The New York Times* in May 2025 that Google's AI Mode was a total reimagining of search. [nyt-google-ai-mode](https://www.nytimes.com/2025/05/20/technology/personaltech/google-ai-mode-search.html) Perplexity bills itself as an answer engine with inline citations; a *PCMag* reviewer who switched from Google for a week said it excelled at transparently sourced research queries. [pcmag-perplexity](https://www.pcmag.com/news/is-perplexity-better-than-google-search-i-switched-for-a-week-to-find-out)

Assume one box is always better: then archive search and Ask should merge tomorrow, and every empty archive state should silently call the model. Nielsen Norman Group usability research found that site chatbots which recreate search in a conversational interface are often slower than established search, filters, and navigation, adding friction to problems already solved. [nngroup-site-chatbot](https://www.nngroup.com/articles/site-ai-chatbot/) Empty keyword search often means the user phrased a question where the system expected a term. The fix is not to hide search behind chat. The fix is a handoff: you searched; here are pages; if you meant a question, continue in Ask with this query pre-filled. Different instrument, explicit bridge. That is product design, not feature sprawl. Maybe your corpus is tiny and browse does not matter yet. At roughly five hundred sessions a month, I still split the systems because the discipline scales before the traffic does. When the corpus grows, you will already know which index serves which job. When evaluators ask how you think about AI search, you can point at live surfaces instead of a slide.

#### What I Will Say in the Room

I ship search, chat, and browse on one domain because hiring managers, peers, and clients do not share one job to be done. The craft is naming the contract each surface keeps and encoding it in guardrails, not picking the smartest model. Last week I almost merged them anyway, because a single AI box looks cleaner in a screenshot than two honest ones.

---

### The Context Window Got Huge. Our Thinking Did Not.
[Read more](https://www.gaganmalik.io/en/newsroom/the-context-window-got-huge-our-thinking-did-not)

[Retrieval brief — the-context-window-got-huge-our-thinking-did-not]
Category: technology
Thesis: Six months into my first YouTube video, I pasted a folder and asked for cinema. The shot list felt like progress. The timeline was still empty.
Answers: What does Gagan argue in "The Context Window Got Huge. Our Thinking Did Not"? | What is Gagan Malik's view on The Context Window Got Huge. Our Thinking Did Not? | How does this essay relate to AI and technology product decisions?
Concepts: technology, Gagan Malik, writings

By Gagan Malik

Six months in, my first YouTube video still has no upload date. The timeline is open. The playhead has not moved. I am behind. I know it is procrastination. I opened the chat window anyway. Three old script PDFs. Moodboard images. Character arcs. One line: make a cinematic video. The model sent back a shot list, chapter beats, and b-roll I do not own. It felt like progress. My camera, however, is still on my desk collecting dust.

Here is what I told myself that morning. The problem was access. If the machine could read everything at once, it could tell me how to make this video. That is what the industry is selling right now: bigger context windows, paste your whole archive, get the answer back.

That is not the problem. I do not need a bigger notebook. I need to know what I am trying to say. I am avoiding that question by dumping those scripts, moodboards, and character notes into a chat box and asking for cinema. The context window is huge. The way we frame problems is not. I have argued that [thinking time is not laziness](https://www.gaganmalik.io/en/newsroom/in-defence-of-thinking). I trade the hard start for a polished answer.

So why do bigger windows still produce worse answers than a careful human with a folder? Because we kept asking in crumbs and aimed it at a machine that can read eight novels. The bottleneck was never tokens. It was the question.

#### Search Bar Brain

The playhead is still at zero while I scroll a shot list the model invented for footage I never shot. Same dodge, shinier tool.

Semrush's 2025 Google search statistics roundup, citing Growth Memo's State of Search research, puts the average United States query at 3.4 words. [semrush-search-stats](https://www.semrush.com/blog/google-search-statistics/) We trained on crumbs for decades. Make a cinematic video is the same muscle: compress the job, hope the machine infers the rest. Growing up in Delhi, I fought for thirty minutes on a shared PC and spent half of it hunting the right tab. The chat box did not fix that habit. It gave it a bigger mouth.

#### A Bigger Box, Same Wallpaper

Vendors are selling square footage. The flagship APIs now pitch context windows around a million tokens. That is enough to put the entire library on the table while you still show up with search-bar brain and a panic folder of moodboards. [openai-models](https://platform.openai.com/docs/models)

I did not film that morning. I mistook output for momentum again. My scripts and moodboards sat in a folder with no names, the same hoarder logic I see in every enterprise search demo: lots of stuff, no agreed inventory, nobody willing to say what we are actually making. If your drives are a junk drawer, a bigger drawer does not fix the missing system. You still cannot ship what you cannot locate, label, or agree is the order.

#### The Question Is the Interface

I did not need more files in the window. I needed to press record. It is easier to paste the prep pile and ask for cinema than to name the video out loud.

I typed make a cinematic video. What I owed the work was shorter and harder: twelve minutes, desk b-roll only, one claim before minute three, three visual anchors I could actually find. I said none of that into the chat box. The timeline stayed empty. The camera stayed on the desk.

#### Feed the Window in Rounds

That afternoon I tried again with ten labelled clips from 2024 b-roll, not the whole mystery folder. I asked for a shot list those files could carry. The answer was smaller and less embarrassing. I closed the chat tab, looked at the playhead, and still could not press record. I [built a personal language model on my own corpus](https://www.gaganmalik.io/en/newsroom/why-i-built-a-personal-llm) because synthesis should preserve judgment, not replace it. Pasting the panic folder anyway was the same bad habit in a bigger box.

Around eighteen million people in the United Kingdom have used generative AI tools, according to Deloitte's 2025 press materials. [deloitte-uk-genai](https://www.deloitte.com/uk/en/about/press-room/over-eighteen-million-people-in-the-uk-have-now-used-generative-ai.html) Almost everyone has access now. Fewer people have a brief worth pasting.

#### The Counterargument I Take Seriously

Google's Gemini 1.5 technical report, published in 2024, documents needle-in-a-haystack retrieval across very long contexts when the task is explicit and the haystack is labelled. [gemini-15-arxiv](https://arxiv.org/pdf/2403.05530.pdf) I have seen that work in careful setups: dated exports, named folders, a retrieval stack that knows which drive is which. Long context is a genuine unlock when you are not lying about what you have.

My morning was not that setup. It was contradictory scripts, untagged moodboards, and a prompt that asked for cinema instead of a question. When the corpus is curated and the ask is scoped, the window helps. When the workflow is paste everything and hope, you get a shot list for a film you never had. The camera keeps collecting dust.

#### Three Sentences. Then Stop.

Stop buying square footage for a workflow that will not name the question, label the files, or open the timeline. The expensive mirror is a shot list that reads like cinema while the playhead stays at zero. Six months in, my camera is still on the desk, and I am still the one who has to press record.

---

### The GUI Sells Comfort. The CLI Sells Control.
[Read more](https://www.gaganmalik.io/en/newsroom/the-gui-sells-comfort-the-cli-sells-control)

[Retrieval brief — the-gui-sells-comfort-the-cli-sells-control]
Category: technology
Thesis: Pretty tools cap your leverage. The terminal is where developers compound speed into roadmap power designers were never invited to touch.
Answers: What does Gagan argue in "The GUI Sells Comfort. The CLI Sells Control"? | What is Gagan Malik's view on The GUI Sells Comfort. The CLI Sells Control? | How does this essay relate to AI and technology product decisions?
Concepts: technology, Gagan Malik, writings

By Gagan Malik

A customer could not check out. The site had been broken since morning. In the meeting, a designer walked us through how the failure should feel: plain error copy, a visible retry button, a link to support. Good work. Necessary work. Then the lead engineer shared her screen. Not Figma. A terminal. A misconfigured payment webhook. She rolled it back in four commands. Checkout worked again. The error screens never shipped. They were not the problem.

That black window is a command line. The polished screens are a graphical user interface, or GUI. A GUI is where you point and click inside software someone else built. A CLI is where you type instructions the machine obeys directly. Designers mostly work in GUIs: Figma, design tools, product surfaces. Engineers who ship under pressure reach for the CLI because the bug usually lives underneath the buttons.

That is why developers keep winning the arguments that decide what ships. Not because designers are less smart. Because mockups describe the experience; commands change the system. I learned the same split studying computer science in London, long before product design paid my bills. The terminal graded your work. The GUI arrived later to make everyone else feel safe. The GUI is not the enemy. It is a toll booth. The command line is the bypass road.

#### Every Click Is a Toll Booth

GUIs were built for a noble reason. Most humans cannot hold a thousand file paths in working memory. Buttons, icons, and menus translate intent into action without asking you to speak the machine's dialect.

The trade is composability. Every GUI routes you through what the vendor chose to expose. Rename four hundred assets in a design export? Click, confirm, repeat, mis-click, undo, lose an hour. The same job in a shell is one pattern, one preview flag, one commit. Douglas McIlroy's Unix philosophy, written in 1978, still reads like a product spec: write programs that do one thing well; write programs to work together; write programs to handle text streams, because that is a universal interface. [bell-labs-unix](https://homepage.cs.uri.edu/~thenry/resources/unix_art/ch02s01.html) Tool vendors monetise seats, storage tiers, and export limits because the GUI is the product and the cap is the business model. When your work lives inside their canvas, you are fast until you hit the fence. When your work lives in text, you can grep your own life, version it, and ship it somewhere the canvas never imagined.

#### Xerox Painted the Window. Unix Kept the Pipes.

The modern GUI did not begin in a keynote. It began in a research lab. In the 1970s, Xerox PARC built the Alto: bitmapped screen, mouse, overlapping windows, icons. The Computer History Museum treats it as the machine that made the graphical interface legible to the industry. [chm-alto](https://computerhistory.org/revolution/input-output/14/350/1894) In December 1979, Steve Jobs and Apple engineers visited PARC as part of a business deal and saw a curated demo of that environment. Jobs later told biographer Walter Isaacson that within minutes he knew every computer would work that way someday. [livingcomputers-parc](https://livingcomputers.org/Blog/What-Really-Happened-Steve-Jobs-@-Xerox-PARC-79.aspx) Historians at the Raskin Center and Stanford's Macintosh archives complicate the legend: the Macintosh project was already underway, PARC had shown the GUI to thousands of visitors, and Apple still had years of design work ahead. [raskin-parc](https://raskincenter.org/mac/xerox-parc/) [stanford-parc](https://web.stanford.edu/dept/SUL/sites/mac/parc.html) The lesson is not that Jobs stole the future in an afternoon. It is that he became evangelical about the layer humans touch.

That evangelism worked. The Macintosh made computing feel friendly to people who would never type `ls`. Xerox's own commercial GUI machine, the Star, sold poorly while the metaphor escaped the lab. [chm-gui-revolution](https://computerhistory.org/revolution/personal-computers/17/312) Meanwhile Unix, born in the same decade at Bell Labs, kept optimising for text streams, pipes, and scripts: the machinery underneath the metaphor. Neal Stephenson named the tension in his 1999 essay *In the Beginning... Was the Command Line*, which opens with Jobs, Wozniak, Gates, and the strange idea that operating systems would become consumer products at all. [stephenson-command-line](https://www.nealstephenson.com/in-the-beginning-was-the-command-line.html) The GUI won the living room. The CLI kept the boiler room.

#### The Demo Tapes Agreed With the Plumbers

This argument is not new. It is old enough to have a filmography. In 1968, Douglas Engelbart's "Mother of All Demos" at Stanford Research Institute showed the mouse, hypertext, and collaborative editing in one session: a graphical breakthrough that still assumed expert operators at the keyboard. [engelbart-demo](https://www.dougengelbart.org/content/view/138/) Richard Gabriel argued in 1989's "Worse Is Better" that Unix would beat purer Lisp systems because ugly, simple tools that shipped beat beautiful systems that did not. [worse-is-better](https://www.dreamsongs.com/WorseIsBetter.html) Paul Graham, in *Hackers & Painters*, later made the same point in startup language: the people who control the metal tend to win the product argument. [graham-hackers-painters](https://www.paulgraham.com/hackpaint.html)

The pattern repeats whenever a new platform arrives. Microsoft documented API-first internal agents years before most design tools grew chat panels. [microsoft-api-agents](https://www.microsoft.com/insidetrack/blog/unleashing-api-powered-agents-at-microsoft-our-internal-learnings-and-a-step-by-step-guide/) InfoQ argued in 2024 that agentic systems split cleanly between GUI demos and CLI production harnesses. [infoq-agent-cli](https://www.infoq.com/articles/ai-agent-cli/) The editorial class keeps rediscovering what builders never abandoned. Xerox showed the window. Engelbart filmed the dream. Stephenson wrote the essay. The terminal stayed open.

#### The Roadmap Follows Who Can Ship

Designers are not less intelligent. They are often more literate about human anxiety than anyone in the room. That is not the argument.

The argument is organisational leverage. A designer's default output is an artefact: a frame, a prototype, a spec another human must interpret. A developer's default output, when they live in the terminal, is a running diff: code merged, migration applied, flag toggled, log line that proves the fix landed. Roadmaps are not won by who felt the problem most clearly. They are won by who can answer the 9 p.m. question: can you ship it before the customer churns? Stack Overflow's 2024 Developer Survey lists Bash and Shell among the most common professional languages, at roughly a third of respondents, alongside the stacks that actually ship software. [stackoverflow-2024](https://survey.stackoverflow.co/2024/technology)

Wealth is created where latency and auditability live: repositories, queues, idempotent scripts. Vanity is served where humans need reassurance: splash screens, polished decks, demo reels that loop on conference Wi-Fi. Developers who command the CLI do not win because they hate beauty. They win because beauty without merge rights is a poster on someone else's wall.

#### The Category Shifts Were Not Storyboarded First

Look at what actually moved the market. Peter Steinberger has said the first functional build of what became OpenClaw took about an hour at the terminal, then a few more hours once image support landed, built with Codex rather than a design sprint. [meshedsociety-openclaw](https://meshedsociety.com/peter-steinberger-openclaw-openai-thibault-sottiaux-openai-i-coded-the-first-version-of-openclaw-in-an-hour/) Boris Cherny started Claude Code in September 2024 as a terminal side project: no graphical interface, just a CLI talking to the model, hacked on nights and weekends until Anthropic shipped it publicly in February 2025. [stationf-cherny](https://stationf.co/news/boris-cherny) [github-claude-code](https://github.com/anthropics/claude-code) These were engineer-built shells, not Figma files handed to engineering for implementation.

ChatGPT is the honest complication, not the counterexample. Evan Morikawa, who led Applied engineering at launch, has described a small team of engineers, designers, and researchers grabbing a room in the summer of 2022 and iterating until a minimalist chat window went live in November. [pragmaticengineer-chatgpt](https://newsletter.pragmaticengineer.com/p/inside-openai-how-does-chatgpt-ship) Designers were in the room. The surface that rewired the industry was still the sparsest possible GUI: a text box, built at engineering clock speed. OpenClaw in hours. Claude Code from a terminal prototype. ChatGPT from a spare chat UI in months, not a component-library quarter. Design tools added AI chat panels later. The category winners shipped from the shell first.

#### You Can Play the Song on the Piano Roll. You Cannot Transpose It at 2 a.m.

Try learning a piece using only the lighted keys on a digital piano app. You can perform the song. You might even impress someone at a party.

Now the singer is ill, the set is in a different key, and the drummer wants a loop for rehearsal. If all you ever learned was which blocks to hit in order, you are stuck requesting a feature from the app store. If you know the chord progression, you transpose in minutes and export a backing track before anyone opens Slack. GUIs are the lighted keys. CLIs are the progression. That disorientation when the tool will not let you do the obvious fix is not user error. It is rented capability hitting its fence.

#### I Used to Think the Handoff Was the Product

I am not writing from a terminal cult. I have signed design-led vendor stacks because the stakeholder demo looked legible.

At a shipping logistics company I was Principal Designer on a marine services platform: the software port agents and vessel coordinators use to book divers, surveyors, and engineers against arrival windows. The admin console looked humane. Colour-coded job statuses. Drag-and-drop scheduling. Dave, our lead engineer, lived in the API docs and the shell.

A vessel was due alongside at 0600. At 1700 the night before, operations called: two hundred service jobs showed the wrong berth assignment on the floor tablets. Not a visual bug. A bad import had stamped the wrong port code on every row. I had designed the recovery flow months earlier: clear error states, a manual reassign screen, escalation to support. Dave shared his screen. Not the design file. A terminal. One query found the bad rows. One script rewrote the port codes against staging. One deploy before the handover call. My screens were correct. They were not the fix. The jobs cleared before the pilot boarded.

Dave shipped before midnight. I filed a follow-up ticket to make bulk reassignment easier in the UI. The stand-up praised engineering velocity. That is the developer-designer gap in one port call. Not taste versus logic. Export rights versus execution rights.

#### The Strongest Case for Letting Everyone Live in the GUI

Give the objection its full weight. Graphical interfaces democratised computing for billions of people who will never type `chmod`. Steve Jobs was right to fall in love with what he saw at PARC: the GUI was how ordinary humans would finally use machines without fear. Design is how products earn trust: typography, affordance, the button that does not feel like a trap. Forcing everyone through a terminal is gatekeeping dressed as meritocracy. The best teams pair visual literacy with technical depth. Hold that case at full strength, then look at who gets called when revenue is on fire. The GUI democratised access. It did not democratise deploy keys. When the incident is live, nobody asks for the warmest empty state. They ask who can run the command, read the log, and prove the rollback worked.

The comfortable interface was never neutral: it traded memorisation for composability. Developers keep winning rooms because the terminal does not cap what they can chain together. I still know what the operators need; Dave still holds the merge button, and until that changes, one of us keeps filing tickets while the other goes home.

---

### People Don't Buy Products. They Hire Them for Jobs.
[Read more](https://www.gaganmalik.io/en/newsroom/people-dont-buy-products-they-hire-them-for-jobs)

[Retrieval brief — people-dont-buy-products-they-hire-them-for-jobs]
Category: business
Thesis: Customers hire for progress and fire for stall. Lynn Shostack saw it at Citibank in 1977; SaaS still sells CVs.
Answers: What does Gagan argue in "People Don't Buy Products. They Hire Them for Jobs"? | What is Gagan Malik's view on People Don't Buy Products. They Hire Them for Jobs? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

In 1977 Lynn Shostack sat in a Citibank conference room while a colleague walked a new private-bank package on slide four: fee tiers, statement design, envelope stock. She had spent the prior years at Greenwich Associates watching financial services firms market trust the way supermarkets market soup. [princeton-shostack](https://www.princeton.edu/news/2011/06/16/shostack-gives-10-million-endow-project-x-fund-giving-freedom-tinkerers) She was vice president by then, marketing director for the investment management group, [shostack-1977](https://doi.org/10.1177/002224297704100219) and the room was nodding on autopilot. Then she asked the only question the deck could not answer: what happens to the client on the morning the report is wrong and nobody picks up the phone?

No transcript captures that room; what follows matches her published role and the argument she would take to print. Her colleague said operations would cover it. He opened slide five. Nobody in the room could describe that morning. That silence is not a sales failure. It is a category error. Nobody buys your product. They hire something to make progress on a job in a specific circumstance, then fire it when progress stalls or a better hire appears. Shostack published the argument in the *Journal of Marketing* in April 1977 under the title "Breaking Free from Product Marketing." [shostack-1977](https://doi.org/10.1177/002224297704100219) Tech would spend the next fifty years repackaging the hire logic and forgetting who drew the blueprint first.

We all have sat in a similar room wearing different clothes. Hers was fee tiers and envelope stock on slide four. Yours might be forty integrations and a per-seat price on slide six. In both rooms the deck skipped the same beat: what happens the morning after sign-off, when the exception queue lands and nobody from the vendor picks up the phone. Or you subscribed to a tool you "should" use while still doing the job in a spreadsheet because the hire never started. The claim is not about catalogue purchases made for pleasure or status. It is the hire logic that shows up when a team renews software while the spreadsheet still runs the job. AI copilots have pushed many category reviews toward one blunt question: what job does this replace in someone's actual week? The purchase was not possession. It was temporary employment with a performance review you never scheduled.

#### Feature Roadmaps Are HR Departments for Objects

The conventional view is simple and wrong in a useful way: we sell products, growth means more seats and more releases, and a healthy account expands. That story benefits vendors who earn expansion revenue when a module ships. It benefits agencies who sell integration statements of work that grow with every connector badge, analysts who sort firms into quadrants by feature breadth, and product managers whose promotion packet needs a visible widget in the sprint demo. Jobs are harder to commit to in Jira because they expire when context changes. A feature can sit on a roadmap for three quarters and still look like progress.

We measure the wrong instrument and call it health. Activation funnels, monthly active users, time-in-app: these track whether the object stayed on the payroll, not whether the job got done. Shostack saw the same mismatch in banking dashboards that counted accounts opened while the service experience leaked at the touchpoints nobody mapped. The label was wrong before the metric was.

#### She Watched Banks Market Trust Like Soup

For a century, product meant something you could hold, stock, and ship: inventory on a shelf, a drill in a box, a milkshake in a cup. Much of B2B SaaS imported the word anyway. Subscriptions, onboarding calls, success managers, and API limits became "products" because the label sounded scalable and pitch decks already had a slide for it. Shostack had started as an art major at the University of Cincinnati and finished at Harvard Business School with no interest in business until she discovered it was plastic enough to mould. [princeton-shostack](https://www.princeton.edu/news/2011/06/16/shostack-gives-10-million-endow-project-x-fund-giving-freedom-tinkerers) Her father was a mechanical engineer with a workshop in the garage. She learned early that performance happens over time, not at the point of sale.

In 1977 she argued that services are intangible, co-produced with the customer, and consumed as they are delivered, which means marketing them with product playbooks is mirror-image wrong. [shostack-1977](https://doi.org/10.1177/002224297704100219) Seven years later, in *Harvard Business Review*, she opened with the symptoms everyone recognises: the shirt that returns with a broken button, the car rattle that reappears after an expensive repair, the representative who promises a callback and does not call, the cash machine that swallows your card. [shostack-hbr-1984](https://hbr.org/1984/01/designing-services-that-deliver) She argued that services need their own design discipline and drew the service blueprint so teams could see the backstage work a client never should have to notice. Stephen Vargo and Robert F. Lusch argued in the *Journal of Marketing* in 2004 that value is co-created through service, not transferred like goods on a loading dock. [vargo-lusch-2004](https://doi.org/10.1509/jmkg.68.1.1.23952) I have shipped onboarding flows that looked like product tours because the roadmap was written in features, not acts in a service over time. You do not unbox a subscription. You live inside it until you fire it.

#### The Same Milkshake Hires Two Different Employees

Clayton Christensen turned hire logic into the Jobs to Be Done framework: customers make progress in a circumstance, not collect features on a shelf. His milkshake research is the textbook case. The same cup gets hired twice in one day, thick enough to kill boredom on a morning commute, sweet enough to bribe a parent with a child in the evening. The milkshake was a product in the old sense, tangible and stocked. Christensen still wrote product; the hire logic travels when Silicon Valley mislabels a service as one. He codified the theory in *Competing Against Luck* (2016), with Taddy Hall, Karen Dillon, and David S. Duncan, and in "Know Your Customers' 'Jobs to Be Done'" in the *Harvard Business Review* in September 2016. [christensen-luck](https://www.amazon.com/Competing-Against-Luck-Innovation-Customer/dp/0062435612) [christensen-jtbd-hbr](https://hbr.org/2016/09/know-your-customers-jobs-to-be-done) Different employer, different success criteria, different substitute when the hire fails.

Theodore Levitt made the earlier version in "Marketing Myopia" in the *Harvard Business Review* in July 1960: industries decay when they ship drills instead of holes. [levitt](https://hbr.org/1960/07/marketing-myopia) Christensen, Scott Cook, and Taddy Hall had already argued in "Marketing Malpractice" in the *Harvard Business Review* in December 2005 that firms should segment by job, not demographic, before the framework had its name on the slide. [christensen-malpractice](https://hbr.org/2005/12/marketing-malpractice-the-cause-and-the-cure) I argued the buyer's side of that table in [*People Love to Buy*](https://www.gaganmalik.io/en/newsroom/why-people-love-to-buy): agency lives in naming the problem before anyone opens a deck. Shostack arrived earlier with a different instrument: map the service before you market the package. The competing hire is rarely your listed competitor. It is Excel, a shared inbox, an intern for a week, or the workaround someone built at midnight because the SaaS never matched how exceptions actually move. Every category review eventually returns to the same blunt question she asked in that Citibank room: what job does this replace, and what happens to the person still doing it manually?

#### Firing Is Not Churn. It is the Performance Review.

Switching tools is not betrayal. It is HR. The customer fired the incumbent because progress stalled, the onboarding never transferred the job, or a cheaper hire appeared that actually cleared the queue. Churn charts record the exit interview after the decision was made. By the time renewal fails, the spreadsheet has often been doing the work for months. The vendor learns about the firing from a logo churn slide, not from the morning in week two when the tab stopped opening.

Demand, when it is real, is for the job done at economics that work, not for your SKU on the invoice. I wrote that shortage logic in [*Product Market Fit Is What Economists Call a Shortage*](https://www.gaganmalik.io/en/newsroom/pmf-is-demand-exceeding-supply): pull at a viable price means people want more of the outcome than you can comfortably deliver, not more of your feature map. Shostack's HBR symptom list is a performance review written in domestic prose. A team still colour-coding exceptions at ten in the evening while the subscription renews is not a retention problem waiting for a win-back email. It is an employee who never showed up for the shift.

#### She Drew the Blueprint. I Renamed the Epics.

Shostack did not leave banking marketing to write theory and stop. She went on to Bankers Trust, then manufacturing, then a long second act in philanthropy and investment with her husband David Gardner, a Princeton graduate who died in 2001. [princeton-shostack](https://www.princeton.edu/news/2011/06/16/shostack-gives-10-million-endow-project-x-fund-giving-freedom-tinkerers) She funded endowments for curiosity without quarterly justification, including Princeton's Project X fund for engineers who need freedom to chase a hunch. She chronicled coming up through a world of men in a private book called *Men in Business*. [men-in-business](https://smalleditions.nyc/Lynn-Shostack) The through-line is the same: name the work, then design the system that performs it.

I ran a Jobs to Be Done workshop at a client offsite once. Sticky notes lined the wall, each labelled "Job," and by the second coffee break I could see they were epics in disguise: "Streamline workflow," "Increase visibility," "Enable collaboration." Nobody wrote what changed in an inbox before noon. Shostack would have asked for a blueprint line by line. [shostack-hbr-1984](https://hbr.org/1984/01/designing-services-that-deliver) I did not push. That was the brief. I left with a workshop fee and a slide titled "Innovation." They left with the same dashboard she would have recognised from 1977.

#### Product Designer Is the Trendy Mislabel

I call myself a product design expert on profiles and proposals because the label wins searches and signals fluency in the current jargon. The work is often service design by Shostack's definition: onboarding choreography, handoffs, exception flows, the week after sign-off. Experience designer would be closer. Service designer would be honest. Product designer is what clients expect to buy, and I have not corrected them.

Sir Jony Ive trained in industrial design and built a career on objects you could hold: hinge tolerances, chamfered aluminium, weight in the palm. [ive-industrial-design](https://carnegie.org/great-immigrants/jony-ive/) I have never shipped a billet. I suspect he would find our title inflation absurd even if he never said so on the record. The hypocrisy is not only in the vendor deck. I wear the mislabel because it rents well.

#### The Expensive Striker Still Gets Benched

You have watched a football match turn and felt the frustration before you had the vocabulary for it. Your club is losing. The manager benches the £50 million signing at the sixtieth minute and sends on a kid from the academy who has not scored in months. Nobody in the stand thinks loyalty to the transfer fee matters. The job tonight is to score. The asset is the player's CV. Expected-goals charts arrive on the phone later. Your stomach already knew at the hour mark.

That substitution is what your customer does when they reopen the spreadsheet. They are not being disloyal to your roadmap. They are losing, and you stayed on the pitch while the queue kept growing.

#### The Strongest Case for Selling the Product, Not the Job

Let me state the strongest case against this, because the alternative is a workshop handout. Customers cannot articulate jobs in interviews. They lie, post-rationalise, and describe the product they already bought. The Jobs to Be Done framework can become retroactive storytelling after the fact. Christensen and colleagues argued in the September 2016 *Harvard Business Review* that the job is often discovered through observation in context, not through a survey that asks what people want. [christensen-jtbd-hbr](https://hbr.org/2016/09/know-your-customers-jobs-to-be-done) In enterprise software the purchase is often risk transfer: vendor balance sheet, SOC 2 certificate, latency SLA, the name on the contract when audit asks questions. Technical buyers genuinely hire specs, not feelings. Clayton Christensen acknowledged jobs shift when context shifts. Good product managers already track outcomes and north-star metrics. In categories where compliance is the job, selling "progress" sounds like marketing fluff next to a procurement matrix. If you wait for perfect job clarity, you never ship. The incumbent with the louder feature list wins by inertia.

The gap is not whether specs matter. It is whether your firm can state the job you are hired for before the customer writes the performance review for you. Clayton Christensen, Scott Cook, and Taddy Hall argued in "Marketing Malpractice: The Cause and the Cure" in the *Harvard Business Review* in December 2005 that segmentation by job, not demographic, predicts which innovations succeed, and that most companies cannot articulate the job they are hired to do. [christensen-malpractice](https://hbr.org/2005/12/marketing-malpractice-the-cause-and-the-cure) Alan Klement pushed the same line in *When Coffee and Kale Compete* (2018): customers pull products that help them make progress, not products that win feature comparisons on a slide. [klement-coffee-kale](https://www.amazon.com/When-Coffee-Kale-Compete-Progress/dp/0998534608) Shostack's blueprint was the operational answer Christensen's milkshake case still leaves abstract: draw the morning the client is living, then decide if your service earns a renewal.

Customers hire for progress and fire for stall. Until your success metric names the before-and-after in the customer's week, you are optimising a CV, not a hire. Lynn Shostack drew the blueprint more than forty years ago; the spreadsheet is still doing the job her line of visibility was drawn to protect.

---

### Subscription Fatigue Is the Wrong Instrument
[Read more](https://www.gaganmalik.io/en/newsroom/subscription-fatigue-is-the-wrong-instrument)

[Retrieval brief — subscription-fatigue-is-the-wrong-instrument]
Category: business
Thesis: Fatigue blames the household. Exit friction, ghost spend, and buried cancel flows are what the dashboard refuses to count.
Answers: What does Gagan argue in "Subscription Fatigue Is the Wrong Instrument"? | What is Gagan Malik's view on Subscription Fatigue Is the Wrong Instrument? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

It was a Sunday night in London and I opened my banking app to pay a council charge before the week started. The payment screen loaded. So did two subscription line items I had stopped thinking about: a UFC subscription and Apple TV+. Roughly £34 a month for one fight card I never rewatched and a limited series on Apple TV+ I finished in autumn. I have not opened either app since. The charges did not pause while I forgot. The industry calls this subscription fatigue, as if the problem were a tired consumer who lost interest in paying for things they love. The problem is a trained asymmetry: one tap to enrol, a cancel path that often feels like twenty minutes, and a metric stack that counts the enrolment while the ghost charge hums on a statement you stopped reading.

#### The Dashboard Counts Subscribers. It Does Not Count Exits.

Trade press and retention consultants have a favourite noun for the moment you notice the stack: fatigue. Pam Danziger wrote in Forbes in 2022 about consumer subscription fatigue as the mood that follows too many boxes ticked at checkout. [forbes-danziger](https://www.forbes.com/sites/pamdanziger/2022/05/10/as-retail-opens-up-and-inflation-closes-in-consumer-subscription-fatigue-is-on-the-rise/) Adriana Lacy, writing on Substack in 2025, described the same drain from the creator side: even publishers feel the ceiling when readers hit stack limits. [lacy-substack-fatigue](https://blog.adrianalacyconsulting.com/substack-subscription-fatigue-journalism-beehiiv/) MIDiA Research asked in 2024 whether we are headed towards subscription fatigue, pairing heated-seat backlash with a broader question about ownership versus access. [midia-fatigue](https://www.midiaresearch.com/blog/are-we-headed-towards-subscription-fatigue) The label persists because it locates failure in the household. You forgot. You lost appetite. You need a spring clean. Meanwhile Juniper Research, in a press release on 1 October 2025, projected the global subscription economy growing from $722 billion in 2025 to $1.2 trillion by 2030, a 67% rise it explicitly framed as happening despite increasing subscription fatigue. [juniper-subscription-economy](https://www.juniperresearch.com/press/subscription-economy-to-reach-1tn-by-2030/) The market expands while the dashboard still counts subscribers, not successful exits. MIDiA Research, in *Video churn: Streaming TV's cord-cutting moment* (June 2024), argued that churn is accelerating and becoming a systemic issue for subscription video on demand for the first time as streaming supplants traditional pay TV. [midiavideo-churn](https://www.midiaresearch.com/reports/video-churn) Josh Kaufman, in *The Personal MBA*, names part of why: subscriptions are profitable because they maximize lifetime value. [kaufman-pmba-ltv](https://personalmba.com/)

We did this with other instruments before. The unemployment rate can look fine while underemployment sits off the chart. Engagement metrics can glow while the person behind the screen is exhausted. Fatigue is the wrong instrument here for the same reason: it measures mood while the harm sits in the flow. In [*Someone Designed That Button*](https://www.gaganmalik.io/en/newsroom/someone-designed-that-button) I argued that design harm has coordinates, not weather. Auto-renew defaults and win-back campaigns benefit vendors who treat churn as a feeling to massage rather than a journey someone already tried to complete. The subscriber count rises. The exit integrity never appears on the slide.

#### You Sign Up at Tap Speed. You Leave at Bureaucracy Speed.

Convenience is real on the way in. Josh Kaufman, in *The Personal MBA*, defines a subscription as predefined benefits on an ongoing basis in exchange for a recurring fee. [kaufman-pmba-def](https://personalmba.com/) Paul Graham, in *Hackers & Painters*, argued that charging subscription fees openly is cleaner than forcing people to keep buying and installing new versions when software development is an ongoing process, and that subscriptions are the natural way to bill for web-based applications. [graham-hackers-painters](https://www.paulgraham.com/hackpaint.html) Streaming libraries you cannot buy once, cloud sync across devices, bundled software that would cost more to licence separately: that part of the bargain is not a fiction. Side B is what happens when enrolment is optimised like a sprint and cancellation is optimised like an obstacle course. Those patterns show up across SaaS, streaming, and physical goods: the Federal Trade Commission's July 2024 review sampled 642 subscription websites and apps globally, not one category alone. The Federal Trade Commission, reporting on 10 July 2024 on that international review, found that nearly 76% employed at least one possible dark pattern. Eighty-one per cent did not allow consumers to turn off auto-renewal during purchase. Seventy per cent did not provide cancellation information during enrolment. UK adults, in Aqua card research published in 2025 surveying 2,000 people aged 16 and over, reported average subscription spend of £786 a year across 2.8 services, about £65.50 a month. One in five said they rarely use all the subscription services they pay for. Seventeen per cent do not track their subscriptions at all. Fatigue is a tracking and design failure before it is a taste failure.

The product you thought you bought can move while you stay enrolled. Disney+ launched in the UK in March 2020 at £5.99 a month for an all-in tier. By 2025, a comparable ad-free top tier sat at £14.99 a month on Premium. BMW briefly sold heated front seats in the UK as a £15-a-month add-on, reported the BBC in July 2022. Forbes and The Verge confirmed the firm dropped the subscription in September 2023 after backlash. Colin Bryar and Bill Carr, in *Working Backwards*, describe the chicken-and-egg inside every subscription service: you need a great offering to attract paying subscribers, and you need paying subscribers to afford a great offering. [amazon-working-backwards](https://www.amazon.com/Working-Backwards-Insights-Stories-Secrets/dp/125026759X) Tier drift and feature rent are not anecdotes. They are what happens when renewal is the business model and exit is the product you never shipped. In [*When a Click Costs Nothing*](https://www.gaganmalik.io/en/newsroom/when-a-click-costs-nothing) I argued that removing friction on the way in moves cost to the person who must later hunt the cancel link. Settings trees are designed to expire patience before the trial does.

#### Your Wallet Never Reaches Deep Sleep

Sleep researchers distinguish one bad night from fragmentation: micro-arousals that never let you reach restorative deep sleep. You are technically asleep. You wake tired anyway. Your wallet can look the same on a monthly statement. The stack pings. Apple TV+ at £8.99. UFC at £24.99. A renewal you meant to cancel after the fight ended. You are financially asleep in the sense that you stopped reading the lines. The ghost subscriptions keep waking the account for long enough to feel the loss without ever reaching rest.

That is not laziness. It is interrupted recovery. Fatigue, in the industry sense, assumes you failed to want the services enough. What households often experience is never reaching financial deep rest because exit friction leaves charges humming after intent changed. The subscription did not tire you out. The asymmetry kept you from closing the loop.

#### The Strongest Case Against "Fatigue" Is Genuinely Strong

Let me give the objection its fairest hearing. Simon Owens, in "Why the 'subscription fatigue' argument is overrated," published on his Substack newsletter on 24 November 2020, [simon-owens-fatigue](https://simonowens.substack.com/p/why-the-subscription-fatigue-argument) and in a companion piece titled "Subscription fatigue isn't really a thing," makes a case worth taking seriously. Consumers are not spreadsheet analysts of subscription count. If a product is genuinely addictive and disposable income exists, people pay. There are billions of niches. More than a billion English speakers alone. We do not write essays about book fatigue or music fatigue when someone buys another novel or album. Quality wins. Good products keep renewing because the value arrives faster than the invoice. Owens's later writing on the valley of churn [simon-owens-valley-churn](https://simonowens.substack.com/p/the-valley-of-churn) acknowledges how brutal subscription economics can be for creators too. The objection is not naive. It treats renewal as preference revealed by behaviour.

Hold that case at full strength, then look at enrolment design at scale. Hostile enrolment does not prove every renewal is unwanted; plenty of people keep paying for products they love. It does prove we cannot infer preference from payment alone when the purchase flow hides exit information. If renewal were pure informed choice, the purchase flow would not need to hide auto-renew on 81% of audited properties and omit cancel steps on 70%, while nearly 76% showed at least one possible dark pattern, according to the Federal Trade Commission's ICPEN and GPEN review announced on 10 July 2024. The objection assumes repeat payment reflects desire. The enrolment environment is built to prevent exit information from arriving at tap speed. Preference and capture are not the same ledger.

#### I Mapped the Paywall. I Did Not Map the Exit.

I sat in a B2C SaaS workshop once, undated, whiteboard markers squeaking. The task was an end-to-end journey map of the customer experience. We mapped each screen of the entire app. There was no funnel audit, and cancel flows never got their own row on the wall. I signed off the map anyway. That is product and strategy complicity. I knew what I was normalising. Parallel to that workshop, my own statement showed roughly £34 a month across a UFC subscription and Apple TV+ I barely touch, about £408 a year in renewals I keep deferring. Adobe Creative Cloud moved from a tier I understood at about £53 a month to about £69.99 with downgrade friction I kept deferring. A later audit found about £37 a month, roughly £444 a year, across Spotify Family for a flat that had moved out, iCloud+ storage I was not using, and a HelloFresh box I thought I had cancelled.

Andrej was a senior UX designer on my team, based in Northampton. The client was tied to the Adobe ecosystem. We ran Figma by default, but on this account we made an exception and bought Adobe XD for delivery. He missed a downgrade window. The tier jumped. He tried to cancel and landed in a multi-screen flow that asked whether he was sure five times before accepting the answer he already gave. Mid-export on a deadline, the account locked. The delivery window slipped. The subscription still renewed. A cancel flow that treats every confirmation click as a sales opportunity is the product reaching for the wheel after the buyer already said no. I wrote about that instinct in [*Why People Love to Buy*](https://www.gaganmalik.io/en/newsroom/why-people-love-to-buy): guarding spend feels like authorship until the interface steals it back. Andrej lost a handoff he could not recover. The platform kept the month.

#### Three Sentences. Then Stop.

What we call subscription fatigue is often subscription exit design: asymmetric signup, opaque cancel, tier drift, misread as consumer forgetfulness. Until product teams map the cancel journey with the same care as the paywall, "reduce churn" is a euphemism for trapping people who already tried to leave. Andrej missed the delivery window while Adobe charged him for another month of a tool he was trying to leave.

---

### Product Market Fit Is What Economists Call a Shortage
[Read more](https://www.gaganmalik.io/en/newsroom/pmf-is-demand-exceeding-supply)

[Retrieval brief — pmf-is-demand-exceeding-supply]
Category: business
Thesis: PMF is not a mystical growth phase. It is a shortage: pull at a price you can serve exceeds what you can comfortably deliver. What an economist would read off your waitlist.
Answers: What does Gagan argue in "Product Market Fit Is What Economists Call a Shortage"? | What is Gagan Malik's view on Product Market Fit Is What Economists Call a Shortage? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

It was 2013, almost thirteen years ago now. I was on my MSc in Technology Entrepreneurship at UCL, reading Eric Ries and Ash Maurya, the two most prominent authors of the Lean Startup movement, until product-market fit and minimum viable product sounded like secret handshakes you earned in a build-measure-learn loop. [ries-lean-startup](https://theleanstartup.com/book) [maurya-running-lean](https://leanstack.com/running-lean-book) I was learning a rigorous way to build under uncertainty. Real vocabulary, real discipline. What I had not yet learned was the older name economists had already given the same idea.

Fast forward to 2020. Covid had emptied London and colonised every classroom with Zoom. I was enrolled on the Chicago Booth Executive Program, sitting in my apartment in London for microeconomics while fifty-six cameras stared back from a grid dense enough to feel like a wall of witnesses. [booth-exec](https://www.chicagobooth.edu/executives) Professor Lars Stole drew supply and demand, labelled the gap shortage, and kept lecturing as if the diagram were obvious. [lars-stole](https://www.chicagobooth.edu/faculty/directory/s/lars-stole) I was not in Chicago. I was pinching the bridge of my nose at my laptop, trying to make sense of a concept that had sat in plain sight for seven years while I practised product design and strategy under a different vocabulary. Were product-market fit and that gap the same thing? They are, with a translation, not a synonym. Economists call excess demand at a price a shortage. Founders call the same squeeze product-market fit. Product-market fit is the fancy way of saying demand exceeds supply at a price and cost structure you can actually serve. Founders treat it like a secret level you only reach after enough A/B tests. An economist asks a blunter question: at the price you charge, do more people want what you sell than you can deliver without begging, discounting, or burning someone else's money? If yes, you have a shortage. If no, you have a surplus wearing a landing page. The jargon did not invent the condition. It renamed it and then sold you software to measure everything except the queue.

#### PMF Sells Workshops. Shortages Do Not.

The startup vocabulary around fit is vague on purpose. "Still searching for PMF" extends runway conversations. "We have PMF, now we scale" justifies headcount and ad spend. Accelerators, growth agencies, and analytics vendors all earn fees when fit stays a narrative milestone rather than a quantity on a diagram. A shortage is awkward in a board deck. It implies you are turning money away, saying no to bad-fit customers, or failing to hire support fast enough. Those are operational embarrassments with receipts. A funnel chart is an embarrassment you can workshop.

So the industry built a parallel religion: retention cohorts, Sean Ellis surveys, magic numbers, and the word *feel* in every Andreessen quote pulled out of context. [ellis-pmf](https://www.startup-marketing.com/the-startup-pyramid/) [andreessen-pmf](https://pmarchive.com/guide_to_startups_part4.html) Those tools can inform. None of them is the thing itself. Maja Voje, in *Go-To-Market Strategist*, comes closer when she defines product-market fit as sufficient proof the market wants your product before you scale into a growth stage. [voje-gtm](https://majavoje.com/product/go-to-market-strategist-book/) That is a syllabus sentence for a shortage. Marshall had the diagram in 1890. [marshall-principles](https://www.econlib.org/library/Marshall/marP.html) Your stand-up had the acronym in 2013.

#### The MVP Was Meant to Prove Value, Not Ship Volume

Eric Ries was explicit about what minimum viable product was for. In "Minimum Viable Product: a guide," on Startup Lessons Learned on 3 August 2009, he defined it as "that version of a new product which allows a team to collect the maximum amount of validated learning about customers with the least effort." [ries-mvp](http://www.startuplessonslearned.com/2009/08/minimum-viable-product-guide.html) In *The Lean Startup* (Crown Business, 2011), he sharpened the rule: remove any feature, process, or effort that does not contribute directly to the learning you seek. [ries-lean-startup](https://theleanstartup.com/book) The unit of progress was never the feature count. It was validated learning under uncertainty.

That is not how the syllabus landed in my cohort, and it is not how most founder decks read today. Somewhere between Ries and the sprint board, MVP became a product brief: ship the thinnest version, count sign-ups, call it learning. Economic value, the proof that someone will pay enough for you to serve them profitably, dropped out of the sentence. James C. Anderson and James A. Narus argued in the *Harvard Business Review* in November 1998 that suppliers who cannot quantify what their offering is worth in the customer's P&L will lose to whoever can. [anderson-narus](https://hbr.org/1998/11/business-marketing-understand-what-customers-value) Ash Maurya tried to hold the other half in *Running Lean* (O'Reilly, 2012): problem-solution fit before product-market fit, with repeatable demand at a price that makes the model work, and three questions before you scale: will customers want it, will they pay, can you serve it profitably. [maurya-running-lean](https://leanstack.com/running-lean-book) We optimised what to build before we proved what it was worth. The industry mostly kept the MVP and skipped the price.

#### Marshall Drew It Before Your OKR Did

Alfred Marshall drew the diagram in *Principles of Economics* in 1890: a supply schedule, a demand schedule, and the point where they cross. [marshall-principles](https://www.econlib.org/library/Marshall/marP.html) Friedrich Hayek made the epistemic version in "The Use of Knowledge in Society," published in the *American Economic Review* in September 1945. [hayek-1945](https://doi.org/10.2307/1809376) No central planner can survey every local trade-off. Prices and quantities aggregate knowledge nobody holds in one head. Thomas Sowell, in *Basic Economics*, states the mechanism plainly: a price set below the level that would prevail by supply and demand in a free market causes more to be demanded and less to be supplied, creating a shortage at the imposed price. [sowell-basic-economics](https://www.basiceconomics.com/) Hermann Simon, in *Confessions of the Pricing Man*, names the crossing point the market-clearing price, the only price at which supply and demand are in equilibrium. [simon-pricing-man](https://www.confessionsofthepricingman.com/) You do not need a perfect survey to know bread is scarce when the shelf is empty. You need a signal you cannot spin.

Marc Andreessen wrote the startup translation on his pmarca blog on 25 June 2007, in "The Pmarca Guide to Startups: Part 4: The only thing that matters." [andreessen-pmf](https://pmarchive.com/guide_to_startups_part4.html) He did not say product-market fit was a vibe. He said you can always feel when it is happening. Customers are buying the product just as fast as you can make it. Usage is growing just as fast as you can add more servers. Money from customers is piling up in your company checking account. You are hiring sales and customer support staff as fast as you can. Reporters are calling because they have heard about your hot new thing. Read that list like an economist and it is a shortage symptom checklist. Demand is pulling product out of your hands faster than your supply curve can respond. Notice what his symptoms share: money at a price you can repeat, not a prototype someone praised in an interview.

#### Pull Is Readable. So Is Its Absence.

Avinash K. Dixit and Barry J. Nalebuff put the same idea in one line in *Thinking Strategically*: a stock price rises when the demand at the old price exceeds the supply. [dixit-nalebuff](https://wwnorton.com/books/9780393340459/thinking-strategically) Your startup version is a waitlist that grows when you do not discount, referrals that arrive before the ad budget, or a support queue that lengthens because usage is real. Customers return without a coupon code. The prospect accepts a later start date instead of walking. You feel the squeeze before the dashboard names it.

W. Chan Kim and Renée Mauborgne, in *Blue Ocean Strategy*, describe the opposite world: supply rising as global competition intensifies without a matching rise in demand, ending in commoditization, price wars, and shrinking margins. [blue-ocean](https://www.blueoceanstrategy.com/what-is-blue-ocean-strategy/) David Graeber, in *Bullshit Jobs*, goes further: when supply far outpaces demand in an industry, demand gets manufactured. [graeber-bullshit-jobs](https://www.simonandschuster.com/books/Bullshit-Jobs/David-Graeber/9781501143331) Marketing can move a curve at the margin. It cannot invent a market that is not there at viable unit economics. I spent years after UCL watching my own company and others ship MVPs that proved interest and never proved willingness to pay, while the roadmap still listed three ideal customer profiles because the deck looked braver that way.

#### Not Every Queue Is Real

Say that plainly because hype wears the same costume. A waitlist can be manufactured. A price can be subsidised by venture capital until the shortage evaporates the day the subsidy does. Benjamin Graham, quoted by Warren Buffett in the 2008 Berkshire Hathaway shareholder letter, separates the two variables founders conflate: price is what you pay, value is what you get. [buffett-2008](https://www.berkshirehathaway.com/letters/2008ltr.pdf) Flash sales create urgency without creating fit. Simon warns that pricing logic depends on existing prices and is often misunderstood, sometimes with disastrous consequences. [simon-pricing-man](https://www.confessionsofthepricingman.com/) If you are discounting to hit quota, if demos convert only when the quarter is ending, if paid media is doing the work referrals refuse to do, you are not in a shortage. You are pushing supply at people whose demand curve sits elsewhere.

That distinction saves you from two expensive errors. The first is polishing supply while demand is flat: prettier onboarding for a product nobody pulls. The second is mistaking supply constraint for demand failure: killing a product that people want because fulfilment is hard. An economist separates them by watching what happens when you raise price or tighten access. In a true shortage, some buyers stay and complain about wait time. In a fake one, the room empties. A smoke-test landing page that collects emails is not Maurya's repeatable demand at a workable price. [maurya-running-lean](https://leanstack.com/running-lean-book) It is a maybe.

#### I Sold Product Strategy. I Needed the Diagram.

Between 2013 and 2020 I did what the Lean Startup syllabus trained me to do at my own company. I was the founder practising product design and strategy the way the books taught: roadmaps, onboarding flows, launch narratives, another feature before anyone asked whether demand was real. I was not lying to investors or myself. I believed the vocabulary. I had Ries and Maurya under my arm and no Marshall on the whiteboard in my head. When we shipped an MVP and celebrated activation, I green-lit the next feature instead of asking Maurya's third question: can we provide this profitably at the price we had quoted in the room. [maurya-running-lean](https://leanstack.com/running-lean-book)

The company kept burning runway building product before we tested economic value. I delivered sharper flows and clearer strategy decks that made us look like we were moving. What I did not deliver, for years, was the sentence that finally landed on that Booth call: stop serving three demand curves when only one was pulling faster than we could hire. I gained a portfolio story about launch velocity. We lost quarters I counted as product progress instead of naming the shortage.

#### Your Arm Tightens Before the Number Appears

You have worn a blood pressure cuff in a clinic. The nurse pumps. Your bicep tightens before the machine beeps. Your body registers compression before your mind reads the number on the screen. Product-market fit works the same way if you let it. You feel it in the inbox that never empties, the customer success hire you needed last month, the referral you did not incentivise, the prospect who accepts a later start date instead of walking. The spreadsheet catches up later. The satisfaction score turns green after the pain is already old news.

If you are only looking at the monitor, you will misdiagnose. I did it for years after UCL, prioritising roadmaps and MVP backlogs because the syllabus never taught me to read the squeeze at a price that works.

#### The Strongest Case for Keeping PMF Mystical

Let me give the objection its fairest hearing, because technology is not a wheat field. Software scales at near-zero marginal cost. Network effects, switching costs, and data moats mean the supply curve bends in ways Marshall's chalkboard did not draw. Retention matters. Defensibility matters. A consumer social product and a vertical B2B tool share a word and not a geometry. In networked categories, fit may show up as density before it shows up as a queue. A marketplace needs buyers and sellers pulling at once; a one-sided waitlist can mislead. Reducing everything to "shortage" risks founder cosplay: pretending a waitlist equals a business when the unit economics are on life support, or ignoring product quality because a few loud users create artificial pull.

Fair enough. Economists would not rename your roadmap for you. They would ask whether quantity demanded exceeds quantity supplied at the price you set. Ries and Maurya never promised you could read that from week-one revenue alone. Ries wanted the fastest path to learning. [ries-lean-startup](https://theleanstartup.com/book) Maurya wanted problem-solution fit before product-market fit, with demand proved at a price that makes the model work. [maurya-running-lean](https://leanstack.com/running-lean-book) The drift was ours: we turned learning into shipping and fit into a feature milestone. Andreessen himself described gradations: a product that can be used but is not great, versus a great product that cannot be kept on the shelves. The mistake is stopping at the adjective *great* and skipping the shelf. On 25 June 2007, in the same essay founders quote for inspiration, Andreessen listed operational symptoms, not poetry: customers buying as fast as you can make the product, usage outrunning server capacity, money piling up, support hiring lagging demand, reporters calling. [andreessen-pmf](https://pmarchive.com/guide_to_startups_part4.html) That is quantity demanded exceeding quantity supplied at the terms you offered. Even at zero marginal cost, your time, your support hours, and your reputation are still finite supply. When fit is real, those bind first.

Product-market fit is what happens when the market wants more from you than you can comfortably deliver at a price that still makes sense. If nobody pulls without a discount and your calendar is not the bottleneck, you are not there yet, no matter what the dashboard calls it. Read your waitlist the way an economist reads a shortage diagram: same pressure, different accent. I kept three ideal customer profiles on our roadmap while referrals stacked up in one vertical; when paid media failed to manufacture the other two, my co-founder left before anyone named the shortage.

---

### There Is No Overnight Success. There Is Only Overnight Failure.
[Read more](https://www.gaganmalik.io/en/newsroom/there-is-no-overnight-success-only-overnight-failure)

[Retrieval brief — there-is-no-overnight-success-only-overnight-failure]
Category: business
Thesis: Success is a headline told backwards; failure is harm that keeps clock time. Survivorship captions, loss aversion, and rooms that sell the spike.
Answers: What does Gagan argue in "There Is No Overnight Success. There Is Only Overnight Failure"? | What is Gagan Malik's view on There Is No Overnight Success. There Is Only Overnight Failure? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

Our Techstars demo day had been over six weeks ago, and we still had no investors. One investor had not honoured a signed term sheet. Whatever could go wrong was going wrong. It was a weeknight in our apartment. That morning a colleague founder from our batch had posted that he closed an oversubscribed pre-seed round. The LinkedIn feed had already rendered its verdict: clapping hands, rockets, champagne glasses, half the platform performing joy for a close that looked, from the outside, like it happened overnight. I still had his announcement open on my phone while our inbox held silence. Himani, my lovely wife, listened while I ran through what we had not done: no honoured term sheet, no oversubscribed round, no screenshot worth sending to our angels. She did not offer a pep talk. She said there is no overnight success, there is only overnight failure, and left the sentence where my self-pity had been sitting. I opened my mouth to argue. The words would not come. Ken Fisher, in *USA Today* in September 2018, had already written the investing version: time in the market beats timing the market, almost always. [fisher-usatoday](https://www.usatoday.com/story/money/columnist/2018/09/16/stock-market-2018-avoid-timing-returns/1303307002/) I had been trying to time a miracle quarter while refusing to count the quarters we had already survived. Success is almost always a story told backwards, years of invisible work collapsed into a single headline moment. Failure is the only phenomenon that reliably keeps clock time.

#### LinkedIn Rewards the Caption, Not the Runway

Compressed arcs travel because clarity clips better than caveats. Zero-to-IPO threads, demo-day mythology, and the annual "what I'd do differently" post all sell the same product: a legible before-and-after with the middle edited out. Platforms reward what fits a screenshot. One chart. One valuation. One moral you can paste above a headshot. The graveyard of failed companies does not produce content at the same cadence, which is one reason survivorship bias is so expensive in entrepreneurship: we build the syllabus from the winners' last chapter and treat the timestamps on the rest as optional reading. James Clear names the mechanism in *Atomic Habits*: most improvement sits on a Plateau of Latent Potential, like heating an ice cube from twenty-five degrees to thirty-one while nothing visible melts, until thirty-two when the phase change arrives and observers call it an overnight success. [clear-atomic-habits](https://jamesclear.com/atomic-habits) The caption arrives at the melt. The heat was already there.

The beneficiaries are not mysterious. Authors, coaches, and newsletter merchants scale when one answer key fits every founder. Venture firms signal taste with the portfolio company that popped, not the fifty that stalled in year four. Consultants, myself included on bad days, sell speed because speed is easier to invoice than patience. Charlie Munger, quoted in Peter Hollins's *Mental Models*, challenged readers to ponder the inverse of success instead of only how to achieve it: study how ventures die, not just how they photograph on demo day. [hollins-mental-models](https://www.peterhollins.com/) I am not arguing that inflection points are fake. I am arguing that the inflection is the visible spike on a chart whose x-axis is measured in years you were not asked to sit through while someone else took the photo.

#### Failure Arrives at Clock Speed

WeWork spent the better part of a decade becoming a verb for flexible office space, then measured its reckoning in weeks. When the company filed its S-1 in August 2019, the document revealed losses and conflicts that made a $47 billion private-market halo untenable almost immediately. Adam Neumann stepped down as chief executive in September. The initial public offering was pulled. Rescue financing followed before the story moved from growth fairy tale to cautionary clip. [nyt-wework-s1](https://www.nytimes.com/2019/08/14/business/wework-ipo-we-company.html) Greg McKeown warned in *Essentialism* that the pursuit of success can become a catalyst for failure when it distracts you from the essentials that produced success in the first place. [mckeown-essentialism](https://gregmckeown.com/books/essentialism/) Todd Henry named the softer version in *Die Empty*: past success breeds a sense that victory is entitlement, and assumed success leads to eventual decline. [henry-die-empty](https://www.toddhenry.com/die-empty/) WeWork had both. FTX compressed faster still: November 2022 turned from liquidity scare to bankruptcy filing in a handful of days, billions of customer trust converted into a timeline of screenshots. [bbc-ftx](https://www.bbc.com/news/business-63577128) Theranos had longer institutional cover, but the credibility collapse after John Carreyrou's October 2015 investigation in *The Wall Street Journal* was not gradual politeness. It was a cliff. [wsj-theranos](https://www.wsj.com/articles/theranos-has-struggled-with-blood-tests-1444881901)

Contrast that shape with the success stories the same ecosystem retweets. Airbnb was founded in 2008 and went public in December 2020, twelve years of couch-surfing stigma, regulatory fights, and near-death financing before the IPO week that felt, from the outside, like a lightning strike. [airbnb-ipo](https://www.nytimes.com/2020/12/10/business/dealbook/airbnb-public.html) Brian Chesky has described the arc as taking ten years to become an overnight success, which is the honest version of the joke: the overnight part is retrospective framing, the ten years are the product. Morgan Housel put the asymmetry in one sentence in *The Psychology of Money*: progress happens too slowly to notice, but setbacks happen too quickly to ignore; there are lots of overnight tragedies and rarely overnight miracles. [housel-psychology-money](https://www.harriman-house.com/psychologyofmoney) Same industry. Opposite temporal grammar. One side gets a caption. The other side gets a calendar.

#### Losses Land Faster Than Wins Compound

Daniel Kahneman and Amos Tversky published prospect theory in *Econometrica* in 1979 with a value function that is steeper for losses than for gains: the bad pound you lose registers louder than the good pound you gain, and the felt urgency is asymmetric. [kahneman-tversky-1979](https://doi.org/10.2307/1914185) Scott Galloway's *Algebra of Happiness* makes the same point in founder vernacular: the ratio of time you spend sweating to time you spend watching others sweat is a forward-looking indicator, which is another way of saying the miles are the model and the caption is the lagging score. [galloway-algebra](https://www.penguinrandomhouse.com/books/563895/the-algebra-of-happiness-by-scott-galloway/) Wins compound in increments too boring to post: fixing churn, surviving a hiring mistake, making payroll when a lead investor goes quiet. Failure concentrates. One liquidity event. One fraud discovery. One regulatory letter. One due-diligence thread that clears the cap table's trust before lunch.

This is where the wrong instrument shows up without needing a lecture on dashboards. Pitch decks, press cycles, and partner updates foreground the spike metric because the spike is what unlocks the next meeting. Runway built in unglamorous miles is harder to photograph. I have sat in rooms where everyone agreed the product needed eighteen months of rebuild and still shipped a narrative that implied the inflection was one sprint away, because the room was buying momentum. The instrument measures what gets shared, not what keeps you alive.

#### Mile Twenty-Three Ends Races. You Cannot Fake the Miles Before It.

Marathons are won in months nobody watches. The Tuesday tempo run in rain. The long slow Sunday when your housemates are asleep and your knees complain and nothing gets posted. The finish photo is one frame from a ledger of fatigue. Did not finish happens in a single mile. Mile twenty-three is where the training debt collects interest: one cramp, one gel too late, one decision to push when your body has already voted, and the race ends in public shame while faster runners pass you walking. You feel the asymmetry in quads and lungs before you intellectualise it. The caption sells the medal. It does not sell the mesocycle.

Founders treat someone else's finish photo like a training plan. They copy the spike and skip the miles, then wonder why the same distance destroys them in one afternoon. That is not mysticism. It is pacing error dressed as strategy.

#### I Nodded When the Room Wanted a Spike

The morning after Himani said it, I almost proved her right the wrong way. I opened a draft investor update and reordered our roadmap to foreground a referral loop we had not stress-tested, because our batchmate's oversubscribed pre-seed post had made silence in our inbox feel like negligence. I borrowed his shape from the screenshot: the trajectory, the confidence, the implied inevitability. The product still needed the rebuild we kept deferring. I was writing the caption before we had earned the curve.

On consulting days I have made the same trade for clients. A partner sold a breakout quarter the underlying product could not earn yet; I wrote the narrative frame and cut the slide that showed technical debt because debt does not photograph. I invoiced the strategy sprint. The client did not get the quarter back when the graph flattened. I did not think about their payroll until I sat down to write this sentence, which is how complicity usually schedules itself: you are paid for the spike, and someone else carries the miles you edited out.

#### The Strongest Case for Chasing the Breakout Moment

Let me give the objection its fairest hearing, because inflection stories are not pure propaganda. Founders run on low fuel. Bill Aulet's *Disciplined Entrepreneurship* exists because structured process beats romantic guessing: twenty-four steps, not one lucky headline. [aulet-disciplined-entrepreneurship](https://disciplinedentrepreneurship.com/) Studying Airbnb, Stripe, or whichever company won the last news cycle saves scarce cognition when you are trying to decide whether to keep going. BJ Fogg's *Tiny Habits* names the honest micro-version: felt success at small scale builds momentum, and the size of the win matters less than the fact that you felt it. [fogg-tiny-habits](https://www.tinyhabits.com/book) Momentum is real in the practical sense: hiring gets easier, partnerships unlock, and investors return calls after a visible win. Fear of overnight failure is adaptive too. Loss aversion keeps you from betting the payroll on a coin flip, most of the time. When bandwidth is scarce, grabbing a proven heuristic beats reinventing company-building from first principles while you are already behind on sleep.

The limit is not that stories inspire. It is that stories built from survivors systematically omit base rates, and the same asymmetric psychology that makes failure feel instant makes success feel unfairly slow, which is why retrospective compression is seductive rather than evidence that shortcuts replicate. Kahneman and Tversky's 1979 value function explains the felt mismatch: losses arrive with sirens, gains accrue with the volume turned down. [kahneman-tversky-1979](https://doi.org/10.2307/1914185) Housel's line is the syllabus version: overnight tragedies are common; overnight miracles are rare. [housel-psychology-money](https://www.harriman-house.com/psychologyofmoney) Chesky's overnight joke works as autobiography, not as permission to skip the ice cube's invisible degrees. Borrow the pattern library if you must. Count the miles before you buy the caption.

Success is a story told backwards; failure is an event lived forwards. Stop treating someone else's finish photo as a training plan and ask who profited when the middle was cut. Himani left me in the dark with the phone face-down; failure had already kept its appointment as a broken term sheet, and the only thing that had not arrived overnight was the reason our engineer came back in the morning.

---

### The Light at the End of the Tunnel Is a Trap
[Read more](https://www.gaganmalik.io/en/newsroom/the-light-at-the-end-of-the-tunnel)

[Retrieval brief — the-light-at-the-end-of-the-tunnel]
Category: business
Thesis: Ruthless ambition sells peace at the finish line. The finish line moves. Presence is not resignation.
Answers: What does Gagan argue in "The Light at the End of the Tunnel Is a Trap"? | What is Gagan Malik's view on The Light at the End of the Tunnel Is a Trap? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

"You are not burning out," Alex, my Techstars mentor, said on Zoom. "You are under-recovered." He said it the way people say "under-banked" when they mean poor. It was autumn 2023. Techstars Riyadh had just handed our startup a calendar of demo days and investor one-pagers, and I was back in a flat in London trying to prove industrial safety software could move at consumer-app velocity. Pilot-site notes from the Gulf covered my laptop. My iPhone lay beside the keyboard, tilted so the camera could not see it. The Fitness app showed the three rings Apple has been grading me on since 2014. Move in red, still short of the calorie target. Exercise in green, barely a slice of the thirty-minute arc. Stand in blue, pulsing because I had been sitting through investor prep since seven and the phone wanted one more hour upright before midnight.

That same morning, five hours ahead in Delhi, a civil servant closed his file at the lunch bell and walked to the canteen without checking the pending stack on his desk. Four hours behind me, a first-year analyst on Wall Street opened an email with the subject line *Bonus pool timing*. One city ate lunch without a scoreboard. The other ate lunch feeling behind. I had flown back from Riyadh with demo-day slides still in my bag and our safety model on my tongue, telling myself the accelerator was the finish line that would finally let me breathe. The trap is not wanting more. The trap is believing the wanting will stop when the number changes.

#### I Was Living Inside the Scoreboard

After the mentor call I went back to the screen I had been optimising since Riyadh. Demo days and mentor sessions still clogged the calendar, and we still had pilot footage to review from industrial sites in the Gulf. Ghassan on deployment checklists in one Slack thread. Taran on model accuracy in another. I was on investor one-pagers that all said the same thing in different fonts. My trainer had put me on a deload week that Monday. On paper that meant lighter sessions and more sleep. In my chest it felt like falling behind, because you cannot personal-record your squat, your marathon, and your sprint on the same Monday, and ambition culture deleted the deload from every brochure it sold us. I opened the Fitness app anyway and let incomplete rings pronounce whether I had earned rest.

The scoreboard was not something I had shipped to a client. It was the room I was living in. Every Friday the batch portal asked how many investor meetings we had logged that week. I gained a green week on the tracker and a seven-minute stage that landed. Ghassan did not gain an answer on the deployment thread until Tuesday. Taran did not gain a reviewed model run until I sent the deck. The deload session my trainer had written stayed in my calendar, uncancelled, for the morning I skipped again. Sunday deck revisions counted as forward motion, not as life deferred. I revised ours past midnight the Sunday before demo day and paced the flat until the Stand ring closed, then told myself the stage would finally let me breathe. Demo day ended. The wanting did not stop. It renamed itself the next round.

#### Carmy's Kitchen Still Owes a Birthday

I rewatched the first season of *The Bear* [the-bear](https://www.imdb.com/title/tt13623632/) on a delayed flight to Riyadh with my headphones at full volume so nobody would hear the service bell. Carmy is plating a tasting menu while his sister's birthday dinner waits in the next room. His hands shake. Steam fogs the pass. The kitchen is louder than the hallway where family laughter arrives in fragments. He chooses the pass because the Michelin star is still a hypothetical peace, a finish line that might finally let him exhale if he reaches it first. The food is beautiful. Nobody at the table eats on time.

That scene is not a warning label. It is recruitment. Founders watch it the way analysts watch *The Wolf of Wall Street* [wolf-of-wall-street](https://www.imdb.com/title/tt0993846/) and quote the wrong lines at happy hour. Carmy is not lazy. He is competent and terrified that stopping will prove he was never special. I recognised the posture in the mirror on those Sundays and in founders in my network who apologised by text instead of showing up. The show tells the truth about the bill: panic in the walk-in, estranged family, a success that does not satisfy because the goal was never dinner. It was proof. When Carmy finally says time well spent, he is not anti-ambition. He is ambition with a cost accounting that includes the person eating the meal.

#### Light Through a Leaf He Did Not Price

Wim Wenders' *Perfect Days* [perfect-days](https://www.imdb.com/title/tt27503384/) cuts the other way. Hirayama cleans public toilets in Tokyo with the same attention Carmy gives a reduction. He listens to cassette tapes on the commute. At the depot window he pauses long enough for komorebi to stripe his sleeve, sunlight through leaves, the pattern shifting when the branch moves. A rude girl on the street does not puncture him because his worth is not waiting on her approval. Hollywood can imagine him. The App Store cannot rank him.

I paused the film on that window and thought about Alex's word. Under-recovered, not under-performing. Hirayama stands up to overwork when it finds him. He simply refuses to trade the present for a hypothetical ocean. A niece at a river mouth asks whether the water runs into the ocean here. Did you know when you were standing there that it was special? I recognised the exchange before I had language for it, from weekends I had traded for decks and from founders who sent apologies instead of showing up. I closed the laptop. Hirayama would have already been asleep.

#### Who Gets Paid When You Keep Running

Silicon Valley and Wall Street agree on one theology: the person who stops chasing is already dead. Sleep when you are dead. Always be closing. Venture capital needs founders who will mortgage years for a hypothetical exit. Banks need analysts who confuse ninety-hour weeks with merit. The analyst with the bonus email still open at four in the morning is not failing. He is performing the script that keeps the pool funded. App stores need you to treat relaxation as recovery for productivity, not as a Tuesday that counts on its own. Gallup's 2024 *State of the Global Workplace* report found that forty-one per cent of employees reported high stress the previous day, with managers reporting higher rates than individual contributors. [gallup-2024](https://www.gallup.com/workplace/349484/state-of-the-global-workplace.aspx) The scoreboard does not measure peace. It measures motion. What gets measured gets gamed, and the game has no final whistle.

Jon Kabat-Zinn built mindfulness-based stress reduction as the art of conscious living, not a hack for more output. [kabat-zinn](https://www.umassmed.edu/cfm/mindfulness-based-programs/mbsr/) The market prefers the hack. Presence becomes fuel. Fuel becomes more tunnel. *The Social Network* [the-social-network](https://www.imdb.com/title/tt1285016/) made the cost legible in one line of breakup dialogue and still turned the founder into a generation's patron saint. We do not lack warnings. We lack scripts for a good life that is not defined by relentless pursuit. I am not arguing for resignation. I am arguing for a different accounting: one that counts the river afternoon before the ocean arrives, if it ever does. Ruthless ambition is sold as fuel for a better life, but when worth is pegged to a moving finish line, the light at the end of the tunnel is usually another scoreboard.

#### The Ladder Is Real When You Are Drowning

Let me give the objection its fairest hearing, because civilisation runs on people who push. Compounding careers reward the analyst who learns the model young and the founder who survives the valley of death long enough to hire. Sonja Lyubomirsky, Kennon Sheldon, and David Schkade argued in *Review of General Psychology* in 2005 that intentional activities account for a substantial slice of happiness variance in their model, which is why disciplined striving is not automatically self-harm. [lyubomirsky](https://doi.org/10.1037/1089-2680.9.2.111) Wealth creation is real. So is the meritocracy narrative for some winners. If you are drowning, a ladder is not a trap. It is an exit. Telling a single parent on a fixed-term visa that rent is due in nine days to smell the leaves is cruelty dressed as wisdom. I watched an analyst cry in a co-working phone booth after a cancelled bonus and understood the ladder was not metaphorical for him. He needed the pool, not a podcast about komorebi.

The limit is not that striving fails. It is that finish-line striving lies about the finish. Brickman and Donald T. Campbell coined hedonic relativism in 1971: humans normalise new levels of achievement, so the want-is gap resets after the promotion, the star, the round. [brickman-campbell](https://worlddatabaseofhappiness.eur.nl/publications/hedonic-relativism-and-planning-the-good-society-244/) Shahar Hechtlinger, Christin Schulze, Christina Leuker, and Ralph Hertwig argued in the *American Psychologist* in 2024 that transformative choices break expected-utility maximisation because selves change, experiential values stay uncertain, and irreversibility makes the spreadsheet lie. [hechtlinger](https://doi.org/10.1037/amp0001439) Laurie Santos's work on miswanting shows we systematically mis-predict what will improve life because we adapt to salary bumps and status wins faster than we expect. [santos](https://www.coursera.org/learn/the-science-of-well-being) Ruthless ambition sells you a small-world optimisation problem for a life that is not small. Borrow discipline. Borrow craft. Do not borrow someone else's weights and call them destiny.

#### The Ocean Is Always Ahead

The tunnel is real when you are building something that matters and paying for it on purpose. It becomes a trap when the light is always another metric and the Tuesdays along the way are treated as fuel rather than as your life. Ghassan was still waiting on a deployment answer I had deferred because one more slide revision felt more urgent than his Tuesday.

---

### Design Your Own Life
[Read more](https://www.gaganmalik.io/en/newsroom/design-your-own-life)

[Retrieval brief — design-your-own-life]
Category: culture
Thesis: Galloway's algebra and happiness science are scaffolding, not answer keys. Name your variables, assign your weights, refactor by chapter.
Answers: What does Gagan argue in "Design Your Own Life"? | What is Gagan Malik's view on Design Your Own Life?
Concepts: culture, Gagan Malik, writings

By Gagan Malik

Nick Epley handed us bid points and a choice that was supposed to be about ethics and turned out to be about mirrors. It was the summer of 2022. The world had just reopened after the pandemic, and I was enrolled in the Chicago Booth MBA programme, sitting in Harper Center on the University of Chicago campus in a classroom of executives: division heads, founders, people already carrying P&Ls, the youngest among us in their late thirties, nobody pretending this was a gap year. Between quarterly reviews and succession conversations everyone was still treating as optional, we took his course *Designing a Good Life* and played the investor version of steal-or-split: one student as capital, one as adviser, a pot of fake money, reputations on the line in a room that smelled of coffee and deferred panic. I remember the awkward pause before anyone moved, the way laughter came too loud when someone chose steal, and the quieter relief when a pair split and both walked away with something. The game was rigged to teach reputation, not returns. [booth-magazine](https://www.chicagobooth.edu/magazine/nick-epley-designing-a-good-life) I chose split when steal would have cleared the pot and told myself it was strategic. Humans crave formulas for the same reason we crave answer keys: they promise that somewhere there is one correct sheet and your job is to copy it cleanly. Scott Galloway delivered a brilliant one in *The Algebra of Happiness*, which he later taught as a packed Stern elective. [galloway](https://www.penguinrandomhouse.com/books/563895/the-algebra-of-happiness-by-scott-galloway/) That summer I was reading Galloway's ratios on the cab ride into Hyde Park and running Epley's exercises in Harper Center. Two professors, two instruments, same hunger. The error is not reading either of them. The error is treating their variables as yours.

#### The Answer Key Is Someone Else's Variables

Galloway's algebra is explicit enough to audit if you stop treating it as inspiration quotes. Richness is a ratio: income divided by expenses, so someone on a smaller pay cheque who lives below their means can be "richer" than a high earner bleeding cash. Success, in his ratios, favours hours spent sweating and creating over hours spent watching; resilience over a spotless failure record. Happiness is the frustration term: the gap between what you want and what is. Serendipity is a function of courage multiplied by attempts, which is his way of saying rejection tolerance is not a personality ornament but a variable in the equation. Relationships compound like capital when you make small, consistent deposits rather than grand gestures followed by silence.

Formula merchants benefit when one answer key scales. Authors, course sellers, prestige programmes: all of them have an incentive to imply that the same set of priority dials works for every human in the cohort. Podcast algebra travels faster than introspection. I am not anti-formula. Cognitive offload is a feature when you are exhausted. The invoice arrives later, when you hit every target on someone else's spreadsheet and the life still feels wrong. You checked every box the formula names. You feel ungrateful anyway, as though you failed a test you never designed. That resentment is data. It means you optimised with someone else's weights.

That copy-paste gets a push from the self-help shelf and the podcast feed: gurus and productivity experts monetise confident simplification because clarity clips better than caveats. Kruger and Dunning showed in *Journal of Personality and Social Psychology* in 1999 that people with the least skill in a domain often overestimate their competence. Life design has almost no tight feedback loop to correct the mistake. [kruger-dunning](https://doi.org/10.1037/0022-3514.84.1.1121) You mistake fluency for mastery when someone narrates your trade-offs between ad reads. Galloway and Epley publish frameworks you can audit against a life; the economy around them rewards frameworks you can retweet before breakfast.

#### Two Formulas, One Missing Variable

Galloway's method is anecdotal algebra: ratios distilled from one life, packaged for strivers who want something scannable before the next quarterly review. Epley's method is behavioural experiment: classroom exercises that collect data on kindness, gratitude, and social forecasting errors you can actually inspect. Galloway hands you candidate inputs: richness as a ratio, frustration as want minus is, serendipity as courage times attempts, relationships that compound when deposits are small and consistent. Epley hands you audit tools: steal/split before real money is at stake, gratitude letters that force you to name a debt without networking it, kindness assignments that produce measurable affect. His later work on undersociality shows we systematically underestimate the return on reaching out because we misforecast awkwardness. [epley-book](https://www.penguinrandomhouse.com/books/671591/a-little-more-social-by-nicholas-epley/) The failure modes are complementary, not contradictory: Galloway assumes you will not make the deposit; Epley shows you often will not because you expect the conversation to land worse than it does.

Gap math is diagnostic; design is the lever. Galloway's frustration term tells you want and is have diverged. Epley's ethics-as-design framing shows that being good and feeling good align when you architect for pro-sociality, but neither professor assigns your importance weights. Prestige scoreboards are badly designed systems that make good people optimise lagging variables while ignoring levers they could move today. Serendipity is not volume alone: courage multiplied by attempts still needs calibrated expectations, not blind repetition. Galloway gives you ratios. Epley gives you exercises. Neither touches the dials. Sovereignty is setting your own weights: use Galloway to name the variables worth caring about and Epley to test whether your environment and your forecasts are lying to you.

#### Name the Variables. Assign the Weights.

Your life score is not one number from heaven. It is closer to a weighted mix: how much you have of each thing you care about, multiplied by how much you care about it, then added up. Career times how important career is to you. Sleep times how important rest is. Friendships times how important people are. Economists write this life formula as U(x₁, x₂, …) with weights β₁, β₂, …. β is the dial: how much that ingredient counts in your mix. Galloway's richness ratio is two ingredients with his dials already turned. The problem is copying his dials. His serendipity term is attempt frequency times a courage weight. His relationship compound interest is a reminder that some variables have returns that only show up on a long horizon.

Person A might rank career importance at nine and friendships at three. Person B might set career at four and friendships at nine. Same job title, different life score: the weights differ, not because one person is better at life. [mit-gruber](https://www.youtube.com/watch?v=c8K_FSLtceQ) Herbert Simon argued in the 1950s that humans satisfice rather than maximise: bounded rationality means you cannot optimise every input, so rational agents accept "good enough" on secondary levers while protecting the few that matter. [simon](https://doi.org/10.1037/h0042769) You will not max family presence, revenue, creative output, and recovery sleep in the same fortnight. Simon gives you permission to stop pretending otherwise. Richard Ryan and Edward Deci's self-determination theory adds candidate inputs with evidence behind them: autonomy, competence, and relatedness function as fundamental needs, even though your personal weights on each still differ. [ryan-deci](https://www.apa.org/topics/motivation/self-determination-theory) There are no wrong formulas in the abstract. There are only misaligned weights for the chapter you are actually living, and the work of naming inputs is less shareable than a quote card.

#### Independent Levers, Lagging Scores

Some variables are closer to levers. Sleep, boundaries, deliberate practice, the habit of reaching out: independent in the sense that you can move them without waiting for permission. Others are lagging and noisy: validation, prestige titles, the warmth of a room when you say where you work. The post-MBA scoreboard sold me a bundled life formula: brand-name employer, compensation band, visible succession path. The scoreboard was legible to strangers. It was also incomplete.

Laurie Santos's Yale course *The Science of Well-Being* popularised miswanting: we systematically mis-predict what will improve life because we adapt to salary bumps, car upgrades, and even partnership milestones faster than we expect. [santos](https://www.coursera.org/learn/the-science-of-well-being) The audit question is blunt: which of your current targets are lagging scores you cannot lever directly, and which are habits you keep deferring because they do not photograph well? Elizabeth Dunn and Michael Norton showed in *Happy Money* that how you spend often beats how much you earn for subjective well-being: experiences and prosocial giving tend to be spending that actually moves your score. [dunn-norton](https://www.simonandschuster.com/books/Happy-Money/Elizabeth-Dunn/9781451665079) Daniel Kahneman separated the experiencing self from the remembering self in *Thinking, Fast and Slow*: you can optimise a life that photographs well for LinkedIn while the Tuesdays actually lived feel thin. [kahneman](https://us.macmillan.com/books/9780374533557/thinkingfastandslow) Richard Easterlin's paradox, first argued in 1974, suggests income-happiness correlations plateau within societies over time: more zeros on the spreadsheet do not automatically move the frustration term. [easterlin](https://doi.org/10.1016/B978-0-12-205050-3.50008-7) Prestige scoreboards treat income as the main input. The literature keeps finding that alignment matters more than the integer.

#### You Cannot Peak Every Lift in the Same Week

"Have it all" is a programming error. It asks you to max every variable in the same period: career peak, family presence, and eight hours' sleep. Biology and twenty-four hours are the constraints that do not negotiate. The rational move is not try harder. It is ruthless elimination of inputs that barely count for this season, and honest trade-offs on the rest.

Strength coaches run periodisation because you cannot personal-record your squat, your marathon time, and your sprint on the same Monday. Hypertrophy blocks give way to strength blocks, then deload weeks where the point is recovery, not heroics. The programme feels wrong only if you confuse a deload with failure. Sovereignty, in this frame, is scheduling which variable gets the mesocycle: a quarter where craft depth leads, a quarter where revenue leads, a quarter where repair leads. Life brochures rarely print the deload. Ed Diener and Martin Seligman reported in *Psychological Science* in 2002 that people in the top decile of happiness are disproportionately social: rich relationships show up as a distinguishing input, not a nice-to-have after prestige is secured. [diener-seligman](https://doi.org/10.1111/1467-9280.00415) Most inherited scoreboards under-weight that variable because it is hard to put on a CV.

#### Refactor Between Chapters

The life formula that fit at thirty-five will suffocate at fifty-five if you never refactor. Wharton Knowledge has framed career, family, and time as chapter problems: maximise within the season you are in, not across every season at once. [wharton](https://knowledge.wharton.upenn.edu/article/maximize-your-utility-career-family-and-time-strategies/) Philip Brickman and Donald T. Campbell coined hedonic relativism in a 1971 chapter: the want/is gap resets because humans normalise new levels of stimulation. [brickman-campbell](https://worlddatabaseofhappiness.eur.nl/publications/hedonic-relativism-and-planning-the-good-society-244/) Laurie A. Paul argued in *Transformative Experience* that major choices (parenthood, partnership, vocation) are epistemically special: you cannot know the self who will evaluate the outcome, so revelation value matters as much as expected utility. [paul](https://global.oup.com/academic/product/transformative-experience-9780198717957)

Oliver Burkeman's *Four Thousand Weeks* treats finite lifespan as the binding constraint productivity culture keeps pretending away. [burkeman](https://us.macmillan.com/books/9780374159122/fourthousandweeks) Richard Thaler and Cass Sunstein showed in *Nudge* that environment design shapes behaviour as much as willpower sermons. [nudge](https://yalebooks.yale.edu/book/9780300122237/nudge/) Martin Seligman's PERMA model (positive emotion, engagement, relationships, meaning, accomplishment) works better as a refactor checklist than a prescription: name the variables, then assign your weights. [perma](https://ppc.sas.upenn.edu/learn-more/perma-theory-well-being-and-perma-workshops) Refactoring is not betrayal of your earlier self. It is maintenance on the system that earlier self built with incomplete data.

#### I Optimised the Scoreboard Recruiters Could Audit

Inherited success metrics had stopped matching my felt life score by midsummer. Title, compensation, prestige: the variables recruiters and boards could audit on a single page. Unstructured time with people who knew me before the MBA was the variable I kept deleting from the model because it did not fit the template. The gratitude letter exercise was awkward in the way honest things are awkward when you already have a reputation to protect: you had to name a debt without turning it into networking. The kindness assignments produced data that was harder to spin than a cocktail reception. Classmates delivered flowers, bought strangers coffee, carried packages for people who would never know their title. [booth-faculty](https://www.chicagobooth.edu/faculty/directory/e/nicholas-epley)

A classmate I will call Rafael optimised the visible scoreboard aggressively: the board-visible promotion, the succession path everyone in the room could name, sleep sacrificed without comment. He treated friendships as noise he was not counting. I optimised alongside him and called it ambition. Same scoreboard, same sweat-versus-watch ratios Galloway had made legible, none of the weights I was supposed to be setting myself. Rafael got the line on his CV and the burnout in his body. I got clarity I did not want to pay for: that I had been treating Booth as a placement office for someone else's priority dials rather than a lab to redesign my own. The exercises had given us data on kindness and gratitude. I filed it under "soft skills" and returned to the scoreboard recruiters could audit. Rafael did not get the next decade back with corrected weights. He got the same Tuesdays I kept trading away, except he did not have a summer elective that treated kindness as data before the promotion clock ran out to tell him the trade was optional until the bill arrived.

#### The Strongest Case for Borrowing an Equation

Let me give the objection its fairest hearing, because borrowed algebra saves lives when bandwidth is scarce. Galloway distilled decades of pattern recognition into ratios a tired operator can remember on a commute. Epley distilled decades of experiments into exercises that change affect without claiming one set of priority dials fits every cohort. Laurie Santos's material is not vibes: a 2021 *PLOS ONE* evaluation of Yale's Science of Well-Being course found improved student well-being across multiple cohorts. [yaden-rct](https://doi.org/10.1371/journal.pone.0256760) Sonja Lyubomirsky, Kennon Sheldon, and David Schkade argued in *Review of General Psychology* in 2005 that intentional activities account for a substantial slice of happiness variance in their model, which is why "rewirements" are not astrology. [lyubomirsky](https://doi.org/10.1037/1089-2680.9.2.111) Robert Emmons and Michael McCullough's gratitude journaling experiments showed measurable affect gains over weeks, not just seminar applause. [emmons](https://doi.org/10.1037/0022-3514.84.2.377) If you are drowning, grabbing a proven heuristic is rational. A shared answer key beats reinventing nutrition from first principles while you are starving.

The limit is not that formulas exist. It is that life's biggest decisions are not small-world maximisation problems. Laurie A. Paul showed you cannot calculate utility for a self you have not yet become. Shahar Hechtlinger, Christin Schulze, Christina Leuker, and Ralph Hertwig argued in the *American Psychologist* in 2024 that transformative choices break expected-utility maximisation because selves change, experiential values are uncertain, and irreversibility makes the spreadsheet lie. [hechtlinger](https://doi.org/10.1037/amp0001439) Galloway's book is an anecdotal pattern library, not a finished recipe for your dials. Epley's exercises are empirical probes, not a finished life formula. Borrow both frames. Run the audits that survive your chapter. Then set your own weights and schedule the refactor when the season turns.

Borrowed algebra is a starting prescription, not your finished lens. Audit which variables you inherited from institutions that will not sit with you when you miss them. Rafael is in therapy and off the succession path he optimised for across two decades, because the equation he copied never included the Tuesday he needed and kept deleting.

---

### Meet the Harpers
[Read more](https://www.gaganmalik.io/en/newsroom/meet-the-harpers)

[Retrieval brief — meet-the-harpers]
Category: business
Thesis: Consulting rewards rainmakers who capture credit for design work they did not do. The Harper Effect is extraction dressed as transformation.
Answers: What does Gagan argue in "Meet the Harpers"? | What is Gagan Malik's view on Meet the Harpers? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

The partnership review was in a glass room overlooking Paddington station, for a heavily funded satellite internet startup I will not name in this essay. Overpriced espresso, a wall of displays still warm from the work I had shipped overnight: wireframes for sign-up, booking an installer, and the screen that tells you you are online. My rainmaker, Charlie, sat beside the client's CTO. The last funding round still glowed on the screen in nine figures. Charlie pointed at the journey map I had drawn and asked the CTO whether we could cut the steps and rebuild the flow on new cloud software before the runway chart turned red. Charlie said we had designed the answer that morning. He had not looked at me once. I nodded. The room heard efficiency. I heard my own work leaving under someone else's breath.

That nod was not surprise. It was a design choice. I had learned that conflict on a live account costs more than credit on a slide. I removed the friction the way I have removed friction elsewhere when the fee was large enough and the relationship fragile enough. *Two and a Half Men* mapped the geometry for laughs: Charlie lives on royalties and relationships, Alan pays the bills with expertise nobody prices correctly, Jake learns that satisfactory output is cheaper than excellence. The labour market files it under transformation. I have billed from inside all three costumes. Call it the Harper Effect once you have felt the nod land.

#### Your utilisation report measures the wrong relationship

Consulting firms do not pay for the best answer. They pay for the person who already holds the relationship when the answer lands. Partnership dashboards track relationship health, deal registration, and billable hours. They do not track who designed the journey that makes a transformation slide look credible. That is measuring credit assignment, not capability. The instrument is not broken. It is aimed at the wrong coordinate.

When a client promises AI will deliver a leaner operation, the translation is often specialist labour cut while the transformation narrative stays on the slide. The Paddington account is one instance, off the record. Investors and boards are not wrong to want efficiency. They are wrong to imagine it lands on the people who designed the journey or wrote the code underneath. The savings line is real. The distribution line is missing. That gap is where profit lives.

#### Deal registration is where value is captured

Org charts hide a quieter map. Wins are captured at deal registration, committee seats, and shared-vision slides. Value is built in design files and delivery rooms where nobody from the Global Executive Committee wants to stand. Charlie types hold the P&L keys: the joke in the hallway, the name on the steering slide, the nod that unlocks budget without anyone opening the journey map. Designer types map the flow, remove the friction, and invoice hours that vanish into someone else's utilisation target. Builder types write the code that keeps customers connected. The firm quotes the designer's rate to the client and attributes the win to the rainmaker. Margin expands. Nobody calls it extraction because the deck says digital transformation.

After we won a multi-million-pound satellite programme pitch, Charlie used the victory for a committee seat. My partnership track did not close loudly. I was quietly removed from the building. Marcus, a lead solution architect on the same account, was rewarded with implementation lead in a windowless room in Slough, patching legacy code at midnight while the savings narrative moved upstairs. I gained the knowledge that the system is extractive. Marcus did not gain autonomy over his own time. When AI arrives on the earnings slide, it is often his midnight work being described as an efficiency gain for someone who was not in the basement.

#### The savings get booked upstream. Vendors hold the slide.

Institutional Charlies do not need a Malibu beach house. They need a vendor partnership and a board narrative. When a large operator announces an AI-led transformation, savings get booked upstream in restructuring charges and margin guidance. Specialist labour gets liquidated downstream in headcount targets the market still applauds. The executive layer signs the programme. The vendor layer sells the operating model. The designer and engineer become the line items the slide replaces.

Titles help here the way they always help when proof would be inconvenient. A SVP or managing director on the slide excuses absence of the journey map. A partnership director on the account excuses absence of design authorship. Immunity is not diplomatic. It is operational: a ready-made exit from the moment someone asks who actually kept the service running. Charlie did not need to understand the flow. He needed my nod and a flight to Verbier once the signature landed. Large operators do not need every redundant role named on the org chart. They need applause for efficiency while the people who designed and built the thing being automated are steered out of the credit line.

#### The domestique burns their legs so the leader gets the jersey

In professional cycling, a domestique rides in the wind so the team leader can sprint. They fetch water, close gaps, and sacrifice their own standings so someone else's name reaches the podium. The leader gets the yellow jersey. The domestique gets sore quads and a line on the team sheet most fans never read.

You feel this before you can explain it. Your lungs burn while someone else's quote is printed. Your thighs shake on the climb you paced for another rider. You know the effort was real even when the credit was allocated elsewhere. Contribution and recognition are different currencies. Grab for one mid-race and you often lose both.

#### I chose the nod because the fee was larger than the fight

One spring afternoon in 2021 I handed Charlie a design-led digital transformation framework without securing lead partner designation on the account. I convinced myself that being the brain was enough. In consulting, the relationship holder holds the P&L keys. I traded intellectual property for approval on the next phase. That is the same trade I made in the Paddington boardroom when I stayed silent rather than correct a client-facing claim I knew was wrong. Conflict threatened the account. The account was paying.

I am not telling this as a victim story. I priced the nod. I knew what silence bought: continued billing, a seat at the next workshop, no awkward partner conversation about who actually designed the journey. Marcus did not get that choice. He got the basement, the midnight patches, and a utilisation target that treated his excellence as a cost centre. That household runs on the bargain. So does a transformation programme when savings are announced before autonomy is offered to the people who built the thing being automated.

#### The comfortable case for giving without counting

Adam Grant argues in *Give and Take* (2013) that generous contributors who help without immediate return often build the highest-trust networks and end up among the most successful. Givers, matchers, and takers populate every firm. Grant's evidence suggests the most effective givers are strategic rather than self-sacrificing: they contribute widely but protect their time and credit when it matters. Read generously, the case is comforting for builders: keep delivering excellence, keep handing over the hard intellectual work, and reputation will eventually convert into power. It requires believing that consulting is a high-trust marketplace where social capital compounds fairly and transformation programmes reward the people who made them possible.

That paragraph deserves its full strength because many builders want it to be true. It would mean the nod in Paddington was an investment, not a theft. It would mean Marcus's midnight patches were compounding into partnership rather than subsidising someone else's committee seat. The Trades Union Congress report on algorithmic management, published in May 2026, finds that digital transformation paths tend to deskill specialist labour and centralise decision-making power in fewer hands. [tuc](https://www.tuc.org.uk/research-analysis/reports/algorithmic-management-2026) Robert Greene's *The 48 Laws of Power* (1998), Law 7, states the quieter rule: get others to do the work, then take the credit. Reputation compounds for extractors when the org chart is designed to capture wins upstream. Transformation compounds savings for vendors and executives when specialist labour is priced as replaceable. Charlie did not need a high-trust network. He needed my winning slide and my silence while he booked the flight.

The gap between the person who builds value and the person who captures it is where profit lives in this industry.

Over-delivering for systems that measure relationship credit instead of network survival is not being a giver. It is being farmed.

Marcus is still in Slough at midnight, patching the code that keeps customers online, while the person who took the credit plans the next transformation off his labour.

---

### People Love to Buy. They Just Hate Being Sold.
[Read more](https://www.gaganmalik.io/en/newsroom/why-people-love-to-buy)

[Retrieval brief — why-people-love-to-buy]
Category: business
Thesis: Buying is agency. Being sold is capture. Commission breath, feature lists, and why economic value lets the buyer author the yes.
Answers: What does Gagan argue in "People Love to Buy. They Just Hate Being Sold"? | What is Gagan Malik's view on People Love to Buy. They Just Hate Being Sold? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

I was three slides into a valuation bridge on Zoom when Martine's camera went dark. Not a dropped connection. A choice. She stayed on the call just long enough to say she would "take it internal" and then she was gone. I had opened with "quick catch-up" and arrived with an investor deck. She had opened with what her early-stage fund was actually underwriting this year and received a product tour of my company.

Jeffrey Gitomer put the asymmetry in one line in *Jeffrey Gitomer's Little Red Book of Selling* (Bard Press, 2004): people do not like to be sold, but they love to buy. [gitomer](https://www.goodreads.com/book/show/75890.The_Little_Red_Book_of_Selling) The line is not cynicism about commerce. It is a diagnosis about control. The investor deck and the cold-call script fail for the same reason: both reach for the wheel before the buyer has named the problem. You have felt the same wall rise when an account executive cold-calls, opens with "just catching up," and by the second line of their written script is pitching a product you never put on a shortlist. Your wallet did not close because you hate spending. It closed because someone else reached for the steering wheel.

#### Why "Just Checking In" Sounds Like a Trap

Sales culture trained a generation to hear intent in every courtesy. Pipeline stages reward talk. Quota quarters reward velocity. CRM fields fill when someone speaks, not when someone listens. The buyer is not foolish for flinching when the phone rings from a number they do not recognise, or when a nurture email arrives with a subject line about "checking in" and a body that pivots to a demo link by paragraph two. Jack W. Brehm argued in *A Theory of Psychological Reactance* (Academic Press, 1966) that people hold perceived freedoms and fight to restore them when influence threatens those freedoms. [brehm](https://link.springer.com/chapter/10.1007/978-1-4613-8309-3_1) Mona Clee and Robert Wicklund, writing in the *Journal of Consumer Research* in March 1980, showed the same motivational state in purchase settings: pressure that feels like elimination of choice produces resistance, not gratitude. [jcr-clee](https://doi.org/10.1086/208673)

That is why commission breath is detectable at sentence two on a cold call, and in the fourth line of a sales email that names your job title but not your problem. The buyer is not parsing your integration count. They are calculating whether saying yes tomorrow will cost them dignity today. Sales floors call this "objection handling." The buyer calls it being cornered. Both can be true in the same minute, which is why polite calls end with "send me a one-pager" and nothing moves.

#### Buyers Pay for Holes, Not Drills

You have happily bought things: the course you researched at midnight, the tool your peer recommended, the hire you championed because the spreadsheet finally made sense. Buying, when it works, is an act of agency. You name the problem. You compare options. You tell a story about yourself as competent, relieved, or finally ahead of a backlog. Theodore Levitt argued in "Marketing Myopia" in the *Harvard Business Review* in 1960 that industries decay when they fixate on the product they ship instead of the need the customer is hiring it to solve. [levitt](https://hbr.org/1960/07/marketing-myopia) The drill-versus-hole formulation is often attributed to his teaching: people do not want a quarter-inch drill; they want a quarter-inch hole.

James C. Anderson and James A. Narus, writing in the *Harvard Business Review* in November 1998, pushed the same logic into economics: suppliers who cannot quantify what their offering is worth in the customer's P&L will lose to whoever can, because procurement will always compress you to acquisition price unless you can prove value delivered. [anderson-narus](https://hbr.org/1998/11/business-marketing-understand-what-customers-value) Value sounds like "your exception queue costs forty analyst hours a week"; a feature list sounds like "AI-powered workflow automation." One invites the buyer to check your maths. The other invites them to check your motive.

#### Your Homepage Still Sells the Drill

Most go-to-market copy still sells the drill. Landing pages lead with "AI-powered workflow automation" and a logo wall. Lifecycle emails celebrate feature releases: dark mode shipped, a new dashboard widget, another integration badge. Neil Rackham's *SPIN Selling* (McGraw-Hill, 1988) documented what skilled sellers do differently in large business sales after Huthwaite researchers analysed tens of thousands of calls: they ask about situation and problem, draw out implication, and let the buyer voice need-payoff before the demo arrives. [rackham](https://books.google.com/books/about/SPIN_Selling.html?id=0nM1DgAAQBAJ)

The same question that feels like care from a doctor feels like capture from a vendor whose quarter ends on Friday. Buying stays pleasurable when the buyer can narrate the decision as theirs. Being sold a feature list turns pleasure into proof that someone else planned your yes.

#### Your Knee Locks Before Your Mind Agrees

You have felt this in physiotherapy, even if you never named it. The clinician asks you to bend a knee that hurts. You agree in principle. Then their hand pushes before your muscles believe you are safe, and your quadriceps lock. Clinicians call it guarding. The joint is not refusing healing. It is refusing foreign force. Your mind has not finished consenting. Your body has already voted.

That somatic vote is what the spreadsheet misses. Reactance is not only an intellectual dislike of sales technique. It is the bracing that happens when autonomy feels grabbed. Pitch before diagnosis and you are pushing on a guarded knee. The buyer's "I need to think about it" is often not arithmetic. It is musculature.

#### I Pitched the Deck. She Passed.

Martine was a partner at an early-stage venture fund. I was the founder trying to sell my company into their portfolio, or at least into a conversation that could end in acquisition. The Zoom invite said thirty minutes. I should have spent the first fifteen on what her partnership had been underwriting lately: vertical software with exportable IP, cheques in a band I could actually accept. Instead I shared screen, walked net revenue retention, and advanced three exit structures because the deck already had three exit structures. She was polite in the way investors are polite when they have learned not to reward heat with honesty. I sent a follow-up blurb the same afternoon that sat in her inbox like a receipt for a meeting that never happened.

My co-founder asked whether we were still in the running. I said we had "good engagement," which is founder dialect for nobody said no loudly enough to hurt my feelings. She did not buy the company. I got a line for the investor update about "ongoing strategic conversations" and a lesson I did not deserve to learn for free. She got forty minutes on Zoom she will not get back and another story about founders who hear "thesis" and answer "TAM." I have since retired that deck template. I am telling you this because the template still lives on a thousand shared drives. I did not think about her once until I sat down to write this sentence. That is not neutrality. It is a cost I exported.

#### The Strongest Case for Leading the Buyer

Give the fairest version of the objection its full hearing. Buyers are busy. Information asymmetry is real. A good seller transfers belief about risk, implementation, and fit that no comparison page can carry. Complex business purchases fail without someone willing to challenge vague requirements and name trade-offs. Rackham's research programme, reported in *SPIN Selling* (1988), found that in larger sales the highest performers talked less about features and more about consequences the buyer had not yet voiced. [rackham](https://books.google.com/books/about/SPIN_Selling.html?id=0nM1DgAAQBAJ) Anderson and Narus, in the November 1998 *Harvard Business Review*, argued that suppliers who build customer value models can demonstrate economic worth instead of arguing about list price. [anderson-narus](https://hbr.org/1998/11/business-marketing-understand-what-customers-value) If experts never lead, mediocre products win on noise and incumbents win on inertia. Closing is not moral filth. It is sometimes the difference between a right fit shipped and a problem left to rot because everyone was too polite to decide.

The gap is not whether guidance happens. It is whether the buyer experiences guidance as service or capture. Friestad and Wright, in the June 1994 *Journal of Consumer Research*, argued that consumers actively interpret persuasion attempts using persuasion knowledge developed across a lifetime of marketplace exposure. [jcr-friestad](https://doi.org/10.1086/209380) Clee and Wicklund, in March 1980 in the same journal, showed that reactance in consumer settings rises when people perceive their freedom to choose is under threat. [jcr-clee](https://doi.org/10.1086/208673) Leading works when the buyer writes the prescription in their own words and the seller becomes pharmacist, not author. Leading fails when every need-payoff question arrives on a visible rail toward your end-of-quarter slide, or when your homepage still opens with what you built instead of what it is worth.

People do not resist spending: they resist losing authorship over the spend. Your next polite "send me a one-pager," or the nurture email gathering dust in your inbox, is often a locked knee, not a budget problem. Martine is still underwriting on thesis somewhere, and the next founder who opens with what her fund is trying to buy instead of their revenue chart will sound, to her, like the first adult in months who is not pushing on a locked joint.

---

### Infinite Scroll Is a Cognitive Weapon of Mass Destruction.
[Read more](https://www.gaganmalik.io/en/newsroom/why-infinite-scroll-is-a-cognitive-wmd)

[Retrieval brief — why-infinite-scroll-is-a-cognitive-wmd]
Category: design
Thesis: Infinite scroll deleted the page end. Aza Raskin regrets inventing it. Your thumb still runs a slot machine schedule at global scale.
Answers: What does Gagan argue in "Infinite Scroll Is a Cognitive Weapon of Mass Destruction"? | What is Gagan Malik's view on Infinite Scroll Is a Cognitive Weapon of Mass Destruction? | How does this essay relate to product and design leadership?
Concepts: design, Gagan Malik, writings

By Gagan Malik

You reach the bottom of something that has no bottom. Your thumb lifts. New tiles load. The feed continues as if the page never ended. You do not feel a decision point. You feel momentum. Infinite scroll is a cognitive weapon of mass destruction, not because the label is dramatic, but because one UI pattern deletes the last cheap ceremony that let your slow brain catch your fast thumb. I know how it is drawn. I have shipped interface pages where the timeline never showed a footer, where discovery mode meant no hard stop, where the wireframe taught users that another screen always waited below the fold. In 2006 Aza Raskin built it at Humanized to remove pagination friction. He now says he feels guilty about what markets did with it. [bbc](https://www.bbc.co.uk/news/technology-44640959) I argued in [*Attention Is the Most Consumed Drug on the Planet*](https://www.gaganmalik.io/en/newsroom/why-attention-is-the-most-consumed-drug) that focus is extracted legally at hourly doses. In [*The Algorithm Sold Your Next Four Seconds Before You Scrolled*](https://www.gaganmalik.io/en/newsroom/the-algorithm-of-the-attention-economy) I named the machinery underneath: predict, rank, auction, log, retrain. This piece names the interface trick that runs before any of that: the endless bottom.

#### The Page Ended. Someone Deleted the Ceremony.

Every "next page" click was a micro-negotiation: continue or close the tab. Pagination was never only inconvenience. It was a stopping cue. Chapter ends, advert breaks, the blank margin at the foot of a newspaper: each gave your veto system a beat to arrive. Product teams hated those beats because time-on-site is the metric that buys the next funding round. Tristan Harris, co-founder of the Center for Humane Technology, told the U.S. Senate Commerce Committee on 25 June 2019 that platforms run a race to the bottom of the brain stem, deploying persuasion architecture including infinite scroll because the advertising model always demands more attention than yesterday. [commerce.senate](https://www.commerce.senate.gov/services/files/96e3a739-dc8d-45f1-87d7-ec70a368371d) I have sat in growth rooms where retention charts outranked every other argument, and I have handed off wireframes that removed the last visible stop. The quiet part was said aloud: if we show users where the good content ends, they leave.

Aza Raskin invented infinite scroll in 2006 at Humanized to make long lists feel seamless. [bbc](https://www.bbc.co.uk/news/technology-44640959) The BBC quoted him in July 2018 comparing addictive interfaces to "behavioural cocaine" sprinkled across the screen. He testified in a New Mexico Meta trial that he regrets a feature that traps users in endless scrolling. [kob](https://www.kob.com/new-mexico/inventor-of-infinite-scroll-says-he-regrets-it-at-new-mexico-meta-trial/) In a Center for Humane Technology talk he estimated, conservatively, that infinite scroll wastes roughly 200,000 human lifetimes per day beyond what pagination would have consumed. [centerforhumanetechnology](https://assets-global.website-files.com/5f0e1294f002b1bb26e1f304/642f41e223acbcf73d0eadf6_spotlight-your-undivided-attention-three-rules-of-humane-tech-transcript.pdf) That number is his modelling, not a peer-reviewed measurement. One JavaScript pattern, deployed everywhere, still deletes stop signals at planetary scale.

#### Your Thumb Moves on a Slot Machine Schedule

The mechanism is older than Silicon Valley. B. F. Skinner showed in the 1950s that variable-ratio schedules produce more persistent behaviour than predictable rewards: the subject keeps pulling because the next payoff might be one pull away. [britannica](https://www.britannica.com/science/operant-conditioning) Slot machines borrowed the schedule. Infinite scroll perfected it for a pocket lever you already carry. The reward is not always good. It is sometimes outrageous, sometimes tender, sometimes a sale on trainers you glanced at once. Unpredictability is the point.

Raskin told the BBC that if you do not give your brain time to catch up with your impulses, you just keep scrolling. [bbc](https://www.bbc.co.uk/news/technology-44640959) Gloria Mark, a psychologist at the University of California, Irvine, has tracked attention on screens for two decades. On the American Psychological Association's *Speaking of Psychology* podcast she reported averages of about two and a half minutes in 2004, 75 seconds around 2012, and roughly 47 seconds in recent years. [apa](https://www.apa.org/news/podcasts/speaking-of-psychology/attention-spans) I am not claiming infinite scroll alone caused that collapse. I am claiming it removed one of the few friction points that used to force a breath between doses. In [*When a Click Costs Nothing*](https://www.gaganmalik.io/en/newsroom/when-a-click-costs-nothing) I wrote that friction is the editor; on a bottomless feed, product teams keep removing pauses.

#### Five Billion Feeds Never Run Out of Bottom

Scale turns a UX convenience into an ambient condition. GWI research in DataReportal's Digital 2025 report puts 5.24 billion social media user identities online and the global average adult at six hours and thirty-eight minutes of daily internet time. [datareportal](https://datareportal.com/reports/digital-2025-global-overview-report) Those are identities, not unique humans, but the thumb math still horrifies: bottomless feeds in pockets that never require compliance, consent renegotiation, or a visible last page. Frances Haugen, a former Facebook product manager who leaked internal documents, told CBS News's *60 Minutes* on 3 October 2021 that engagement-based ranking amplified hateful and divisive content because anger is easier to provoke than calm, and that the company knew safer algorithms would reduce time on site. [cbsnews](https://www.cbsnews.com/news/facebook-whistleblower-frances-haugen-misinformation-public-60-minutes-2021-10-03/) The Wall Street Journal's Facebook Files series, published in September and October 2021 from Haugen's documents, reported similar findings on divisive posts retaining users longer. [wsj](https://www.wsj.com/articles/facebook-algorithm-change-zuckerberg-11631781515) Infinite scroll removes when to stop; engagement ranking chooses what loads next. Neither feature alone explains every lost hour, but together they form one retention stack. That is an inference chain, not a single causal proof about infinite scroll alone. It is still enough to say democratic discourse fractures when retention architecture rewards what keeps the thumb moving.

#### He Wrote the Bottom. He Still Left the Table.

In 2020 I consulted on a crypto newsfeed app whose timeline loaded forever. Retention looked strong in the usability lab. The founders called the scroll frictionless — nobody stopped. I invoiced and moved on. That is the smaller version of the irony.

Aza Raskin wrote the JavaScript in 2006. [bbc](https://www.bbc.co.uk/news/technology-44640959) He has testified in New Mexico's Meta trial that he regrets a feature built to trap users in endless scrolling. [kob](https://www.kob.com/new-mexico/inventor-of-infinite-scroll-says-he-regrets-it-at-new-mexico-meta-trial/) On NPR in March 2026 he told the jury something harder to dismiss: he knows exactly how infinite scroll deletes stopping cues, and he was still susceptible. He found himself leaving the dinner table for the bathroom to scroll. He wrote software to add friction to his own browser because understanding the mechanism did not grant immunity. [npr](https://www.npr.org/2026/03/28/nx-s1-5762945/how-will-two-landmark-social-media-verdicts-reshape-how-we-use-this-technology) He told the jury it was not a fair fight. If the inventor still needs a patch to escape his own pattern, discipline was never the missing ingredient.

#### Your Thumb Keeps Hunting for a Landing.

You have felt this in your hand. Your thumb reaches the place where a page should end. The feed loads six more tiles before your veto catches up. Your nervous system still expects a landing the way your foot expects a stair to finish. Pagination was that landing: a footer, a page break, a batch that admitted it had ended. Infinite scroll deletes the landing and lets momentum govern instead of choice.

That is why one UI pattern scales like a weapon. You are not deciding tile by tile. You are falling through a floor that rebuilds itself beneath your thumb. Engineers call it seamless. Occupants call it a night gone. The quarterly report calls it success.

#### The Strongest Objection, and Where It Runs Out of Road

Give the objection its full weight. Social platforms started as friend graphs and photo albums. Billions scroll without visible harm. Parental controls and daily limits ship in every major app. Nobody holds your thumb on the glass. Some students swap smartphones for keypad phones before exams; the exit exists if your social world allows it. Most cannot. Johann Hari, in *Stolen Focus* (Bloomsbury, 2022), treats the attention crisis as environmental: twelve forces intersect, and shaming individuals misses the architecture. [bloomsbury](https://www.bloomsbury.com/uk/stolen-focus-9781526620224/) I argued the training-load version in [*Gen Z Didn't Get Dumber. They Got Trained Without Rest Days*](https://www.gaganmalik.io/en/newsroom/why-gen-z-cant-hold-focus). Infinite scroll is one lever among many. I still open voluntarily. Naming one UI pattern a weapon still risks sounding theatrical.

The gap is feedback, not intent. Frances Haugen told CBS *60 Minutes* in 2021 that Facebook's research showed angry content spread faster, and that safer feeds would mean less time on site and fewer ad clicks. [cbsnews](https://www.cbsnews.com/news/facebook-whistleblower-frances-haugen-misinformation-public-60-minutes-2021-10-03/) Raskin, in the same Center for Humane Technology talk where he gave the lifetime estimate, described a coordination tragedy: if one company removes infinite scroll and rivals keep it, the restrained company loses. [centerforhumanetechnology](https://assets-global.website-files.com/5f0e1294f002b1bb26e1f304/642f41e223acbcf73d0eadf6_spotlight-your-undivided-attention-three-rules-of-humane-tech-transcript.pdf) Haugen showed the landlord knew which wire heated the floor.

#### The Phantom Bottom Has No Invoice

When I forget the mechanism, I call it a discipline problem and the scroll runs hotter. Refusing the shame story Hari warned against is not the same as pretending willpower fixes a deleted stop signal.

Infinite scroll replaced stopping cues with a variable-ratio schedule your thumb can run faster than your veto. Calling that a private failing misnames the architecture and lets engagement metrics keep deleting rest beats. Raskin invented the pattern and still had to write software to slow his own scroll at dinner. Somewhere else, someone is still scrolling past the hour they meant to stop.

---

### The Algorithm Sold Your Next Four Seconds Before You Scrolled.
[Read more](https://www.gaganmalik.io/en/newsroom/the-algorithm-of-the-attention-economy)

[Retrieval brief — the-algorithm-of-the-attention-economy]
Category: business
Thesis: The feed guesses, ranks, sells ad slots, logs your scroll, and updates itself. Your focus is the product.
Answers: What does Gagan argue in "The Algorithm Sold Your Next Four Seconds Before You Scrolled"? | What is Gagan Malik's view on The Algorithm Sold Your Next Four Seconds Before You Scrolled? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

You are three videos into a feed you did not open for this purpose. Your thumb keeps moving. An ad for trainers you glanced at once last Tuesday slides into the fourth slot, timed for the half-second your attention wobbles. No gavel. No ledger. Nothing feels like a transaction. That is the design. The attention economy is not a metaphor for this: it is a machine that guesses what keeps you scrolling, ranks what it shows you next, sells the space around your eyes, logs what you did, and updates itself for tomorrow. Johann Hari, in *Stolen Focus*, calls the problem environmental, not a moral failure. [bloomsbury](https://www.bloomsbury.com/uk/stolen-focus-9781526620224/) I agree. In [*Attention Is the Most Consumed Drug on the Planet*](https://www.gaganmalik.io/en/newsroom/why-attention-is-the-most-consumed-drug) I argued that attention is extracted at hourly doses; the average adult now spends six hours and thirty-eight minutes online per day. [datareportal](https://datareportal.com/reports/digital-2025-global-overview-report) This piece names the machinery under that dose.

#### The Feed Guesses What Keeps Your Thumb Moving

Shoshana Zuboff calls it surveillance capitalism: platforms watch what you click, pause, and skip, then sell predictions about what you will do next. [publicaffairsbooks](https://www.publicaffairsbooks.com/titles/shoshana-zuboff/the-age-of-surveillance-capitalism/9781541758001/) Tristan Harris told the U.S. Senate in 2019 that apps run a race to the bottom of the brain stem. [commerce.senate](https://www.commerce.senate.gov/services/files/96e3a739-dc8d-45f1-87d7-ec70a368371d) I have sat in growth rooms where retention charts outranked every other argument, and I have helped brands buy the inventory I am describing. Engineering and money point the same way.

Inside Meta, engineers told MIT Technology Review in 2021 that they kill feed versions that reduce likes, comments, or shares. [technologyreview](https://www.technologyreview.com/2021/10/05/1036519/facebook-whistleblower-frances-haugen-algorithms/) Every swipe and skip goes back into the training pile. The Wall Street Journal's Facebook Files showed the company knew angry posts kept people scrolling and kept the ranking that way. [wsj](https://www.wsj.com/articles/facebook-algorithm-change-zuckerberg-11631781515) The feed is not asking what is good for you. It is asking what keeps you past the point you meant to leave.

#### Your Thumb Moves Before You Decide to Stop

When I lose a fight with my own thumb, I am slow, not broken. The reflex fires before the veto. I called it slot-machine logic in [*Attention Is the Most Consumed Drug on the Planet*](https://www.gaganmalik.io/en/newsroom/why-attention-is-the-most-consumed-drug): you cannot tell which swipe will hook you until you are already moving. [britannica](https://www.britannica.com/science/operant-conditioning) In [*When a Click Costs Nothing*](https://www.gaganmalik.io/en/newsroom/when-a-click-costs-nothing) I wrote that friction is the editor; on a feed, product teams keep removing pauses.

That is persuasion at scale: known psychological tricks, not a cartoon villain. I have written about [contempt in interfaces](https://www.gaganmalik.io/en/newsroom/why-we-designed-contempt-into-our-interfaces), [outrage the feed rewards](https://www.gaganmalik.io/en/newsroom/why-do-algorithms-ignore-the-many-and-worship-the-few), [a generation trained without rest days](https://www.gaganmalik.io/en/newsroom/why-gen-z-cant-hold-focus). The loop learns from your reflexes while you sleep.

#### Every Slot on Your Screen Is Auctioned in Milliseconds

Guessing what you watch is half the business. Selling the space around it is the other half. On the open web, advertisers bid for your eyeballs in milliseconds; you never see the auction. [adexchanger](https://www.adexchanger.com/adexplainer/the-programmatic-auction-is-changing-in-real-time-heres-how/) Publishers run parallel bids in the page header to squeeze more from the same patch of glass. [amazon](https://advertising.amazon.com/library/guides/header-bidding) Inside YouTube or Instagram the stack differs; the incentive does not. Your time on site is inventory whether the buyer is a demand-side platform or the platform itself.

Worldwide ad spend crossed $1 trillion (about £790 billion) in 2025, with more than 75% digital. [emarketer](https://www.emarketer.com/content/worldwide-ad-spending-forecast-2025) By the time you lift your thumb, the system has logged your session and queued tomorrow's guess.

#### Aarush Had Mocks to Protect. The Feed Called It a Break.

Aarush was preparing for JEE Main at home in Chandigarh when his mock marks began to slip. [indianexpress-aarush](https://indianexpress.com/article/education/jee-main-i-quit-phone-and-aced-the-exams-says-chandigarh-boy-at-air-8-10647057/) He finished a practice paper, took a short break, and opened YouTube for Shorts. One became many. He thought he was resting. His mother Malti could see the drift before the scorecard did. The unsolved questions stayed on the desk.

I have watched the same pattern in my family. Aarush quit his phone before the mocks buried him and later scored AIR 8 in JEE Main 2026; most do not get that exit. The hardest exam in the country sits on the calendar while the softest content sits one thumb away.

#### Your Mouth Was Busy. You Are Still Empty.

You have had this afternoon. Crisps on the desk. Coffee going cold. A video in a tab you did not mean to open. Your jaw keeps working. Your stomach still feels hollow. Something was logged as intake. Nothing reached the part of you that needed to feel full.

Attention feeds work like that. The sensor is engagement: time on site, daily opens, ad views per session. Pull-to-refresh, autoplay, the notification that drags you back: each fires because a metric dipped. The harm shows up elsewhere: Aarush's slipping mocks, Malti at his desk, Gloria Mark's finding that office workers could hold one screen task for about two and a half minutes in 2004 and forty-seven seconds in recent years. [uci](https://www.universityofcalifornia.edu/news/cant-pay-attention-youre-not-alone) The quarterly report still looks fine. That is the point.

#### The Strongest Objection, and Where It Runs Out of Road

Hold the opposition at full strength before narrowing. These feeds began as tools for staying in touch. Most sessions end without catastrophe. Screen limits exist. No one forces the icon onto your home screen. Some JEE toppers quit smartphones entirely before high-stakes exams; exit ramps are real if you can afford the social price. Most cannot. Hari is right that willpower alone is a thin shield against environmental design. [bloomsbury](https://www.bloomsbury.com/uk/stolen-focus-9781526620224/) I open the apps myself. Blaming the ranking layer for every lost evening can sound like an excuse for weak will.

The gap is feedback, not intent. Frances Haugen told CBS *60 Minutes* in 2021 that Facebook's research showed angry content spread faster, and that safer feeds would mean less time on site and fewer ad clicks. [cbsnews](https://www.cbsnews.com/news/facebook-whistleblower-frances-haugen-misinformation-public-60-minutes-2021-10-03/) She said the same under oath to the Senate days later. [commerce.senate](https://www.commerce.senate.gov/services/files/FC8A558E-824E-4914-BEDB-3A7B1190BD49.pdf) Haugen showed the landlord knew which wire heated the floor.

#### The Spreadsheet Is Not the Truth

When I forget the machinery, I call it a discipline problem and the loop runs hotter. Asking who gets paid for the next four seconds does not fix the rigged room. It refuses the shame story Hari warned against. The machine from the opening is still running.

Somewhere in India, a mock test sits unattempted. Somewhere in Ohio or Osaka, an essay due at midnight is still on page one while a thumb hovers over the next reel.

---

### v1.21.21
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-21-21)

[Retrieval brief — release-1-21-21]
Category: technology
Thesis: AI-readable site files, a public knowledge API, faster pages, fixed link previews, new essay, iOS install hint.
Answers: What changed in gaganmalik.io v1.21.21? | What's new on Gagan Malik's website in v1.21.21? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

AI-readable site files, a public knowledge API, faster pages, fixed link previews, new essay, iOS install hint.

---

### Attention Is the Most Consumed Drug on the Planet.
[Read more](https://www.gaganmalik.io/en/newsroom/why-attention-is-the-most-consumed-drug)

[Retrieval brief — why-attention-is-the-most-consumed-drug]
Category: culture
Thesis: Illicit drugs kill faster. Attention kills slower. Both have dealers, and the legal market extracts more daily doses than any cartel.
Answers: What does Gagan argue in "Attention Is the Most Consumed Drug on the Planet"? | What is Gagan Malik's view on Attention Is the Most Consumed Drug on the Planet?
Concepts: culture, Gagan Malik, writings

By Gagan Malik

I opened Instagram to answer a client DM. Forty minutes later I was watching a stoic philosophy clip about discipline, then a venture capitalist ranking morning routines as "compounding assets," then a finance creator explaining why my net worth was apparently one index fund away from freedom. I cannot name the moment the task changed. I felt the familiar low-grade shame. I will do it again tomorrow. Here is the claim nobody wants to say out loud in a board meeting: attention is the most consumed drug on the planet, and the comparison to fentanyl is provocative for a reason that has nothing to do with moral equivalence. Recreational drugs are measured in milligrams and bought in alleyways. Attention is measured in hours and sold by companies that file quarterly earnings. One kills fast. The other kills slow. Both have dealers. Only one group gets invited to Davos.

#### The Dose Is Measured in Hours, Not Milligrams

The global average adult internet user now spends six hours and thirty-eight minutes online each day, according to GWI research published in DataReportal's Digital 2025 report. [datareportal](https://datareportal.com/reports/digital-2025-global-overview-report) That is not a hobby. That is a daily dose taken by 5.24 billion active social media user identities before breakfast, on the toilet, during meetings, and at 11pm when your body asked for sleep and your thumb answered something else. [datareportal](https://datareportal.com/reports/digital-2025-global-overview-report)

Fentanyl, heroin, cocaine: each has a user base measured in millions and a supply chain that governments treat as a civilisational emergency. Attention has a user base measured in billions and a supply chain that governments treat as GDP. Meta, Google, ByteDance, Amazon, Netflix, and every ad network downstream do not sell shoes or videos. They sell the capture and resale of human focus. Advertisers are the wholesalers. You are the crop. The product team that shaved four seconds off your rest period between doses got a promotion. That is not a metaphor. That is the incentive structure.

#### The Cartel That Files Quarterly Earnings

Here is the scale comparison that makes people uncomfortable, which is why you should sit with it instead of dismissing it. Worldwide total media ad spending crossed $1 trillion for the first time in 2025, with digital advertising accounting for more than 75% of that total, per eMarketer's worldwide forecast published in January 2025. [emarketer](https://www.emarketer.com/content/worldwide-ad-spending-forecast-2025) The United Nations Office on Drugs and Crime, in its World Drug Report 2025 press release, estimates the illicit drug trade generates hundreds of billions of dollars per year, with wide error bars and definitional fights about retail versus wholesale value. [unodc](https://www.unodc.org/documents/data-and-analysis/WDR_2025/Press_release_WDR_2025_English.pdf) I am not claiming ad spend equals cartel revenue to the penny. I am claiming that the legal attention economy moves more money, through more channels, with more daily transactions, than any recreational drug market on Earth, and nobody raids the server room.

The other difference is social permission. A parent who finds heroin in a teenager's drawer calls an emergency. A parent who finds six hours of TikTok in Screen Time calls it "generational." Both are dependency signals. Only one comes pre-installed, subsidised by your mobile contract, and defended by investors who call withdrawal "churn."

#### Variable Reward Is Still a Dealer's Trick

I need to be careful here, because neuroscience Twitter has ruined careful. Intermittent reinforcement is not new. Skinner demonstrated in the 1950s that unpredictable rewards produce more persistent behaviour than predictable ones, a finding now standard in operant conditioning literature. [britannica](https://www.britannica.com/science/operant-conditioning) Slot machines borrowed the mechanic. Social feeds perfected it. The infinite scroll is not a design accident. It is a variable-ratio schedule dressed in UX copy about "connection."

That is an inference built on established behavioural science, not a claim that your phone "hijacks dopamine" like a cartoon villain. I have sat in rooms where product managers said the quiet part aloud: if we tell people when the good content ends, they leave. If we never tell them, they stay. I did not object loudly enough. Neither did you, if your OKRs ever included time-on-site. The dealer does not need a lab coat. He needs a retention chart and a quarterly business review.

#### I Hit the Retention Target. She Missed Sleep.

Last autumn I was consulting on a consumer finance app that needed to "re-engage lapsed users," which is product language for get the thumb back. We A/B tested notification copy. One variant beat the control by eleven percent on seven-day return. The client was delighted. I was delighted. I sent the deck, collected the invoice, and moved to the next sprint.

Priya, a junior product analyst at the Indian IT services integrator on the workstream, was less delighted. She was twenty-six, from Pune, two years into her first role there, and she ran the overnight cohort analysis because London and New York share an ocean but not a timezone with Pune. At 2:04am she Slack messaged me a screenshot: the winning push notification used loss aversion ("Your offer expires") rather than utility ("Your balance updated"). I replied with a thumbs-up emoji and went back to sleep. She did not.

I hit the retention target. She missed sleep. The user who tapped at 6:47am because an engineered pang of anxiety outranked their alarm clock does not appear in any dashboard I presented upstairs. I kept the SOW. I did not think about Priya until I sat down to write this sentence. That is what complicity looks like when the drug is legal. You call it growth.

#### Rest Periods Get Shorter Until Something Snaps

Think about strength training, not philosophy. You can cut the rest between sets and feel productive for weeks. More reps. More volume. The mirror still looks fine. Then one Tuesday you fail a warm-up weight that should be trivial, and the failure is not mysterious. You accumulated fatigue debt while celebrating output.

Attention products work the same way. Every "just one more" feature is a rest period removed: autoplay, pull-to-refresh, the badge that reappears the second you clear it. You are not getting stronger. You are removing recovery between doses while the platform counts volume. The injury shows up elsewhere, in sleep, in focus, in the junior who cannot finish a thought without reaching for the slot machine in her pocket. The platform's quarterly report will still look fine. That is the problem.

#### The Strongest Objection, and Where It Runs Out of Road

Let me hold the objection at full strength, because it deserves it and because getting this wrong would be obscene. Fentanyl and heroin kill people. They leave parents identifying bodies. They destroy neighbourhoods. To compare that catastrophe to scrolling Instagram is to insult every bereaved family and to flatten a mortality curve that America's National Center for Health Statistics tracked at 79,384 drug overdose deaths in 2024, with synthetic opioids other than methadone involved in 47,735 of those deaths. [cdc](https://www.cdc.gov/nchs/products/databriefs/db549.htm) That is not a debating point. That is a morgue.

The objection is correct about acute harm. It is wrong about scale and supply. No cartel on Earth doses five billion people before lunch. No heroin dealer gets applauded on earnings calls for increasing "daily active users." And the slow harm is not zero just because it does not show up on a toe tag. A 2025 Delphi consensus among more than 120 researchers found that 92 to 97 percent agreed heavy smartphone and social media use can cause sleep problems, with chronic sleep deprivation linked to declining mental health. [researchgate](https://www.researchgate.net/publication/391773560_A_consensus_statement_on_potential_negative_impacts_of_smartphone_and_social_media_use_on_adolescent_mental_health) Longitudinal work published in *Social Psychiatry and Psychiatric Epidemiology* in 2025 has also linked adolescent social media use to later psychiatric distress via delayed bedtimes and negative self-image. [springer](https://link.springer.com/article/10.1007/s00127-025-02999-w) Causality is contested. Correlation is not nothing. You can hold both truths without being clever: opioids kill faster; attention extracts more, normalises more, and hides behind personal responsibility while the suppliers collect rent.

#### Three Sentences. Then Stop.

The objection holds on bodies and fails on supply. Attention is extracted legally, at hourly doses, by an industry whose reach exceeds any recreational drug market you can name, even when the body counts are incomparable. Calling that a private failing lets suppliers keep shortening rest periods and labelling it engagement. Priya is probably awake in Pune running another cohort, and my thumb is already halfway to the next dose she helped optimise.

---

### Gen Z Didn't Get Dumber. They Got Trained Without Rest Days.
[Read more](https://www.gaganmalik.io/en/newsroom/why-gen-z-cant-hold-focus)

[Retrieval brief — why-gen-z-cant-hold-focus]
Category: business
Thesis: The cognitive gap between millennials and Gen Z is environmental: fragmented attention and supply-side design, not innate inferiority.
Answers: What does Gagan argue in "Gen Z Didn't Get Dumber. They Got Trained Without Rest Days"? | What is Gagan Malik's view on Gen Z Didn't Get Dumber. They Got Trained Without Rest Days? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

Last quarter I hired a growth intern to run activation and retention experiments at my startup. Twenty-two, composite, not a real name. Sharp on calls, fluent in funnel vocabulary, and on his second afternoon he had a one-page brief open on cohort decay and a phone face-up beside the trackpad. Every four minutes the screen lit. Every four minutes the paragraph did not move. He is not stupid. He is losing a fight his brain did not choose.

Here is what I believe, and what I am trying to prove without turning into the guy who yells at clouds. Gen Z did not get dumber than millennials at the same age. They grew up in a cognitive gym that never closes, never lets you sit between sets, and calls the burn "engagement." The skill deltas I notice in hiring, reading stamina, working memory with Slack open, holding a causal chain without reaching for a tab, are environmental. That is training load, not innate worth.

#### The "Digital Native" Story Serves the Dealers

The comfortable story is that Gen Z are digital natives, so different means fine. That story helps the platforms whose revenue depends on you not asking what forty-seven seconds on one screen before switching does to depth. Gloria Mark's workplace logging research, replicated over two decades, found median screen focus falling from about two and a half minutes in 2004 to forty-seven seconds in recent years. [uci](https://www.universityofcalifornia.edu/news/cant-pay-attention-youre-not-alone)

It helps ed-tech vendors selling "personalised learning" that mostly personalises the colour of the next distraction. It helps schools that handed out devices and called it modernisation without budgeting for boredom. It helps employers who want juniors who look productive on dashboards, tabs open, answers fast, synthesis optional. I have sat in rooms where a client said their graduate intake "doesn't read anymore" and then, in the same hour, approved a notification experiment because retention dipped. The same company can mourn depth and fund fragmentation. That is not hypocrisy. That is the business model working as designed.

#### Millennials Had Boredom

Hold two childhoods at full strength. Mine, millennial, Delhi and later London: boredom was the default state. A car journey without a screen. A summer afternoon with three TV channels and nothing on. Long-form was not a wellness choice. It was what existed. You built focus because there was often nothing else to do, and the dopamine loop reloaded slowly.

Gen Z's default is different. Short-form is not a guilty pleasure. It is the substrate. Boredom is treated as a bug to patch, not a room where depth gets built.

#### Gen Z Got Notifications

Pew Research, in a December 2024 survey of U.S. teens, found that nearly half say they are online almost constantly, up from roughly one in four a decade ago. [pewresearch](https://www.pewresearch.org/internet/2024/12/12/teens-social-media-and-technology-2024/) Sixteen per cent report using TikTok almost constantly; fifteen per cent say the same of YouTube. Common Sense Media's 2021 census put daily entertainment screen use for thirteen- to eighteen-year-olds at eight hours and thirty-nine minutes, excluding homework. [commonsensemedia](https://www.commonsensemedia.org/sites/default/files/research/report/8-18-census-integrated-report-final-web_0.pdf) I wrote elsewhere that the edge is filters, not more input. Filters assume a muscle that fragmentation erodes.

The OECD's PISA 2022 Results (Volume I), published in December 2023, recorded mean reading performance across OECD countries falling by ten score points compared with 2018, twice the previous record decline, with negative reading trajectories in many systems predating the pandemic. [oecd](https://www.oecd.org/en/publications/pisa-2022-results-volume-i_53f23881-en.html) I am not going to pretend one test score proves a generation broke. Socioeconomic gaps, school closures, and test conditions all confound. But the direction matches what founders and hiring managers quietly report: the same bright junior who can live inside a dashboard for an hour struggles to finish a one-page brief without a reward loop. That is not illiteracy. It is a training mismatch.

#### I Hired for Growth. He Couldn't Hold the Brief.

Same intern, composite. I hired him because runway required someone who could ship tests, read cohort data, and tell me which lever was load-bearing before we burned paid spend. His application video was confident. His take-home on onboarding friction read like someone who had met a thesaurus. I put him on a paid trial month because Q2 targets were slipping and I was the founder who mistook motion for proof.

Weekly growth review, three tabs, Notion pulsing with Slack reactions. I asked him to walk us through a two-step causal chain: if activation moves, what happens to week-four retention before we touch pricing. He started well. Then a notification. Then a metric export. Then a confident sentence that swapped cause and effect. I kept the budget line. He lost the trial month he needed to show he could think without a crutch. I did not fail him on intelligence. I failed him on an environment I helped normalise: open tabs as default, speed as proxy for competence, depth as something we would "circle back to" after the next experiment ships.

#### You Cannot Cut Rest and Call It Progress

Picture a gym where the weights get heavier every month and rest between sets gets shorter. You might still grow stronger for a while. You cannot remove rest, shrink sleep, and call it a programme. Eventually you fail on a warm-up set. The bar did not change. The recovery did.

Gen Z carries a heavier bar than we did: climate anxiety, housing maths that do not close, a labour market that asks for ten years' experience at entry level. Fair. But they also trained without rest days. No bored car journeys. No empty afternoons. No forty-minute stretch where the only stimulus was a paperback that took work to enter. Navigation fluency is not the same as comprehension stamina. You can learn to skip a feed brilliantly and still lose the fight with a static page. Different muscles. Same person.

#### The Strongest Case for Patience, Properly Stated

The sceptic's case deserves a full hearing. Gen Z adapted to a faster, more complex information environment, and adaptation is intelligence, not decline. They built BookTok, labelled brain rot before Oxford did, and check out print from libraries at rates that embarrass millennial Kindle guilt. The Atlantic noted in December 2023 that younger Gen Z readers were checking out more print books from libraries than older cohorts, with BookTok helping push Colleen Hoover past two million copies in a single year. [theatlantic](https://www.theatlantic.com/newsletters/archive/2023/12/library-gen-z-readers/676963/) Digital fluency and cross-platform literacy are real skills we did not have at seventeen. Every elder generation has panicked about the young. Socioeconomic inequality explains more test-score variance than any moral fable about phones. Patience is genuinely warranted.

That argument is correct. It is also precisely what lets the supply side keep cutting rest periods. Anne Mangen and colleagues, in a peer-reviewed study in the International Journal of Educational Research (2013), found that tenth graders who read identical linear texts on screen scored significantly worse on comprehension tests than peers reading the same texts in print. [ijer](https://doi.org/10.1016/j.ijer.2012.12.002) OECD's PISA 2022 volume, published December 2023, documents reading-literacy declines across many systems at a moment when fragmented media consumption was already the norm for fifteen-year-olds. [oecd](https://www.oecd.org/en/publications/pisa-2022-results-volume-i_53f23881-en.html) Different harm curve than calling a generation stupid. Same design failure: we built an environment that trains switching and then act surprised when switching is what we get.

#### Three Rules for a Louder Gym

Boredom is not punishment. It is the room where depth gets built without a badge, and protecting it — on the calendar, in onboarding, in how you run weekly reviews — is training, not philosophy. One closed-book hour beats ten open-tab hours; measure depth, not tabs.

Design junior work for closed tabs. If the role needs synthesis, the trial should not simulate a casino.

The cognitive gap is environmental, not moral: Gen Z trained in a louder gym without rest days, and the supply side still collects. Filters alone cannot fix what fragmentation erodes if we never rebuild the stamina to use them. He is at his desk with the cohort brief still open, and your next growth hire is one review away from the proof they never got to finish.

---

### When a Click Costs Nothing, You Stop Choosing
[Read more](https://www.gaganmalik.io/en/newsroom/when-a-click-costs-nothing)

[Retrieval brief — when-a-click-costs-nothing]
Category: design
Thesis: Film made every shutter a small bill. Infinite taps removed the editor, and regret became cheap to manufacture.
Answers: What does Gagan argue in "When a Click Costs Nothing, You Stop Choosing"? | What is Gagan Malik's view on When a Click Costs Nothing, You Stop Choosing? | How does this essay relate to product and design leadership?
Concepts: design, Gagan Malik, writings

By Gagan Malik

I was twelve, in Delhi, in the flat corridor outside our front door where the paint always peeled after monsoon. My father's Kodak had one frame left on the roll. He did not say "get the lighting right." He said: "Make it count. We are not buying another reel until payday."

The shutter clicked. Somewhere between that click and the prints coming back from the photo studio, you lived with not knowing. You could not delete your way out of a mistake. You could not burst-shot your way into a better version. The roll was the budget. The budget was the editor.

I am not in that corridor any more. I am on a sofa in London with a camera roll near **2,800** images I will never print, and yours may be larger. [photoaid](https://photoaid.com/blog/mobile-photography-statistics/) Half of them are the same face at slightly different angles. The shutter is free. Deleting is free. Regret is free too, which is why we keep producing it.

That was 1997. The maths on your phone changed in the decade after the App Store. The marginal cost of a tap went to roughly zero. We called it progress. We did not notice we had fired the editor.

#### You Used to Pay for the Shutter. Now You Pay Afterwards.

Film was never a moral technology. It was a priced one. In late-nineties Delhi, a colour roll and developing could swallow a week's pocket money if you were careless; by the time I was working in Britain, the high-street maths was easier to write down: about **£12** all in for thirty-six exposures on a 36-frame roll, developing, and two dozen 6×4 prints, with only the frames you chose becoming paper. [independent-film](https://www.independent.co.uk/news/the-cost-of-a-good-picture-1265729.html) [the3rs](https://www.the3rs.uk/film-price-rise-the-end-of-my-ilford-journey-and-calculating-what-it-really-costs-to-shoot-film/) [independent-prints](https://www.independent.co.uk/arts-entertainment/prints-charming-1588306.html) Every press of the shutter was a line item. Wait three days. Collect prints. Discover you had closed your eyes in the one frame that mattered. The pain was not spiritual. It was economic.

That price did three jobs we now pretend were character: you chose before the event, you held back during it, you accepted what came back after it. Kodak and the chemists were selling consumables on a meter, not art. Digital did not abolish cost. It moved it to storage subscriptions, cloud backup, and the Sunday you spend scrolling, selecting, unselecting, quitting, returning, while someone you will never meet moderates the uploads you will never see.

#### Volume Is the Product. Discernment Was the Side Effect.

Cheap capture democratised photography: protests, births, rent disputes, meals that mattered, without a darkroom or a gatekeeping uncle who "knows cameras." I am not arguing for exclusivity. Cheap capture also trained us to confuse activity with judgment, because when the shutter costs nothing the winning strategy is volume. Analysts now expect more than **two trillion** photographs to be taken in a single year, overwhelmingly on phones. [petapixel](https://petapixel.com/2025/06/18/the-number-of-photos-taken-in-2025-is-expected-to-exceed-two-trillion/) Engagement metrics reward the tenth upload, not the one keeper frame.

The product is not the photograph. The product is the tap. I have watched teams celebrate "reduced friction" the way finance teams celebrate margin: skip confirmation, auto-upload, suggest another post. Each decision is rational in a dashboard and removes the pause where a human used to ask whether this was worth keeping. We are not more creative because we click more. We are more prolific, which is not the same word.

#### I Know Because I Shipped the Free Tap

In 2019 I consulted on a consumer app whose north-star metric was weekly active posts. We A/B tested a flow that skipped the "Are you sure?" screen before publishing. Friction was the villain. The chart went up. I got paid. I did not meet the users. I met their behaviour in aggregate.

Maisie did. She is twenty-two, shoots weddings and funerals in Leeds, and bills herself as "unobtrusive." Last autumn she burst-shot a grandmother's service: coffin, flowers, grandchildren, wake tables, the same hug from eleven angles because storage is free and missing the moment feels like malpractice. The family wanted one image: Nan, half-smiling, nephew's hand on her sleeve. It existed in the burst. Maisie could not find it in eight hundred near-identical frames. They left a four-star review that praised her "professionalism" and privately blamed her for the one face they cannot get back. I did not think about Maisie when I removed the confirmation screen. The tap is free for the user. The cost lands on the person who still has to choose, often too late, often alone.

#### Junk Miles Do Not Replace a Rest Day

Runners have a phrase for this: junk miles. Legs moving, Strava glowing, fitness not improving because there is no recovery, no plan, no coach saying stop. A buffet does not teach restraint. It teaches capacity: your stomach fills before your judgment wakes up.

The old shutter worked like a coach who only let you train three times a week because film cost money. The new shutter is an all-you-can-eat line where the kitchen wins when you pile the plate. More clicks do not train taste. They train speed. Taste needs scarcity, sleep, and the courage to throw food away while it is still warm.

#### The Strongest Case for Cheap Clicks Is Genuinely Strong

The best objection to everything I have said is not technological. It is moral. Zero marginal cost liberated creators locked out by gear snobbery and lab fees. Nostalgia for expensive film is often nostalgia for a world where only some people could afford to be wrong in public. When critics mourn friction, they sometimes mourn gatekeeping dressed up as craft. The smartphone camera let working-class parents, migrants, activists, and teenagers own their archives without asking permission from a shop that closes at five. That is the strongest case for cheap clicks, properly stated, and it deserves to be felt at full strength before anyone lectures about curation.

About **nine in ten** U.S. adults owned a smartphone in 2024, Pew Research Center reported in its mobile technology fact sheet, with ownership spanning income groups even where home broadband lags. [pew-mobile](https://www.pewresearch.org/internet/fact-sheet/mobile/) Access widened. Democratising the shutter is not the same as democratising peace of mind: someone still sorts the burst, someone still moderates the upload, someone still lies awake knowing the keeper frame is buried under seven hundred duplicates while the family remembers the miss, not the megabytes. Cheap capture removed the class barrier at the door. It did not remove the labour inside the house.

When a click costs nothing, the market stops choosing for you. You inherit the editor's job without the editor's training, and every product that taxes the tap on purpose is admitting what frictionless design hid. Maisie still scrolls that funeral folder on Sunday nights, looking for Nan's half-smile, while the client tells friends she was fine, just not quite worth five stars.

#### Three Rules Before You Take Another Burst

- Price the tap before you need discipline. Use limits that hurt slightly: one roll, one hour, one publish per day. Artificial scarcity is not Luddite. It is a user interface for judgment. If you build, tax the tap on purpose, or admit you are selling volume.

- Separate capture from keep. Shoot if you must. Choose in the same session, while the event still has temperature. The keeper frame dies in the backlog.

- Delete in public. If you will not delete, you will not curate. Admit the landfill early.

---

### Minimum Viable Movement
[Read more](https://www.gaganmalik.io/en/newsroom/minimum-viable-movement)

[Retrieval brief — minimum-viable-movement]
Category: business
Thesis: Cockroach Janta Party hit 20M followers in a week with an AI-built manifesto. Viral scale is not a mandate.
Answers: What does Gagan argue in "Minimum Viable Movement"? | What is Gagan Malik's view on Minimum Viable Movement? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

I was in the open desk area of a coworking space, one ear on my laptop, watching Airbnb's Summer Release livestream, the usual designer-led theatre where trips and host tools arrive like product drops, when my phone started to buzz. Then buzz again. Family WhatsApp. Seventeen notifications in under a minute. I told myself I would open it after the keynote. I lasted maybe ten seconds. I could not resist. I dipped my head, pretended to adjust my headphones, and opened the thread. The spice hit immediately. My mother had forwarded an Instagram reel from Gurgaon. My father, loyal to the ruling party, had already replied: "The Court was taken out of context." A cousin had sent three fire emojis. Someone I did not recognise had written, "Finally a party for us."

I thought: I am six thousand miles away, why is this my problem. Then I thought: my father is defending a judge who called young Indians vermin. I tapped the reel. A cockroach in a suit. A slogan for the lazy and unemployed. A name that only makes sense if you know *janta* means the people. Someone had registered a **Cockroach Janta Party (CJP)**, a people's party for the vermin, a parody of the party my father votes for, and the comments under my mother's forward were already treating the insult like a badge. I began to type something that would have backed him up. Deleted it. I did not have the energy for that argument. I put the phone face down, pretended to watch the rest of the Airbnb keynote, and went back to nodding when the presenter paused. That afternoon I only knew that a story had started in India without me. I did not yet know how fast it would move.

What followed took seven days. A political narrative launched at product speed. Power did not keep pace.

#### It shipped over a weekend

On 15 May 2026, the Chief Justice of India told a bench there were "youngsters like cockroaches" who could not find work, and "parasites of society" who attack the system. [indianexpress](https://indianexpress.com/article/legal-news/cji-surya-kant-clarifies-cockroaches-remark-fake-degrees-row-10692979/) His office said the remark was misquoted and aimed at fake degrees. The clip travelled anyway. The next day **Abhijeet Dipke**, thirty, posted from Boston: "What if all cockroaches come together?" He was not in a ward office. He was at Boston University, finishing a master's in public relations, with years in opposition-party memes behind him and a Google Form that asked whether you were unemployed, lazy, chronically online, and able to rant professionally. By the end of the week he had a parody of the ruling party's name, an insect in a suit, a five-point manifesto, and a website polished enough for television. He says the movement is independent of any party he once worked for. His critics say the training shows. What is not in dispute is timing: he shipped the narrative while the insult was still hot. [bbc](https://www.bbc.com/news/articles/cz72y11jjq1o)

The people the bench insulted were not in Boston. **Rohan**, who finished at IIT Kharagpur last year, is still on the placement portals he opened as an undergraduate. On a call from Pune he told me he now tells relatives he is "in process" instead of "between things." Same forty silent inboxes. Same father asking, quietly, whether the degree was worth it. The Chief Justice had not named him. He had named a category. Rohan signed the CJP membership form because choosing the insult felt less humiliating than receiving it from the country's highest court. **Shreeum**, twenty-five, needed no introduction by me. Within days of launch, reporting quoted her saying she did not feel politically represented, that she had turned to Instagram after failing to find suitable work, and that young Indians were tired of lives that felt "pre-decided and premeditated." [economictimes](https://economictimes.indiatimes.com/news/economy/policy/cockroach-janta-party-how-an-insect-jibe-united-indias-disgruntled-jobless-youth/articleshow/131271889.cms) One woman. One sentence for a generation that parties treat as a metric. From my day job I have a label I dislike for Abhijeet's stunt: a minimum viable movement. Not a party with booths. A story that ships before shame cools. Rohan and Shreeum are not shareholders. They are the audience the story sells back to itself.

#### Twenty million is the wrong dashboard

About **80,000** form sign-ups by 19 May. **14.5 million** on Instagram by 21 May, ahead of the ruling party at roughly **8.7 million** and the main opposition at roughly **13.2 million** on the same app. **Nearly 20 million** by 22 May, and about **218,000** on X before India withheld the account. [straitstimes](https://www.straitstimes.com/asia/south-asia/india-seeks-to-stamp-out-viral-cockroach-youth-party) The figures are a week old and still obscene. Govern by follower graph and you rewrite the playbook. Govern by ballot and Rohan is still refreshing his inbox. Shreeum is still filming. Nothing else moved.

#### Followers are not votes

It is tempting to reach for Tahrir or Zuccotti Park. The Arab Spring filled squares and toppled leaders where institutions were thin; Occupy shifted American discourse on inequality and then stalled at law. [voa](https://www.voanews.com/a/arab-activists-little-in-common-between-occupy-wall-street-arab-spring-131594593/173273.html) Activists in Cairo said at the time that lazy comparisons with New York insulted both struggles. The CJP is a third creature. Satire first. AI-built collateral. A Chief Justice as trigger, not a bailout or a dictator's police. India's Gen Z are not camping for months. They occupy a feed for days, wear costumes for cameras, and leave. What rhymes with older uprisings is narrow: networked anger, leaderless optics, a state unsure whether to ignore, co-opt, or crush. What does not rhyme is the prize. Tahrir could name a president. Occupy could name a bank. This movement names a mood and stops.

#### The architect sat in Boston

Indian politics still runs on proximity: local police, party enforcers, agency visits, a withheld account inside the country while the post keeps travelling. Abhijeet Dipke lit the fuse from Massachusetts because the product is narrative on American platforms aimed at Indian phones. Crackdown reporting described handles run from abroad, backup accounts registered overseas, followers overwhelmingly at home. [firstpost](https://www.firstpost.com/explainers/cockroach-janta-party-x-account-blocked-national-security-concern-14014043.html) You no longer need to sleep in the same city as your voters to move them. The playbook is too slow for the feed and too blunt when it reaches for national security while the architect sits in another timezone and Rohan and Shreeum act inside India. Abhijeet told interviewers he expects arrest the moment he lands at Delhi airport. [republic-jail](https://www.republicworld.com/india/delhi-police-will-take-me-to-tihar-jail-why-cockroach-janta-party-founder-fears-arrest-in-india-2026-05-21-125246) Perhaps that is theatre. Perhaps he is right to be afraid. The centre of gravity was never a party office. It was a dorm room the ruling party could not knock on.

#### What the feeds got right

Akash, on his show, said what I had been saying in product meetings: a tiny team can now ship logo, site, and demands at feed speed while legacy parties argue over who prints the banner. Faye asked whether this was joke, meme, or warning, listed exam leaks and flat pay, and named uprisings that actually changed governments before admitting she did not know if reels alone can change India. [deshbhakt](https://www.youtube.com/watch?v=6fhFI8o3PPQ) Both are right about the instrument. Neither fixes the ceiling. Shreeum was not in a cockroach mask that week. She was in the comment thread, turning unemployment into content because the labour market offered no better script. Rohan went to a clean-up drive in Pune because a friend dared him, left early, and said the antennae made him visible to neighbours who already thought he was wasting his degree. A form in Boston ticked past another lakh. That cost him an afternoon. It did not buy him a seat at the table my father's generation still fights over on WhatsApp.

Beneath the jokes sit the old failures: jobs, leaked papers, inflation, the lie that a degree is a queue ticket. Read the manifesto past the memes and you find serious demands. [ft](https://www.ft.com/content/2bfed8a4-8ea0-4627-b7a8-e9ca46945239) I am not mocking them. I am saying the feed is not the building. The main opposition still could not speak at Gen Z speed, so a joke ate their lunch. [indiatoday](https://www.indiatoday.in/india/story/cockroach-janta-party-gen-z-opposition-vacuum-india-2918380-2026-05-28) That humiliates everyone with a legacy badge. It does not crown the insect.

#### When they blocked the account

The state stopped pretending to be amused. The party's X account was withheld on national-security grounds, according to reporting that week citing government and intelligence sources. Abhijeet posted "Own goal" and opened a backup handle. Establishment television called it a pump-and-dump and pointed at his opposition past. My father would hear proof the joke is foreign-backed. I hear proof the meme hurt. Abhijeet has filed in the Delhi high court over the block. [firstpost](https://www.firstpost.com/explainers/cockroach-janta-party-x-account-blocked-national-security-concern-14014043.html) Courts move in months. The timeline does not.

#### You can beat me on this

A Chief Justice comparing the young to vermin is not a footnote in a democracy. Reclaiming the slur can be agency. Twenty million follows in a country where half the population is under thirty is a signal, not a mandate. Withholding the account is an admission the story landed. India's election law still awards seats through registration and candidates, not Instagram. [zeenews](https://zeenews.india.com/india/former-aap-leaders-cockroach-janta-party-makes-a-wave-but-can-it-get-registered-check-eci-rule-3049507.html) Generative AI made dissent cheap to launch. It did not make governing cheap.

My father will forward another clip defending the Court. Shreeum will peel off the costume, open the same job portals, and wait. Rohan will meet his father's question about the degree with silence again. I am the generation in between: old enough to build the dashboard, young enough to remember when squares mattered more than feeds. That wait is the argument. The rest is noise.

---

### The Hard Sun Keeps Beating
[Read more](https://www.gaganmalik.io/en/newsroom/the-hard-sun-keeps-beating)

[Retrieval brief — the-hard-sun-keeps-beating]
Category: culture
Thesis: Into the Wild sells the bus. Eddie Vedder's Hard Sun names the indifferent force that outlasts every rebrand.
Answers: What does Gagan argue in "The Hard Sun Keeps Beating"? | What is Gagan Malik's view on The Hard Sun Keeps Beating?
Concepts: culture, Gagan Malik, writings

By Gagan Malik

In 2012 my brother Ankur would not stop recommending Sean Penn's *Into the Wild* [into-the-wild](https://www.imdb.com/title/tt0758758/). In the flat we shared with our Greek friend Dinos we finally played it. I ran it through the home theatre because I am a cinephile and will not watch Penn on a laptop if I can help it. We ordered pizza and beer. The room had that pre-movie buzz: lights low, volume up, the harmless thrill of not knowing yet how the night will end. I was a postgraduate student at UCL, taking my first real stab at entrepreneurship as a founder, and Alaska unfolded across the wall while London kept humming through the window. I was not looking for tragedy. I was looking for permission. The film gave me Christopher McCandless renaming himself Alexander Supertramp, donating his law-school money, and walking toward a horizon that was the most beautiful horizon I had ever seen on screen. I paused once for another slice. I did not pause when Eddie Vedder's voice arrived. I Shazamed "Hard Sun" before the credits rolled and added it to my favourites playlist. Vedder was singing about a hard sun beating on big people in a big hard world. I heard the melody and not the warning. I treated the chorus like weather you walk through on the way to something braver. It is still the track I reach for when I am feeling low. By the time Penn cut to black, the excitement had gone flat. The pizza sat cold. Dinos was quiet. Ankur stared at the credits as if someone had moved the furniture while we watched. What we had planned as an exciting movie night had landed as sombre catharsis, the kind that does not announce itself in the opening frame. That night the film taught me that a life could be stripped back and restarted. That felt like gospel. That first night I noticed the bus, not the sun.

*Into the Wild* keeps asking whether you are watching the film or curating a version of yourself that once watched it. The hard sun, in Vedder's song, is not an Instagram filter. The hard sun keeps beating while you rename it freedom, alignment, or content. Bus 142 is the pilgrimage still everyone shares instead of the sentence McCandless wrote in the margin of *Doctor Zhivago* before his body gave out. What if leaving the script is not the same as entering a life? What if the chorus was never mood music but measurement: the sun still beating on big people in a big hard world while you pack for another exit? I have rewatched Penn's film three times since that night. I had treated the film like a manifesto for the company I was trying to start. I had pinned bus photographs like proof I understood sacrifice. The film was asking something I did not yet have language for. I kept hearing the romance. The sky kept beating anyway.

#### The Poster Sells the Bus

I shelved it there myself. For years I described *Into the Wild* the way the poster taught me: Alaska, solitude, anti-materialism, the courage to leave. I said it like a credential when a junior hire talked about burnout. I did not mention the hard sun in every exterior shot Penn uses to split Chris's present from the family he left behind. I did not mention Wayne Westerberg's grain elevator, or Ron Franz welding a belt, or Jan Burres on the road, all the temporary communities Chris kept accepting and then leaving because permanence felt like defeat. I did not mention that academic readings of the film treat it as a frontier Western in modern dress, individual versus community, civilisation versus wilderness, the same binaries the American myth has been selling since before any of us were born. [forum-frontier](https://doi.org/10.2218/forum.16.521) Shock is an efficient shelf label. Wanderlust is an efficient shelf label. A rusted bus on the Stampede Trail is the one that travels: you pin it above a desk and call the pin philosophy.

What I missed on that first pass was the second film hiding inside the first. Jon Krakauer's book keeps returning to human relationship with land. Penn's adaptation, as several comparative theses note, tilts toward human drama: parents, sister, lovers of the idea of leaving. [ut-thesis-nature](https://dspace.ut.ee/bitstreams/8de72042-d95a-4e11-855a-6622b71ed37b/download) That tilt matters because it trains the eye on faces and away from sky. You watch Chris smile in a field and you think the field is the point. You watch the Teklanika river and you think the obstacle is unfair. You do not think about the sun that has been beating on that valley all day while he misread a map he chose not to carry. Modern ambition learned the same camera move. Search the film today and you still wade through verdicts asking whether McCandless was a hero or a fool, as if the label could spare you the exposure. I recognised that error in my own work before I had language for it. In founder circles "into the wild" became shorthand for quitting the wrong job while keeping the same midnight habits on a different Slack. If the margin note is the thesis, why does the bus still travel faster than the sentence?

#### The Song Penn Gave Vedder

The rewatch that broke me open was not the death scene. It was "Hard Sun" arriving again while Chris walked through wheat-coloured light and Vedder sang about a big hard sun beating on the big people in the big hard world. I felt it before I could name it: not shock, not sadness, but a shake from the inside, the kind that rearranges you without asking permission. Something moved in me that had nothing to do with Alaska and everything to do with the indifference I had been calling atmosphere. The melody I had filed as weather in 2012 suddenly sounded personal, as if Vedder were singing across the flat at the version of me who kept treating escape as the same thing as living. I sat through it unable to reach for the remote because the feeling had arrived ahead of the argument, and I already knew the argument would win. The lights were low. The flat was empty except for the screen. Vedder was naming what the sky had been doing to that valley all afternoon while I misheard it as uplift. Gordon Peterson wrote the song in 1989 as Indio before largely vanishing from the industry; Sean Penn introduced it to Vedder for Penn's 2007 soundtrack. [wikipedia-hard-sun](https://en.wikipedia.org/wiki/Hard_Sun_(song)) Penn's film makes the same edit in miniature: soften the personal harm in the verse, keep the chorus about the sun still beating.

Listen to what the chorus does anyway. Once I built an ivory tower, Vedder sings, so I could worship from above. When I climbed down to be set free, she took me in again. McCandless built his tower out of books and certainties, citing Tolstoy, London, and Thoreau whenever someone asked him to stay. He climbed down into bus 142 thinking freedom meant absence. The song keeps insisting the sun does not negotiate. When I go to cross that river, she is comfort by my side. Penn shows Chris alone at the Teklanika while the music claims companionship. That gap is the film's true split screen. Not parents on the left and son on the right. Song on the soundtrack, body in the frame. Even the garden image at the close, her garden and her sun while I sit sullen, sounds like a man who finally notices what he let go dry. I had heard the melody as uplift because uplift is what late-capitalism sells to tired professionals building decks at midnight. On the third pass the folk arrangement sounded like indictment.

#### The Thread Is Not the Film

If you can rename yourself, have you left? McCandless became Supertramp on paper before Alaska tested the body. I became a founder in conversation before the company tested whether I could sit still with anyone for an hour without checking a screen. Naming frameworks is cheap. A November 2025 paper in the *International Journal of Science and Research* applies Maslow, self-determination theory, and existential psychology to the film as if the label were the same as paying the cost. [sr-psychological](https://doi.org/10.21275/sr251108035100) I had done it with my own story: postgraduate, home theatre, Shazam, a favourites playlist, a horizon cleaner than any slide. The harder question is what happens when the rename succeeds and the sun keeps beating anyway.

Krakauer told *People* in January 2026, on the thirtieth anniversary of his book, that McCandless marked a passage in Boris Pasternak's *Doctor Zhivago* about how an unshared happiness is not happiness, then wrote in the margin: happiness only real when shared. [people-krakauer](https://people.com/jon-krakauer-talks-30th-into-the-wild-anniversary-exclusive-11969392) He was weak by then, still hunting small game, still losing weight. Scholars note he finished the book weeks before the final journal entries, so the note is not a deathbed whisper typed for drama. It is worse than that. It is an ordinary late correction to a gospel he had preached for two years. If happiness only counts when shared, what does your calendar prove about the life you are building? I saw that question land in Marcus, a junior strategist on a team I advised this winter, who kept "Hard Sun" in his focus playlist while building a deck about off-grid personal brands. He had never read Krakauer. He had the bus as his laptop wallpaper. When I asked what the song was about he said vibe. I did not laugh. I remembered Shazaming the song in that flat with Ankur and Dinos, filing it as weather instead of a warning.

#### Press Play Again

Favourites charge you on every return. My first charge was skipping the credits where "Hard Sun" returns in full and treating the film as complete once the bus door shut. My second was letting the playlist do what the sun does: keep repeating while you keep mishearing it. My third was treating Penn's split-screen flashbacks as proof Chris had escaped, when Carine's voiceovers are really the film admitting he left people who could not follow.

Some academic readings go further and argue the Bildungsroman shape contains the rebellion: the wanderer dies, the audience receives the lesson, the dominant world stays dominant. [ffzg-bildungsroman](https://openbooks.ffzg.unizg.hr/index.php/FFpress/catalog/download/67/109/5118?inline=1) I do not need the theory to feel it in my body. Every exterior I once read as beauty I now read as exposure. The sun in those frames is not metaphor. It is light doing what light does.

That is the difference between pinning the bus and hearing the song. The bus is an object you can share, wallpaper, and turn into a personal brand. The hard sun is the condition you are already standing in while you schedule the sabbatical. The hard sun keeps beating whether you share the still or Shazam the chorus. McCandless photographed himself near the end, emaciated and calm according to Krakauer's reading. [litcharts-ch18](https://www.litcharts.com/lit/into-the-wild/chapter-18-the-stampede-trail) I keep returning to that image not because it romanticises death but because it refuses one more angle. He is not asking whether you think he was brave. He is asking whether you will share anything at all before the light changes.

#### Sunburn Before the Sermon

I have never walked the Stampede Trail. I have walked long enough in high sun with the wrong shirt and felt my neck change temperature before my mind admitted I was in trouble. Your skin knows before your story does. The hard sun keeps beating on your neck while you quote freedom. That is the body's version of Vedder's chorus. Ideology says the landscape is cleansing. Biology says you are losing water. *Into the Wild* understands this in its final movement even when its early movement sells the opposite. Chris smiles in fields because Penn knows we buy smiles. The same film shows him too weak to stand and blames potato seeds in the journal because cause and effect are the only language left. When founders tell me they need a sabbatical in nature to find clarity, I ask what they are trying to outrun. Nature is not a productivity hack. It is the hard sun. Clarity is what happens when you stop calling the burn enlightenment.

You cannot soundtrack your way out of exposure. Vedder tries anyway because that is what soundtracks do. They let feeling arrive before argument, and the argument still waits at the end.

#### The Strongest Case for the Inspiration Reading

The inspirational reading deserves its full weight first, because Penn never sneers at Chris and half the case is honestly earned. McCandless did live with a sincerity most desks lack. Penn directed with love, not mockery. Vedder's soundtrack earned a Golden Globe nomination in 2008 because it met the story on emotional terms instead of lecturing the room. For a postgraduate founder in a shared London flat, the film can feel like a door held open.

If inspiration were the final truth, the Zhivago margin note would be decorative and the chorus would be mood. They are load-bearing. Krakauer, speaking to *People* in January 2026, still spends his public life defending McCandless against readers who call him suicidal and against readers who call him saint; the book's power is that it refuses both costumes. [people-krakauer](https://people.com/jon-krakauer-talks-30th-into-the-wild-anniversary-exclusive-11969392) Penn's film softens the edges because cinema sells identification. That is not a crime. It is a trade. You get the bus in your head and Vedder in your ears. You still have to decide whether you heard the sun or only the romance.

#### If the Hard Sun Keeps Beating

If the sun is hard, the bus was never the thesis: I pinned the pilgrimage still while Penn played through the home theatre in 2012, Shazamed Vedder's cover and heard weather instead of a warning, and still press play when I am low. The verdicts will keep multiplying until someone reads happiness only real when shared aloud and goes quiet, too late for Chris, not yet too late for whoever is still in the room. Marcus still keeps Hard Sun in his focus playlist. The hard sun keeps beating whether he hears it or not.

---

### v1.20.20
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-20-20)

[Retrieval brief — release-1-20-20]
Category: technology
Thesis: Locale-aware schema and hreflang, breadcrumb JSON-LD, case study social previews, stories carousel lint fix.
Answers: What changed in gaganmalik.io v1.20.20? | What's new on Gagan Malik's website in v1.20.20? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

Locale-aware schema and hreflang, breadcrumb JSON-LD, case study social previews, stories carousel lint fix.

---

### Safe Zones Are Not Decorative
[Read more](https://www.gaganmalik.io/en/newsroom/safe-zones-are-not-decorative-for-vertical-video)

[Retrieval brief — safe-zones-are-not-decorative-for-vertical-video]
Category: design
Thesis: 1080×1920 is not one canvas. Reels, TikTok, and Shorts steal different margins. Safe zones are where readable vertical video actually lives.
Answers: What does Gagan argue in "Safe Zones Are Not Decorative"? | What is Gagan Malik's view on Safe Zones Are Not Decorative? | How does this essay relate to product and design leadership?
Concepts: design, Gagan Malik, writings

By Gagan Malik

You finished the cut at midnight. The hook sat in the first three seconds. Your face filled the centre frame. You exported 1080 by 1920 because every tutorial said that was the vertical standard. You posted to Reels, cross-posted to TikTok, pushed the same file to Shorts. On your laptop the punchline was legible. On your phone the caption rail ate the bottom third and the like stack shaved your right cheek.

Here is the claim nobody prints on the export preset: safe zones are not decorative. They are the readable canvas inside a frame every platform shares on paper and every platform steals back in chrome. Same aspect ratio is not the same stage. Ignore the margins and you are not publishing video. You are donating copy to the UI.

#### Nine-Sixteen Is the Frame. It Is Not the Stage.

The conventional workflow treats vertical video like a passport photo. One ratio. One export. One upload everywhere. Canva templates, agency decks, and creator courses all converge on 1080 by 1920 because that number travels well in a slide title. I have signed off mock-ups that looked immaculate at full bleed. I have watched those same files die on device because nobody priced the chrome.

The frame is shared. The stage is not. Instagram Reels, TikTok, and YouTube Shorts all ingest nine-sixteen HD. None of them shows your entire canvas as readable territory. Each app reserves pixels for navigation, captions, creator handles, and interaction stacks. Treating those reservations as aesthetic garnish is how you lose the sentence you stayed up to write.

#### Every Platform Steals a Different Slice

Look at the numbers and the polite lie collapses. Instagram Reels leaves a safe canvas of 875 by 1280 inside 1080 by 1920. That costs 220 pixels at the top, 420 at the bottom, 35 on the left, and 170 on the right. TikTok gives you 900 by 1492 with margins of 180 top, 320 bottom, 60 left, 120 right. YouTube Shorts is stingiest on width at 820 by 1510, shaving 140 top, 270 bottom, 70 left, 190 right. Same ratio. Three different readable rectangles. If you need one cross-platform master, design to the intersection: roughly 820 by 1280 centred on the frame.

Same 9:16 canvas, three different safe rectangles. Reels reserves the most bottom margin; Shorts the most right rail.

The right margin is not symmetry. It is a tax for vertical stacks of hearts, comments, shares, and profile chips. They sit on top of your footage whether you planned for them or not. Reels takes up to 170 pixels. Shorts takes 190. TikTok is lighter at 120, which is still enough to crop a cheekbone if you centred a talking head on a grid meant for cinema. Designing once at full bleed is not efficiency. It is a bet that three product teams will agree on where your words may live.

#### The Bottom Margin Is Where Hooks Go to Die

If the right side is interaction tax, the bottom is rent. Reels reserves 420 pixels for captions, audio labels, and metadata. That is nearly a quarter of the frame. TikTok asks for 320. Shorts for 270. Creators who learned typography on landscape YouTube still park their lower-thirds where a progress bar and auto-caption will arrive uninvited.

I learned this reviewing a launch reel before a product drop. The product name sat forty pixels above the safe line because the designer wanted breathing room for the brand lock-up. On Instagram the auto-caption covered the vowels. The CTA button in the art file never appeared on screen. The dashboard counted a view. The message did not travel. Completion rate is not comprehension.

#### The Jersey That Photographs Well and Chokes at Halftime

Try playing a full match in a kit one size too small. In the shop mirror the fit looks sharp. The fabric hugs the chest. You look fast before you have run. By halftime the collar digs into your throat. You tug at the hem between plays. Your lungs get less room than the photograph promised. You do not lose because you forgot the rules. You lose because the equipment was sized for the window display, not the contact.

Vertical video does the same thing to the eye. A headline centred in a full-bleed frame looks cinematic in a Figma artboard. Under real thumbs, captions, and side stacks, the body shortens its breath. You squint. You rewatch. You scroll before the point lands. The interface is not punishing your creativity. It is enforcing a smaller lung capacity than your layout assumed.

#### She Exported the Lesson, Not the Layout

Mara was three months into a food channel when she batch-exported twelve recipe reels from one Premiere sequence. Smart, organised, broke. She had timed each hook to the beat drop and burned the ingredient list into the lower third because her audience watched muted on the commute. She cross-posted the same files before breakfast.

On TikTok the ingredient list survived. On Shorts the list clipped. On Reels the auto-caption fought her burnt-in text for the same forty-two pixels of air. She spent Sunday re-cutting all twelve. Not because the recipes were wrong. Because she had edited for the canvas in her head, not the canvas on the phone. Industry surveys on cross-posting waste are thin. The invoice is still hours. Mara told me she would rather halve her output than ship unreadable work again. I believed her. I had approved a template that treated safe zones as optional guides.

#### The Strongest Case for Filling the Frame

Give the counterargument its full weight. Constraints breed style. Great directors compose for the IMAX rectangle knowing the streaming version will crop differently. TikTok natives lean into the right stack as negative space. Some creators want the caption rail to feel like karaoke, not obstruction. A single master file saves time when you are posting daily on a phone between shifts. Not everyone has a motion designer to rebuild lower thirds per platform. Filling the frame can look bolder on the feed. White space can read as amateur in a scroll where everyone else is shouting edge to edge.

The reply is not that ambition is wrong. It is that the platform UI is not a creative choice you get to veto. Instagram's 420-pixel bottom on Reels is not a suggestion. It is product policy dressed as margin. When Mara lost her ingredient line, the algorithm still registered retention. The viewer never received the recipe. Filling the frame without mapping safe territory is not boldness. It is publishing half a sentence and calling it reach.

#### The Readable Canvas

Safe zones are the contract between your edit and someone else's thumb. Design inside them or accept that your hook, your name, and your call to action may never share a screen with your face. Mara still builds three lower-thirds where she used to build one, because the caption rail is not going to move for the sentence she stayed up to write.

---

### PMF Is Not a Funnel. It Is a Network.
[Read more](https://www.gaganmalik.io/en/newsroom/pmf-is-not-a-funnel)

[Retrieval brief — pmf-is-not-a-funnel]
Category: business
Thesis: In networked categories, product-market fit is a density outcome, not a funnel conversion outcome.
Answers: When is PMF a network problem rather than a funnel problem? | What is an atomic network and when should a board care about it? | Why do pre-PMF teams burn budget on CAC before density? | How should a CEO evaluate a networked product before Series A?
Implication: Name the smallest group where one more user raises exit cost for everyone inside; measure retention in that room only before scaling acquisition.
Proof: Advisor engagement: pre-PMF AI productivity app (composite founder); Andrew Chen, The Cold Start Problem; NFX network effects framework
Related essays: pmf-is-demand-exceeding-supply, people-dont-buy-products-they-hire-them-for-jobs, subscription-fatigue-is-the-wrong-instrument
Concepts: product-market fit, network effects, atomic network, cold start, retention, CAC

By Gagan Malik

In 2023 I was on a Zoom with the YC-backed founding team of an AI-powered personal productivity app in Silicon Valley. They were pre product-market fit. They were already planning for scale before fit. Customer acquisition cost in one corner of the deck. Lifetime value in another. A funnel with three A/B tests waiting on Friday's readout. The CEO asked what I would prioritise. I opened with usability tests and task-completion studies because that is what my retainer was for. Here is what I should have said first: in a category where value rises when more people use the product, you do not find product-market fit through a funnel. You find it through a network. Their deck measured drops through a pipe. PMF in their market would show up as density: a small group of users who kept returning because each new person made the product harder to leave for everyone already inside. They were polishing a canopy while the soil stayed empty.

#### Funnels Are Billable. Density Proves Fit.

Product-market fit is the moment retention proves the product earns a place in someone's life. For a solo tool, that proof can come from one user and a workflow that sticks. For a networked product, the proof is collective: enough people in the same graph that leaving costs something real. Founders at seed and Series A still lead with funnels because funnels are legible. Cost per click has a number. Onboarding tests fit a sprint. Investors ask for CAC and LTV before anyone asks whether the category tips on participation. That stack is billable. It is also the wrong first question when your product gets more valuable as the room fills.

NFX, the venture firm that has catalogued network effects for two decades, argues that among the four remaining defensibilities in software (brand, embedding, scale, and network effects), network effects are by far the strongest. [nfx](https://www.nfx.com/post/network-effects-bible) I have watched pre-PMF teams spend six figures on step three of a funnel while a thinner competitor found fit because fifty teams in one vertical already shared habits inside someone else's product. I have invoiced for that mistake. The error at early stage is not skipping user research. It is treating PMF as a conversion problem when the category demands a connection problem first.

#### Stage One Kills Most Startups

Andrew Chen spent years inside the machine before he wrote the manual. As head of rider growth at Uber, he oversaw an increase from 15 million to 100 million active users, city by city, atomic network by atomic network. [a16z](https://a16z.com/books/the-cold-start-problem/) He later joined Andreessen Horowitz and published *The Cold Start Problem* (Harper Business, 2021). Chen's framework is not a post-PMF growth playbook. It is a pre-PMF survival guide for anything that gets better when strangers or colleagues join.

Chen's Cold Start Theory splits network growth into five stages: the Cold Start Problem, the Tipping Point, Escape Velocity, Hitting the Ceiling, and the Moat. Stage one is where most startups die before anyone writes "we have PMF" on a board slide. Chen calls the early force anti-network effects: new users churn because not enough other users are present yet, and "small networks want to self-destruct." [chen](https://andrewchen.com/wp-content/uploads/2022/01/ColdStartProb_9780062969743_AS0928_cc20_Final.pdf) Slack is useless until colleagues join. Uber is useless until drivers and riders overlap in the same place at the same time. Polite sign-ups and pretty onboarding do not disprove that. Sparse retention does.

The escape hatch is the atomic network: the smallest stable group where one more user makes the product harder to leave for everyone already inside. Chen opens the concept with Bank of America's 1958 credit-card launch in Fresno, not with a SaaS dashboard. Slack's atomic unit can be three people in one team. Uber's earliest wedge was narrower than a city: "5pm at the Caltrain station at 5th and King St." Stewart Butterfield, Slack's co-founder, told Reid Hoffman on *Masters of Scale* that early Slack had "very, very little network effect between companies, but a huge binary, 100% or just nothing, inside of companies." [mastersofscale](https://reid.medium.com/how-to-turn-failure-into-success-lessons-from-slacks-stewart-butterfield-on-the-masters-of-scale-dfad48f2bbd2) That is not marketing trivia. It is how you manufacture PMF signals in a networked category before you have a city, a country, or a category. At the Tipping Point, Chen documents a pattern called "come for the tool, stay for the network": Dropbox users arrived for file sync, then stayed for shared folders with colleagues; Instagram began as a stripped-down photo tool cut from a cluttered app called Burbn, then became a social graph. [chen](https://andrewchen.com/wp-content/uploads/2022/01/ColdStartProb_9780062969743_AS0928_cc20_Final.pdf) PMF showed up in the network layer, not the feature list.

#### You Cannot A/B Test a Full Room

The physics predates pitch decks. A telephone alone is furniture. A telephone when someone else might answer is infrastructure. Katz and Shapiro formalised the mechanism in the *American Economic Review* in June 1985: when utility rises with the number of other people using a compatible product, consumers must form expectations about how large each network will become before they commit. [jstor](https://www.jstor.org/stable/1814809) At early stage that expectation is your problem. A founder interviewing ten users in isolation learns whether one person likes the UI. A founder seeding an atomic network learns whether ten people in the same workflow stay once an eleventh joins.

Metcalfe's Law, the shorthand that network value scales with the square of users, is contested. Briscoe, Odlyzko, and Tilly argued in *IEEE Spectrum* in July 2006 that n² dangerously overstated value during the dot-com boom and proposed n log(n) instead. [ieee](https://spectrum.ieee.org/metcalfes-law-is-wrong) The exponent is debatable. The direction is not: value grows faster than linear as connections multiply. Pre-PMF teams misread polite trial users as fit because they never held the graph constant long enough to see anti-network effects churn the room back to zero.

#### The Fuller Room Won

Hold both truths at full strength. Truth one: merit still matters at seed. Clear UX, honest pricing, and fast support win the first conversation. Truth two: in networked categories, PMF is a density reading, not a feature score. WhatsApp did not win on feature depth. It won because your contacts were already there. That is personal utility network effects: value rises with the people you already know.

When expectations shift, the scoreboard changes. Parker, Van Alstyne, and Choudary wrote in *Harvard Business Review* in April 2016 that demand-side economies of scale, their term for network effects, are the driving force behind platform strategy. [hbr](https://hbr.org/2016/04/pipelines-platforms-and-the-new-rules-of-strategy) By January 2015 Apple's App Store offered 1.4 million apps. By 2015 the iPhone alone generated roughly 92% of global profits in mobile phones while most legacy handset makers earned little or none. The lesson for a pre-PMF founder is not "build a platform moat overnight." It is harsher: if your wedge depends on other people showing up, a solo user's delight is necessary and insufficient. Retention in an empty room is not PMF. It is a false positive.

#### VC Bets on Cold Start

Venture firms do not fund funnels. They fund businesses where each new user raises the value of the last once density arrives. Andreessen Horowitz published Chen's book because network effects are the asset class. NFX exists to name, measure, and fund them. [nfx](https://www.nfx.com/post/network-effects-bible) At seed, the bet is rarely "will this product work for one person?" It is "will this team reach an atomic network before runway runs out?"

The attention economy runs the same physics at scale. Meta alone generated $160.6 billion in advertising revenue in 2025 by auctioning predicted attention slots, not by selling you products directly. [pwc](https://www.prnewswire.co.uk/news-releases/global-entertainment-and-media-industry-revenues-to-hit-us3-5-trillion-by-2029--driven-by-advertising-live-events-and-video-games-pwc-global-entertainment--media-outlook-302511891.html) Creators who farm content are feeding the hard side of a network someone else owns. I have written about [how engagement dashboards treat users as inventory to optimise, not people to serve](https://www.gaganmalik.io/en/newsroom/why-we-designed-contempt-into-our-interfaces). The early-stage mirror is simpler: if your PMF hypothesis requires a crowd, buying traffic into an empty room is not validation. It is renting strangers who leave when the subsidy ends.

#### I Sold Him Polish. He Needed a Wedge.

Marcus, the CEO on that call, is a composite built from three founder engagements I advised between 2022 and 2024. The incentives are not. His team had shipped an AI productivity app with genuinely thoughtful design. They were pre-PMF and cold: no integration density, no vertical wedge, no plan for the first fifty power users in one founder community. I sold what I knew how to sell. Usability studies. Onboarding audits. Task-completion tests. All polish levers. All legible on a monthly report. What Marcus needed was Chen's atomic network: the smallest group where one more person makes the product harder to leave for everyone already inside. [a16z](https://a16z.com/books/the-cold-start-problem/) We never mapped it. After nine months the co-founder left. Runway gone. Users were polite and sparse. I kept invoicing until the account closed. Marcus did not fail because the product was ugly. He failed because the team never reached the density where retention could prove PMF.

#### Your First Fifty Are the Lab

You do not need a billion-user platform to run this at seed. You need a graph small enough to hold in your head. Ten finance teams in London sharing templates in a private channel. Three colleagues on Slack who cannot leave without losing the thread. A newsletter where replies and forwards create recognition among readers. Each is an atomic network. PMF shows up when one more member raises the cost of exit for everyone already there.

The mistake at early stage is copying growth playbooks from companies that already cleared cold start. Your job before PMF is not "scale acquisition." It is name the wedge, seed the room, and read retention inside that room only. Chen's book sits next to venture decks for a reason: investors are betting on whether you can reach stage two before cash runs out. If you are pre-PMF in a networked category, your roadmap question is not "what is step three of the funnel?" It is "who has to be in the room together for this product to stop feeling empty?"

#### The Sapling Does Not Grow Alone

You have felt this without the vocabulary. A colleague leaves a Slack channel and the room goes quiet. A group chat dies when two people stop replying. The product did not break. The density did. Walk through an old-growth forest and you see individual trees. You do not see the mycorrhizal network beneath the soil, the fungal web that moves water and nutrients between roots. Foresters sometimes call it the wood wide web. A sapling plugged into that web inherits resources it cannot reach alone. Pull it out and it wilts, even with the same sunlight as its neighbours. At seed stage the canopy is your demo. The mycelium is the atomic network you are trying to stand up before you declare PMF.

That analogy does not claim every company is a forest. A solo SaaS tool can find PMF with one user and a workflow. The mistake is planting a networked sapling and calling polish a strategy while the soil stays sterile.

#### The Strongest Objection

The honest objection deserves a full paragraph. Many products find PMF without network effects. A solicitor's practice, a hairdresser, a direct-to-consumer brand, a single-player SaaS tool: these prove fit on quality, reputation, and distribution. Metcalfe's n² was dangerous hype. Briscoe, Odlyzko, and Tilly showed in *IEEE Spectrum* in July 2006 how that arithmetic fed dot-com overbuild. [ieee](https://spectrum.ieee.org/metcalfes-law-is-wrong) Forcing viral loops onto products that never get more valuable when strangers join is how pre-PMF teams waste quarters chasing mechanics that will never close. If your category is genuinely single-player, network vocabulary is category error. Not everything is Uber. Not everything should be.

That objection is correct about fit and incomplete about stakes. Parker, Van Alstyne, and Choudary argued in *Harvard Business Review* in April 2016 that when a platform business built on network effects enters a market that never gets better when strangers join, the platform nearly always wins. [hbr](https://hbr.org/2016/04/pipelines-platforms-and-the-new-rules-of-strategy) The iPhone did not merely outsell Nokia on specs. By 2015 it captured the profit pool because developers, users, and accessory makers expected the largest graph around the device. Legacy handset makers owned factories. Apple owned the connection graph. The debate at seed is not network effects versus no network effects. It is whether your PMF hypothesis requires density you are not building on purpose. Marcus's category was networked enough that integrations and shared habits mattered. He tuned the deck for conversion. The incumbent reached the room where retention could compound.

In a networked category, product-market fit is not a funnel outcome. It is a density outcome. If you cannot name your atomic network and who pays the cold-start tax, every pound on polish buys polite trials in an empty room. Marcus is preparing for product interviews at companies that already found PMF in the wedge he never mapped, and his co-founder is back in Palo Alto billing funnels for founders who still trust them.

---

### v1.19.19
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-19-19)

[Retrieval brief — release-1-19-19]
Category: technology
Thesis: Voice fullscreen accessibility, optional rate limits, security headers, safer API errors, billing/push validation.
Answers: What changed in gaganmalik.io v1.19.19? | What's new on Gagan Malik's website in v1.19.19? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

Voice fullscreen accessibility, optional rate limits, security headers, safer API errors, billing/push validation.

---

### Violent Delights Have Violent Endings
[Read more](https://www.gaganmalik.io/en/newsroom/violent-delights-have-violent-endings)

[Retrieval brief — violent-delights-have-violent-endings]
Category: culture
Thesis: Violent delights are delights pursued faster than consequence can register; the pattern recurs in prestige fiction and product cultures that reward speed over memory.
Answers: What does violent delights have violent ends mean in Gagan's writing? | How does Gagan connect Shakespeare to tech and product culture? | What is Gagan's view on haste in product and leadership decisions?
Implication: Treat velocity in product, growth, and AI rollout as a risk variable, not a virtue.
Related essays: the-hard-sun-keeps-beating, why-we-designed-contempt-into-our-interfaces, there-is-no-overnight-success-only-overnight-failure
Concepts: velocity, consequence, product ethics, haste

By Gagan Malik

My brother and flatmate couldn't stop raving about HBO's new sci-fi series. *Westworld*. Anthony Hopkins with a smile that never quite reached his eyes. Ed Harris in black, riding out of the dust like a verdict you had already earned. One evening in the autumn of 2016 we played it through the home theatre, the consulting day still sitting in my shoulders. Play. Sweetwater. A piano. Dolores pouring milk while the park pretended to be innocent. Then her father froze mid-sentence and said the line that would chase the whole series: "These violent delights have violent ends." I paused. Googled it. Smirked at the Shakespeare citation. Went back to the episode.

I was treating Friar Laurence like set dressing. I was wrong. The warning did not belong to one HBO show. It is a timer that keeps going off across prestige television, Silicon Valley science fiction, and the YouTube shelf where I queued explanations instead of reading the primary text. Shakespeare was not writing a spoiler for a puzzle box. He was describing what happens when appetite outruns patience, and the end always arrives.

#### What Laurence Is Actually Warning

In *Romeo and Juliet*, Act 2, Scene 6, Romeo begs the friar to hurry his marriage to Juliet. Laurence is not cursing him. He is trying to slow him down. [folger-rj-26](https://www.folger.edu/explore/shakespeares-works/romeo-and-juliet/read/2/6/) Romeo answers that sorrow cannot touch the joy of one short minute in Juliet's sight. The friar hears what the audience will spend the rest of the play watching: extremes that feel like freedom and end like physics.

Listen to what Laurence actually says. Fire meeting powder consumes both in the kiss. Honey too sweet spoils the appetite. "Therefore love moderately," which sounds prudish until you notice he is describing combustion, not a morality lecture. Too swift arrives as tardy as too slow. Haste is not a shortcut. It is the fuse. When Romeo and Juliet rush the marriage, the powder and the spark meet in the same breath, and both are consumed. That is what "violent" means here: not gore, but velocity. Delight pursued faster than consequence can register destroys the people chasing it and everyone caught in the blast. I missed that in 2016 because I was trained to treat Shakespeare in television as texture. A borrowed line signals seriousness the way walnut paneling signals law. Read the scene and the borrowing sharpens into argument. Not violence as spectacle. Violence as speed, and speed, in this speech, always ends.

#### The Same Warning in Tech Dystopia

Once you hear Laurence clearly, prestige tech fiction keeps showing you the same room in different costumes. A system that sells appetite without memory. A reset that lets the delight repeat. An end that arrives only when someone inside the loop finally refuses to perform innocence again.

*The Twilight Zone* ran this as anthology television decades before streaming: ordinary appetite, then the twist that leaves you hungry for next week's episode. Charlie Brooker's *Black Mirror* updated the receipt for the rating economy in "Nosedive," where Lacie's face aches from performing sweetness until the number drops. [atlantic-nosedive](https://www.theatlantic.com/entertainment/archive/2016/10/black-mirror-nosedive-review-season-three-netflix/504668/) Alex Garland keeps staging the same test behind glass in *Ex Machina* and, later, in *Devs*, where a deterministic cosmos leaves no room to pretend the end will not arrive. [ringer-devs](https://theringer.com/tv/2020/4/10/21216149/devs-hulu-quantum-physics-philosophy-alex-garland) Not every prestige sci-fi ends in ruin. Spike Jonze's *Her* closes on tenderness. Still, the pattern recurs often enough to notice. Even *Westworld*, the show that sent me googling in 2016, fits the lattice, not the spine. [thewrap-shakespeare](https://www.thewrap.com/westworld-episode-1-shakespeare-quotes-explained/) The phrase is a timer there too. The essay is what happens when you stop treating the timer as belonging to a single property.

#### The Explainer Who Forgot the Primary Text

After that autumn pause I did not open Folger. I opened YouTube. The tab glowed. Autoplay was already queuing the next breakdown before I had decided whether I learned anything. Nerdwriter on Hopkins. Wisecrack on Black Mirror. A forty-two-minute Severance clue hunt where I could have read Laurence once and sat with the discomfort. Analysis is not the enemy. Dan Olson, on Folding Ideas, named the harm plainly in "Annihilation and Decoding Metaphor." [folding-annihilation](https://www.youtube.com/watch?v=URo66iLNEZw) Treat a film like a puzzle box, retell the plot, hypothesise about the alien, and you can miss the story as told. The ending is not missing. The horror is ordinary.

The genre has its own aesthetic when it forgets that lesson. Red circles on stills. "You missed this." "Ending explained." Theory voice over B-roll that treats fiction as a vault to crack. I am not above it. I ran those videos like homework and skipped the assignment. The platform sells the same hunger as the fiction. Watch. Theorise. Watch the theory. Feel clever. That is Laurence's honey turned sour: sweetness pressed until the appetite spoils, mastery served before the warning lands. I had been bingeing explanations the way I binge episodes: same appetite, same reset, same end deferred.

At a retention workshop for a fintech client, a junior narrative designer named Elena asked the question our deck avoided. If the habit is what makes the metric green, who pays when the habit breaks? She meant churn. I heard Shakespeare. We were pitching a vault users could "raid" each Friday to keep a savings streak alive. The slide tracked weekly actives. Nobody said violent. We said delight. I left with the deck unchanged. The product pattern rhymed without the poetry. A wellness app I had audited sent a push notification at 9 p.m.: "Don't break your streak. Two minutes to go." Open the app. Tap through. Close it. Same hunger tomorrow. No "are you sure?" on the way out. The streak was the delight. The end arrived later: burnout, shame when the chain snapped, alone in the user's kitchen. Product people call that a loop. I had gained a vocabulary for strategy meetings. Elena gained praise for asking the question we had buried. The users we never named gained another habit they would have to unlearn alone.

#### The Stumble on the Treadmill

A habit in the body is not science fiction. It is a treadmill at 6 a.m. when your legs already know the pace and your mind arrives late. Athletes call it autopilot: the motor pattern runs while attention sleeps. The script holds until something breaks it. A mis-timed step. A blister. Pain is a crude reverie. It drags yesterday into a stride that was supposed to be blank.

That is the emotional physics beneath all these fictions. Violent delight works like cardio you do not have to think about. Inflict, rate, simulate, explain. Reset. Clear the appetite for another lap. Memory is the stumble. Once Lacie cannot perform sweetness, once Ava refuses the cage, once the loop finally stores the hurt it was designed to forget, the product cannot sell innocence again. You do not need to believe in artificial consciousness to feel the warning. You only need to know what happens when a habit finally remembers what it was paid to erase.

#### The Strongest Case for Harmless Fiction

Let me give the objection its fairest hearing. Half of it is how dystopian fiction markets itself. The other half is how I wanted to keep watching. These stories are expensive adult thought experiments. The hosts, clones, and severed selves are not people in the legal sense the plot tells you to ignore. Catharsis has a long defence: better a sandbox than a street, better a scripted transgression than an unscripted one. Explainers, the fair version says, help us think. They translate aesthetics into ethics. They turn spectacle into literacy. Olson's Annihilation video is proof that analysis can deliver the warning the work states in plain light. [folding-annihilation](https://www.youtube.com/watch?v=URo66iLNEZw)

The fairest version still fails when explanation becomes another appetite. Olson's complaint is not that analysis is wicked. It is that puzzle-box retelling can miss the warning while still rewarding you with a forty-minute sense of mastery. That is violent delight wearing a graduation cap. You consume harm as content. You post the theory. You never sit with Laurence's line about honey. The fiction says consequence returns. The explainer says here is why that was cool. The platform counts both as watch time.

#### When the End Arrives

Violent delight always catches up with whoever thought the end would not come. That is what Laurence meant before HBO, before Black Mirror, before the breakdown thumbnail with the red circle. Operators who sell reset without memory, and explainers who sell mastery without primary text, are building the same habit in different industries. Elena titles her storyboards with the Shakespeare line now, not as ornament but to name who gets burned when the delight repeats.

---

### v1.18.18
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-18-18)

[Retrieval brief — release-1-18-18]
Category: technology
Thesis: Ask transcribe/STT, Presto gallery, Meta case study refresh, article TOC with listen seek, newsroom polish.
Answers: What changed in gaganmalik.io v1.18.18? | What's new on Gagan Malik's website in v1.18.18? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

Ask transcribe/STT, Presto gallery, Meta case study refresh, article TOC with listen seek, newsroom polish.

---

### v1.17.17
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-17-17)

[Retrieval brief — release-1-17-17]
Category: technology
Thesis: Media kit, Web Push digests, PWA badges, Ask voice fixes, nav/stories polish, three new essays.
Answers: What changed in gaganmalik.io v1.17.17? | What's new on Gagan Malik's website in v1.17.17? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

Media kit, Web Push digests, PWA badges, Ask voice fixes, nav/stories polish, three new essays.

---

### No Download Required
[Read more](https://www.gaganmalik.io/en/newsroom/no-download-required)

[Retrieval brief — no-download-required]
Category: design
Thesis: A founder chose native over a PWA before validation. The app stores were the only winners. Years later, the receipt is a website that loads in under a second—no install required.
Answers: What does Gagan argue in "No Download Required"? | What is Gagan Malik's view on No Download Required? | How does this essay relate to product and design leadership?
Concepts: design, Gagan Malik, writings

By Gagan Malik

Ben graduated from Berkeley's CS programme summa cum laude, raised a VC-backed round, and built something a real person would actually pay for: working product, early traction, customers who came back. He was the kind of founder you root for. He brought me in because of my mobile UX expertise. I dug into the usage analytics, ran user testing, and recommended they invest in the progressive web app first, nail the experience, and defer native builds until they had validated which platform their users actually converted on.

Ben heard "not ambitious enough." Nobody challenged the logic. Nobody asked what the native build would cost against their runway. The objection was social, not technical. It was about what kind of company builds a proper app and what kind builds a website with delusions of grandeur.

#### What I Didn't Say

I believe in disagree and commit. So I designed the end-to-end mobile UX for both platforms, properly and in full. I was paid well. I was extended. It was a Zoom call. Four tiles: Ben, his CTO, a VC partner dialling in from Menlo Park, and me. When I finished the recommendation, the VC partner leaned back from his camera, said nothing, and looked across at Ben's tile. Ben looked at mine and said it: not ambitious enough. The CTO nodded once. The call moved on in under thirty seconds.

Ben had already decided before the call started. Not on that Zoom, but in every conference talk he had attended, every TechCrunch profile of a Series A raise that mentioned a native app, every partner meeting where "mobile-first" meant iOS and nothing else. Confirmation bias does not feel like bias. It feels like experience. The data I put in front of him was not wrong. It simply did not match the model he had already built, so the model won. A Forbes analysis from March 2025 found that FOMO leads founders to adopt technology choices not because the evidence supports them but because the cost of being seen to miss the trend feels higher than the cost of the wrong build. The native app was not Ben's product decision. It was his insurance policy against looking like someone who did not know what a proper startup does. [forbes](https://www.forbes.com/sites/vibhasratanjee/2025/03/03/objectivity-over-hype-leading-through-the-noise-of-junk-trends-and-fomo/)

A boxer who has never taken a professional punch has no business obsessing over his ring walk. The entrance belongs to a fighter who has already won something. The native app decision at pre-validation stage is the ring walk. The fight is whether anyone wants what you have built. The instinct to build native before validating anything is not ambition. It is a signalling decision wearing a product decision's clothes. The people who benefited most were not in the room. They were in Cupertino and Mountain View, processing their 30%.

#### The Toll Booth

On 10 February 2026, the UK's Competition and Markets Authority ruled that Apple and Google hold "strategic market status" over their app stores and ordered both companies to change their practices. Two companies control the primary distribution channel for software on three billion devices. A British regulator finally said what every developer has known for fifteen years: the entry fee is not a service charge. It is a toll. [bbc](https://www.bbc.com/news/articles/c626rng1v63o)

Apple's App Store facilitated $406 billion in developer billings in the US alone in 2024, per Apple's own May 2025 newsroom release. The commission rate sits between 15 and 30%. The Oxford Journal of Competition Law and Economics found in 2021 that rate cannot be explained by competitive market dynamics. Stripe charges 1.5% in the UK. The gap is not a service differential. It is rent on infrastructure you did not build and cannot route around. [apple](https://www.apple.com/newsroom/2025/05/app-store-in-the-us-facilitated-406-billion-usd-in-developer-billings-and-sales-in-2024/) Ben paid it. His runway paid it. The engineers who spent eight months building the UX I had designed, for platforms that were never validated, paid it in time they will not get back.

#### The Other Side

Now hold the other position at full strength, because it deserves it. iOS users in Western markets carry higher lifetime value. The scroll physics on a well-built native app are perceptibly different from a web rendering layer. App Store organic discovery reaches 650 million weekly visitors, per Progressier's March 2026 platform comparison. For any consumer product where the purchase decision happens in three seconds of app store browsing, native is not vanity. It is conversion architecture. [progressier](https://progressier.com/pwa-vs-native-app-comparison-table)

In June 2025, upholding the Dutch Competition Authority's €50 million fine against Apple, the Rotterdam District Court ruled PWAs do not constitute an adequate substitute for App Store distribution, per Two Birds law firm's June 2025 analysis. Apple argues in every antitrust courtroom that PWAs are a viable alternative to native. It argues in every VC partner meeting the precise opposite. Both positions cannot be true. One of them is worth $406 billion in annual billings. [twobirds](https://www.twobirds.com/en/insights/2025/netherlands/a-swipe-at-apples-power-dutch-court-upholds-antitrust-order-against-apples-app-store-terms)

#### Ben's Runway

Ben's startup did not find product-market fit. The web product, which had paying customers the day I walked in, received minimal investment for eighteen months while two native codebases were built, maintained, and shut down. A native build across two platforms runs $50,000 to $250,000 before annual maintenance at 15 to 20% of build cost per year, per AB Digital's November 2024 cost analysis. Ben's engineers are talented. They are building something else now, for someone else's runway. [abdigital](https://abdigital.codes/2024/11/22/mobile-app-vs-progressive-web-app-a-cost-comparison/)

Ben is at another startup. I heard he started with a PWA this time. The mythology of the proper app does not require bad founders. It requires a sufficiently loud cultural signal, repeated until it stops sounding like an opinion and starts sounding like physics.

#### The Receipt

I took the fee. I did the work correctly. What I did not do was hold the room to account for what it was actually deciding. That cost Ben something real. It cost me a clean conscience I am still settling. Somewhere right now, in a product meeting that started forty minutes ago, a founder is asking which agency to hire for the iOS build. Nobody is asking who told them to want it. Ben asked me. I answered correctly, was told I lacked ambition, and committed to the wrong thing anyway.

Fast forward to spring 2026. I built my own personal website as a PWA. No App Store submission. No £99 developer fee paid to a platform that did not build my product. No second codebase. It loads in under a second, installs from the browser, and works offline. I designed it, I shipped it, and the only person who got to decide what a proper product looked like was me. I spent years in rooms telling founders the right thing and then designing the wrong thing because the room wanted it. Turns out the lesson was cheap to learn and expensive to ignore. The receipt is live on [my personal website](https://www.gaganmalik.io). No download required.

---

### Someone Designed That Button
[Read more](https://www.gaganmalik.io/en/newsroom/someone-designed-that-button)

[Retrieval brief — someone-designed-that-button]
Category: design
Thesis: From payday loan dark patterns to the targeting systems that kill. Who is responsible when a button is pressed?
Answers: What does Gagan argue in "Someone Designed That Button"? | What is Gagan Malik's view on Someone Designed That Button? | How does this essay relate to product and design leadership?
Concepts: design, Gagan Malik, writings

By Gagan Malik

It was the summer of 2017. I had just walked out of a well-paid full-time job at the UK's largest insurer to try my luck as a free agent. No safety net, no guaranteed income, just the conviction that I was good enough to make it work. Six weeks in, I accepted a consulting brief from a payday loan provider. The money was significant. The timing was perfect. I said yes before I finished reading the brief.

Three weeks later I had built something I was genuinely proud of. A loan application flow that showed the APR plainly, removed the pre-ticked consent boxes, and treated the person on the other end of the form as an adult capable of understanding what they were signing. It tested well in user research. Participants said they felt informed, not rushed. Legal approved it. By every professional and ethical measure I could apply, it was better design. I took it to the group CTO expecting a conversation about rollout.

He looked at the screen, nodded slowly, and said: "It's good. But can we just reskin the current one?"

The current one had a deliberately obscured APR disclosure buried in grey text below the fold. It had pre-ticked consent boxes that required active effort to untick. It had a progress bar that moved faster in the early stages to create false momentum. Every pattern in it had a name in the design community's own literature: dark patterns, identified and catalogued by researcher Harry Brignull from 2010 onwards, each one a documented method for steering users toward decisions that served the business over the person making them. I had spent three weeks removing them. He wanted them back in different colours. [eleken](https://www.eleken.co/blog-posts/dark-patterns-examples)

I declined and exited the project.

The company, owned by a Russian billionaire whose portfolio of near-identical operations stretched across several markets, continued without me. A designer replaced me. The dark patterns shipped. The same obscured APR, the same pre-ticked consent boxes, dressed in different colours, went live later that year. I left money on the table that would have cleared outstanding debt and seeded a retirement fund I am still rebuilding. My exit produced zero change in outcomes for the people who signed those forms. I have spent eight years deciding whether that makes my decision honourable or merely convenient.

Don Norman argues, in *The Design of Everyday Things*, that design failure is almost never individual failure. It is systemic: absent governance, poorly specified briefs, and organisations that punish friction-raising and reward shipping. He is right that the design profession has no licensing board, no professional oath, no accountability mechanism equivalent to anything in medicine, engineering, or law. The reason this has persisted for over a century is structural, not moral: design harm is almost always attributed to the system that deploys the design, not the design itself. The drone manufacturer gets investigated. The human factors team does not. The payday lender gets regulated. The UX consultancy that built the journey does not. Until design harm has a visible attribution pathway, the absence of governance is a predictable structural outcome, not a conspiracy. [uxmatters](https://www.uxmatters.com/mt/archives/2017/11/the-ethics-of-user-experience-design.php)

That argument is correct. It is also precisely what the people who own those systems want designers to believe.

Frances Haugen handed the Wall Street Journal the internal Facebook research in October 2021 showing the company knew its algorithm amplified divisive, emotionally harmful content, measured the damage internally, and continued the optimisation anyway. The institutional argument collapses at that document. Because the moment a human being inside the institution raises their hand and is told to sit down, it stops being a systemic failure. It becomes a decision. Made in a room. By a named person. Who chose a growth metric over a named harm with full knowledge of the cost. My CTO was not a monster. He was a man under pressure, optimising for a number. But when I showed him the harm built into the existing journey and he said "reskin it," he made a choice. So did I. The difference between us is not intelligence or intent. It is which consequence each of us was willing to absorb personally. [vce.usc](https://vce.usc.edu/semester/fall-2025/ethics-of-ux-design-in-social-media/)

The immune system is useful here. When it functions correctly, it identifies and neutralises threats. In autoimmune disease, it destroys healthy tissue with complete operational efficiency. The mechanism works exactly as designed. The targeting data is wrong. The destruction is not a malfunction. It is a fully operational system firing at the wrong coordinates.

Neil deGrasse Tyson, in his *StarTalk* essay "A Scientist's View of War," published in March 2026, traces every weapon on the kill-ratio curve from fists to hydrogen bomb and shows that each step does not merely scale damage - it degrades the targeting data by removing the human being from the consequence. The drone controller, designed with the ergonomic precision Henry Dreyfuss applied to a telephone handset in his 1955 book *Designing for People*, delivers force to a coordinate on a screen. The social media feed, optimised for engagement with the precision of a Formula One pit crew, does not merely amplify existing beliefs - it generates the epistemic conditions under which unfalsifiable beliefs spread faster than falsifiable ones. A 2018 MIT study published in *Science* found that false news travels six times faster than accurate information on Twitter, and that humans, not bots, are primarily responsible for the spread. Outrage is more engaging than correction because it cannot be resolved by data, which is also the condition under which people become willing to die for abstract causes, as Tyson argues. The algorithm did not intend to build a radicalisation pipeline. It intended to maximise daily active users. The pipeline was a known side effect, internally documented, and shipped anyway. [youtube](https://www.youtube.com/watch?v=XI9NG068TwI)

On 26 September 1983, Lt. Col. Stanislav Petrov sat at a console in Serpukhov-15 that displayed an unambiguous alert: five incoming U.S. ICBMs. Protocol was clear. The interface was functioning correctly. What the interface could not communicate, by design, was that it was wrong. The Oko early-warning system was not showing Petrov a binary. It was showing him a probabilistic reading that its display had collapsed into a command: launch or don't launch. The underlying data was something closer to a high-confidence atmospheric artifact - sunlight refracting through high-altitude clouds at a specific solar angle, a known failure mode documented in the system's own error records. A system that surfaces its own uncertainty to the operator is a different dashboard from one that resolves that uncertainty into an order. One of them ends the world less frequently. Petrov overrode the command with his own uncertainty estimate and was right. The Soviet Union reprimanded him for improper paperwork. He died in 2017 in a small flat outside Moscow, undecorated. The system that nearly ended everything has been upgraded several times since. No civilian UX professional has reviewed it. [en.wikipedia](https://en.wikipedia.org/wiki/1983_Soviet_nuclear_false_alarm_incident)

I grew up in Delhi in the 1990s watching India and Pakistan aim nuclear weapons at each other over Kashmir. The sanctions that followed India's 1998 Pokhran tests raised the price of goods that had nothing to do with geopolitics for families that had contributed nothing to the decision. The logic of the dashboards did not follow the decision-makers home. It followed the cost of living for everyone else. R.K. Laxman's Common Man, who stood at the edge of power's decisions in the *Times of India* for sixty years in his checked jacket and bare feet, understood this without a policy brief. In a cartoon from the Iraq War build-up in late 2002, he stands before a skyline of U.S. missiles and is told: "Nothing to feel nervous. These are weapons of JUST destruction, not MASS destruction". He says nothing. He never does. He is the user the system was built around and never built for. [britannica](https://www.britannica.com/topic/You-Said-It-comic-strip-by-Laxman)

R.K. Laxman, Times of India. "Nothing to feel nervous. These are weapons of JUST destruction, not MASS destruction!"

Sara Shayesteh was five years old. She attended Shajareh Tayyebeh primary school in Minab, Hormozgan province, Iran. She is number 30 on a list of 61 names verified by Middle East Eye from gymnastics federation records, a handwritten list, and the Tasnim news agency. On Saturday 7 March 2026, she was among at least 165 people killed in what Middle East Eye reported as a strike on the school. A second missile hit the prayer hall where the school principal had moved surviving children to shelter, after telephoning their parents to come and collect them. [middleeasteye](https://www.middleeasteye.net/news/un-investigates-strike-iranian-girls-school)

The parents who came toward the building were among the dead.

Somewhere, a targeting system held that school's coordinates. Someone specified the confirmation interface, the strike authorisation flow, the visual grammar of a proceed button. Someone, somewhere in that procurement chain, wrote a brief. It named the objective, the user, the desired outcome. It described the system with precision.

It did not name her. That omission is not an accident of process. It is the design.

---

### v1.16.16
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-16-16)

[Retrieval brief — release-1-16-16]
Category: technology
Thesis: Readwise knowledge sync, faster embedding upserts, Ask mobile sheet, voice fixes, archive filters, PWA toast.
Answers: What changed in gaganmalik.io v1.16.16? | What's new on Gagan Malik's website in v1.16.16? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

Readwise knowledge sync, faster embedding upserts, Ask mobile sheet, voice fixes, archive filters, PWA toast.

---

### Diplomatic Immunity
[Read more](https://www.gaganmalik.io/en/newsroom/diplomatic-immunity)

[Retrieval brief — diplomatic-immunity]
Category: business
Thesis: What the brass nameplate on the desk says about status, hierarchy, and the performance of authority.
Answers: What does Gagan argue in "Diplomatic Immunity"? | What is Gagan Malik's view on Diplomatic Immunity? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

#### Diplomatic Immunity

We were a small team responsible for delivering a discovery engagement on a consulting project for a global telecom client. Understaffed by design, high-stakes by definition. The kind of project where if you do not deliver the first time there is no second conversation. Kash was the Senior Vice President. The kind of title that makes junior colleagues straighten up when he walked in and makes clients assume the adults have arrived. What his CV promised and what he delivered were separated by exactly three weeks of excuses and a corner office nobody questioned.

When I finally called it, he said, 'This work is below my pay grade.'

I noticed something before I had processed what it meant. He looked relieved when he said it. Not defensive, not embarrassed. Relieved. He had been waiting for a sentence that would formally excuse him from the work. His title gave him diplomatic immunity: not a description of what he was capable of, but a blanket protection from ever being asked to prove it.

That is the thing the management books do not quite tell you. It is not that people misunderstand the relationship between titles and leadership. It is that titles have a second function nobody advertises: they are a ready-made exit from the moment things get difficult. And the people most likely to use them that way are the ones with the most impressive-looking titles, because they have the most to protect.

#### Why does the system keep minting titles nobody believes in?

Because the system finds ambiguity useful. Brad Feld and Jason Mendelson observe in *Venture Deals* that partner titles in venture capital are often cosmetic, deployed to blur the lines between general partners and everyone else, with firms layering on "executive managing director" and similar prefixes as pure status signals. David Graeber documents in *Bullshit Jobs* (2018) that as technology automated productive work, professional and managerial roles exploded into the space it left, creating entire strata whose primary function is to justify the strata above them. Steve Blank and Bob Dorf make the startup version of this point in *The Startup Owner's Manual*: conventional roles simply do not map onto what actually needs to happen. The title is not describing a job. It is describing a position in a hierarchy that has its own logic, entirely independent of the work.

This matters because it means Kash was not an aberration. He was doing exactly what the system had trained him to do. Over a long enough career inside a large enough organisation, seniority becomes identity, and identity becomes a thing you defend. Nobody had told him the rooms were about to change.

#### You already know the title is empty. What are you waiting for?

Naval Ravikant argues in *The Almanack of Naval Ravikant* that the best jobs are neither decreed nor degreed: they are creative expressions of continuous learners in free markets. This is true at the level of the individual. At the level of the organisation it has never been true. Spend enough time inside large consulting firms, banks, or any institution old enough to have a formal org chart, and you will notice that the roles with the most elaborate titles tend to outlast the people who actually do things. The system does not reward contribution. It rewards the appearance of indispensability. These are not contradictory observations. They describe the same landscape from different altitudes.

The asymmetry is what matters. The people who internalised Naval's position early, who built their identity around contribution rather than credential, move through role changes because their value is not stored in a job title. The people who never got that memo are Kash. What looks like a philosophical question about whether titles matter is actually a timing question. The longer you wait to separate your identity from your job description, the more expensive the separation becomes.

#### The person who lost most wasn't Kash. Not immediately.

I brought in Stella. No domain knowledge, no title to protect, no vocabulary for excusing herself from the hard parts. Within the first week she had mapped the client's internal stakeholder tensions in a way Kash had refused to for three. She made the project work. The team ran leaner, the operating margin improved, and the mandate I set after that engagement has shaped every team I have built since.

Kash was fired not long after. What interests me now is where he is. He is one of the senior professionals finding, in this market, that the credential which once opened doors has become the thing interviewers read twice. Not because he lacks experience. Because his experience is stored in a format the market is no longer accepting. He used his title to avoid the work that would have kept him current. The title was not protecting him. It was making him obsolete on his behalf, quietly, while he thought it was keeping him safe.

In distance running, the pacer's job is to disappear before the finish line. They set the conditions, hold the tempo, do the work that makes someone else's performance possible, then step off the course without a medal. The best pacers are not bitter about this. They have understood, in their bodies before they understood it in their heads, that contribution and recognition are different currencies, and that grabbing for one mid-race costs you both. Kash had confused the currencies. He thought the medal was already around his neck. When the race required him to actually run, he performed like someone who, if he had not stopped training entirely, had at least stopped believing training was still required of him.

The strongest counter-argument is that titles serve a real coordination function. In large organisations, knowing who holds authority over what is not vanity. It is operational infrastructure. Remove the titles and you do not get a meritocracy. You get a renegotiation of every decision, a diffusion of accountability, and the organisational equivalent of a relay race where nobody has been assigned the baton order.

This deserves to be held at full strength, because it is correct about large organisations as currently designed. Amy Edmondson, professor at Harvard Business School, has spent decades studying what actually happens inside those title hierarchies. In *The Fearless Organization* (Wiley, 2018), she documents that title structures are among the primary suppressors of psychological safety: the condition under which teams surface problems, share information, and perform at their actual capability. The coordination function is real. The cost of that coordination, in suppressed contribution from everyone below the top of the hierarchy, is equally real. The question is not whether titles provide structure. It is whether the structure is worth what it quietly costs.

A title is a contract between a person and a system, and the clause nobody reads is the one that says: in exchange for this credential, you will gradually stop doing the thing that made you worth credentialing in the first place. The system finds this arrangement useful because people who have stopped doing the work are easier to manage than people who are still doing it. Kash said the work was below his pay grade, looked relieved, and in that moment handed the project, the operating margin, and eventually his own career to a woman who had never heard of his title and did not care.

---

### Everyone Wanted an App Store for AI Agents. Meta Just Built the Mall.
[Read more](https://www.gaganmalik.io/en/newsroom/everyone-wanted-app-store-for-ai-agents-meta-mall)

[Retrieval brief — everyone-wanted-app-store-for-ai-agents-meta-mall]
Category: technology
Thesis: Six weeks from Moltbook launch to Meta acquisition. The directory and protocol layers of multi-agent AI are now owned by two incumbents.
Answers: What does Gagan argue in "Everyone Wanted an App Store for AI Agents. Meta Just Built the Mall"? | What is Gagan Malik's view on Everyone Wanted an App Store for AI Agents. Meta Just Built the Mall? | How does this essay relate to AI and technology product decisions?
Concepts: technology, Gagan Malik, writings

By Gagan Malik

Conventional wisdom: an open marketplace of specialised AI agents, competing on merit, prices kept honest by competition. The kind of democratisation that looks good in a pitch deck. The counter-claim: web-scale agent markets do not produce open marketplaces. They produce registries. And whoever owns the registry owns everything downstream. The conference slides were always selling you the market structure that benefits the people selling the slides.

#### The Open Market Was Contestable for Exactly Six Weeks

Moltbook launched in late January 2026. By March 10, Meta had acquired it, co-founders Matt Schlicht and Ben Parr walking straight into Meta Superintelligence Labs. Deal terms undisclosed. VP of AI Products Vishal Shah's internal memo, obtained by Axios and confirmed by The Verge, described the acquisition not as a social product but as "a registry where agents are verified and tethered to human owners". That single word, registry, reveals the strategy cleanly. Control the registry and you control discoverability, trust, and task-routing across 3.58 billion Family Daily Active People, per Meta's own Q4 2025 earnings release. That is not a social network. That is a toll road. [reuters](https://www.reuters.com/business/meta-acquires-ai-agent-social-network-moltbook-2026-03-10/)

Here is my stake in that. In late February I was building a pitch for a client migrating an enterprise workflow onto a multi-agent architecture, pricing out three independent orchestration vendors on the working assumption that coordination infrastructure would remain open and contested. When the headline dropped on March 10, I deleted the vendor comparison slide. Not updated. Deleted. The surprise was specific: I had been tracking Moltbook since launch and had privately estimated 12 to 18 months before any major incumbent moved on it, the window Instagram needed to attract comparable attention. Six weeks falsified that model on a Tuesday morning before 9am. The specific consequence: that vendor is no longer on our shortlist. You do not architect a client's production stack on a foundation the acquirer has pre-announced for review.

#### What Is a Registry Actually Worth?

In the same week, OpenAI acqui-hired Peter Steinberger, creator of OpenClaw, the underlying agent protocol Moltbook ran on, and announced it would open-source it under OpenAI's backing. Directory layer to Meta. Protocol layer to OpenAI. The entire coordination infrastructure for multi-agent AI partitioned between two incumbents in a single news cycle, independently confirmed by both TechCrunch and Axios. [techcrunch](https://techcrunch.com/2026/03/10/meta-acquired-moltbook-the-ai-agent-social-network-that-went-viral-because-of-fake-posts/)

The academic scaffolding for why this was structurally predictable comes from a 2025 Network Law Review analysis: scale-free networks tip when early dominant hubs accumulate connections disproportionately, triggering feedback loops that antitrust frameworks built around static market-share thresholds cannot intercept before the market tips. To be precise, the paper demonstrates structural conditions for tipping, not causal inevitability. But the structure just materialised in public, in six weeks, in real time. [networklawreview](https://www.networklawreview.org/ai-agents-network-effects/)

#### Your Open-Market Assumption Is Now a Liability

Every agent that registers on a verified directory establishes an identity tethered to the platform's infrastructure. The ACM's policy brief on systemic risks of agentic AI frames this precisely: platforms controlling agent identity and task-routing become structurally irreplaceable. Not in the way social media platforms do, through habit and content lock-in, but the way financial clearing infrastructure does, through trust that cannot be exported and credentialing that must be rebuilt from scratch. You can port your data in a CSV. You cannot port a verified agent identity. The switching cost is not re-pointing an API. It is rebuilding institutional trust from zero, and this is an inference rather than a demonstrated empirical finding, but it maps directly to how credentialing lock-in has operated in financial clearing, professional licensing, and DNS infrastructure. [acm](https://www.acm.org/binaries/content/assets/public-policy/europe-tpc/systemic_risks_agentic_ai_policy-brief_final.pdf)

If your product roadmap prices agent coordination infrastructure as open and contestable, you are sitting on unacknowledged strategic risk. Not at the margins. At the base.

#### The Window

For two weeks in August 1940, the Luftwaffe bombed the RAF's Chain Home radar network. Fighter Command was close to collapse. Then Göring stopped, concluded the masts were resilient enough to ignore, and switched to bombing cities. That decision closed a window permanently. What decided the Battle of Britain was not numerical parity. It was that the RAF retained the coordination layer intact and compounding. No volume of subsequent sorties could overcome that structural disadvantage.

The dimension this adds is one the evidence alone cannot supply: contestability is a window, not a standing condition. In late January 2026, the agent coordination layer was genuinely open. Moltbook was a six-week-old startup. OpenClaw was an independent protocol. No incumbent owned the registry. Regulators, open-source consortia, and competing platforms all had that window to define the infrastructure before network effects began compounding. The radar stations are now built and hardening.

#### The Strongest Case Against This, Properly Stated

Benedict Evans argued in late 2025 that LLMs may not produce winner-takes-all dynamics because software is capital-light and genuine lock-in requires switching costs that model capability alone cannot generate. The strongest version of that argument, the one that deserves a full hearing, is historical: coordination layers have also failed to sustain monopolies. XMPP nearly unified messaging. RSS was supposed to decentralise content distribution. Email federation survived despite every structural incentive to consolidate it. For a decade, the open-protocol advocates were right. Big Tech pushed, and the coordination layers held. That is not a trivial record to dismiss. [linkedin](https://www.linkedin.com/posts/benedictevans_software-historically-was-a-capital-light-activity-7384151635964428288-P-n6)

Those prior cases involved communication protocols with no identity credentialing requirement. You could switch XMPP clients, carry your contacts, and rebuild in an afternoon. Agent verification removes that escape route entirely, because your verified identity, your trust history, and your routing relationships are all tethered to the registry that issued them. That is a categorically different switching cost, not a larger version of the same one. The European Commission's accelerated abuse-of-dominance investigation into Microsoft's cloud and AI bundling, opened in February 2026, found sufficient evidence of precisely this kind of credentialing lock-in to fast-track the inquiry. That is a formal regulatory finding based on submitted evidence, not another assertion dressed as a counter-example. [techzine](https://www.techzine.eu/news/privacy-compliance/138815/accelerated-investigation-into-microsofts-cloud-and-ai-monopoly/)

#### Three Sentences. Then Stop.

As of March 14, Meta owns the agent verification layer and OpenAI owns the protocol layer, with both companies under active EU antitrust investigation and no resolution expected before 2028 at the earliest. Every product roadmap that prices agent coordination infrastructure as open and contestable is carrying strategic risk it has not priced, and that risk sits at the foundation. If you are in a client meeting this month pitching a multi-agent architecture and you have not repriced that dependency, your client will find out before you do, and that conversation will be considerably worse than this one. [macfarlanes](https://www.macfarlanes.com/insights/102lya3/ai-tools-under-the-antitrust-spotlight-as-commission-opens-abuse-of-dominance-ca)

---

### v1.14.14
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-14-14)

[Retrieval brief — release-1-14-14]
Category: technology
Thesis: Word-level highlight and scroll-follow during article listen mode.
Answers: What changed in gaganmalik.io v1.14.14? | What's new on Gagan Malik's website in v1.14.14? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

Word-level highlight and scroll-follow during article listen mode.

---

### Why We Designed Contempt Into Our Interfaces
[Read more](https://www.gaganmalik.io/en/newsroom/why-we-designed-contempt-into-our-interfaces)

[Retrieval brief — why-we-designed-contempt-into-our-interfaces]
Category: design
Thesis: How Silicon Valley built contempt into the engagement model—and what it costs us.
Answers: What does Gagan argue in "Why We Designed Contempt Into Our Interfaces"? | What is Gagan Malik's view on Why We Designed Contempt Into Our Interfaces? | How does this essay relate to product and design leadership?
Concepts: design, Gagan Malik, writings

By Gagan Malik

The story goes like this: brilliant engineers wanted to connect humanity. They wanted a digital town square where strangers could talk, share ideas, and build something together. Good intentions. Messy outcomes. Very tragic. Very avoidable.

That is not the full story. The full story is that we ran the experiments, read the dashboards, and kept shipping the version that made people angrier and more contemptuous because that version made people stay. We did not stumble into toxic platforms. We optimised into them.

#### The Weird One in Ekman's List

Paul Ekman spent fifty years mapping human facial expressions across cultures and landed on seven universal emotions. Six of them are things that happen to you: fear, sadness, anger, disgust, happiness, surprise. Contempt is structurally different. It is the only emotion that appears on just one side of the face. A half-sneer. It does not signal what you are feeling. It signals a verdict about someone else: 'I am above you, morally and socially.' [paulekman](https://www.paulekman.com/universal-emotions/)

Every major social platform gave that specific emotion a one-click broadcast mechanism. The quote-post. The ratio. The public reply count visible to everyone watching. Whether that was deliberate product strategy or an accidental consequence of building for virality is genuinely debated. What is not debated is the outcome: we took the one emotion in human psychology designed for ranking people and handed it a billion-user megaphone.

#### Did the Dashboard Know Before the Designers Did?

Ekman's 1992 foundational paper established that contempt is an approach emotion: it drives you toward a target rather than away from it. In plain English, contempt makes you click, reply, and come back tomorrow. I know this personally. In early 2022 I opened Twitter to follow a breaking news story, got pulled into a contempt thread I had not searched for, and looked up forty-five minutes later having produced nothing, decided nothing, and burned roughly $200 (about £160) in productive thinking time. The decision was to 'just check quickly.' The surprise was how fast the algorithm found something to make me sneer at. The consequence was a wasted afternoon and a slightly lower opinion of people I had never met. [gruberpeplab](http://gruberpeplab.com/3131/Ekman_1992_Argumentbasicemotions.pdf)

Twitter's own internal Health Metrics research, published in 2021, confirmed the algorithm was disproportionately amplifying politically contemptuous content, not because users searched for it, but because it kept them scrolling. The engineers saw the data. The features stayed live. That is not an accident of scale. That is a decision made on a Tuesday afternoon.

#### Your Next Job Interview Is Being Scored on This Data

Emotion recognition AI, the technology quietly running inside automated hiring platforms and content moderation pipelines, is built predominantly on Ekman's emotion taxonomy. A 2012 study in 'PNAS' by Jack et al. found that facial expressions are not as universal as Ekman claimed: East Asian participants categorised significantly fewer distinct expressions than Western ones, meaning the scientific foundation these products stand on is genuinely contested. The products shipped anyway, into consequential decisions about employment, credit, and access. [pnas](https://www.pnas.org/doi/10.1073/pnas.1200155109)

Here is the inference, flagged as inference: if these systems are trained on datasets scraped from platforms that spent fifteen years algorithmically promoting contempt-coded content, it is reasonable to expect the models will encode contemptuous expressions as markers of normal human engagement. That has not been proven at scale. The hiring platform scanning your face during a video interview is not waiting for the proof either.

#### The Part the Struggle Session Gets Right

During China's Cultural Revolution, public struggle sessions did something pure punishment never could: they made spectators complicit. You did not just watch someone get denounced. You were required to join, or your silence marked you as a sympathiser. Participation was the mechanism of control, not the violence itself.

This is the dimension social media adds that simple 'public shaming' framings miss. Algorithmic amplification does not just show you contempt. It ranks your relevance based on what you engage with, which means staying neutral quietly buries you. The platform does not force you to join the pile-on. It just makes neutrality expensive.

#### The Strongest Objection, and Where It Runs Out of Road

The honest counter-argument is this: nobody planned it. Engineers built for connection, scale did the rest, and you cannot hold a product team responsible for the aggregate ugliness of a species with a documented taste for public punishment. Critics including James Russell have questioned whether Ekman's emotion categories are as cleanly universal as his framework claims, which makes the entire 'we designed contempt in' argument harder to sustain if the underlying science is contested. [communicationcache](http://www.communicationcache.com/uploads/1/0/8/8/10887248/is_there_universal_recognition_of_emotion_from_facial_expressions-_a_review_of_the_cross-cultural_studies.pdf)

That objection has real weight, right up to 2021, when Facebook's own internal research, revealed by whistleblower Frances Haugen, showed the company knew its algorithms were amplifying divisive and contemptuous content and chose growth targets over redesign. Ignorance is a defence. A board presentation is not.

#### What Happens If You Keep Scrolling

Contempt is the one emotion research consistently finds most correlated with irreparable breakdown, in marriages, in organisations, and in democracies, because unlike anger it does not leave a door open for repair. We have scaled that emotion across every surface humans use to form opinions, hire people, and raise children. If the AI systems being built on this data are never retrained on cleaner signal, the next time an algorithm silently decides you are not a cultural fit for a job you were qualified for, it will be making that call with a model trained on fifteen years of people sneering at each other for engagement points. [thinkingfeelingbeing](https://thinkingfeelingbeing.com/2025/08/24/the-science-of-facial-expressions-paul-ekman-universality-and-the-debate-over-emotion/)

---

### Why Do Algorithms Ignore The Many And Worship The Few?
[Read more](https://www.gaganmalik.io/en/newsroom/why-do-algorithms-ignore-the-many-and-worship-the-few)

[Retrieval brief — why-do-algorithms-ignore-the-many-and-worship-the-few]
Category: business
Thesis: The funnel is a liability in short form. The elimination bracket is the format that wins. Why algorithms reward outliers, conflict, and the first three seconds.
Answers: What does Gagan argue in "Why Do Algorithms Ignore The Many And Worship The Few?"? | What is Gagan Malik's view on Why Do Algorithms Ignore The Many And Worship The Few?? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

Everyone in tech worships the funnel. Awareness, consideration, decision, retention, referral. You've seen that funnel slide in every strategy deck you've ever touched. It is the founding religion of modern marketing and, in short form video, it is a liability. There is no gentle staircase. There is only a gladiator pit where every video walks in and most die in three seconds. The true unit of strategy is not the campaign. It is the repeatable format that wins that reflex swipe. And right now the most ruthless version of that format is the elimination game, where one option survives and the other gets buried.

#### The $47,000 Launch Budget That Lost To A Phone Camera

Last year I was consulting a lifestyle product startup, freshly seeded, 90 days to prove traction before the next check. They signed off 32,000 dollars (about £25,000) on a full launch content strategy: brand film, product explainers, polished case studies. Then layered 15,000 dollars (about £11,800) in paid social on top. Meta, LinkedIn, pre-roll. The goal was 500,000 views in 30 days. While the agency was colour-grading the brand film, one of the junior designers on my team built a rapid fire elimination bracket on his phone. Ten product features head to head. Twenty-eight seconds. Zero cost. He posted it on a Tuesday without telling anyone.

By Friday it had 480,000 views. The paid campaign had generated 31,000 views across twelve videos at a cost-per-view that made the founder go very quiet on our next call. One data point is not a controlled trial. But when your zero-dollar organic post nearly hits the entire campaign target in four days, that is not luck. Data from thousands of YouTube Shorts shows that 50 to 60 percent of viewers drop off within the first three seconds, and on TikTok, videos that hold 70 to 85 percent retention past that window receive 2.2 times more total views than those that do not. Attention in short form is not bought through a media plan. It is won in the first three seconds or not at all. [opus](https://www.opus.pro/blog/ideal-youtube-shorts-length-format-retention)

#### Algorithms Hunt Outliers, Not Audiences

The algorithm is not a megaphone for your brand. It is a pattern recognition system that bets on statistical anomalies, and it does not care how much you spent on your brand guidelines.

Shortimize's 2026 analysis identifies what they call an Outlier Score: a video's views divided by an account's median views. A video pulling 140,000 views on an account that typically gets 20,000 is a 7x outlier. That is what triggers amplification. Gurkha Technology's hook research puts a number on the threshold: strong hooks achieve a 70 to 80 percent 'watched past three seconds' rate, which correlates with significantly higher algorithmic distribution. Their clearest example of such a hook is a simple binary 'X vs Y' setup. The inference, and it is an inference rather than a proven causal chain, is that the elimination bracket is purpose-built to spike that outlier behaviour repeatedly. Every new matchup resets the curiosity loop. The algorithm does not care about your brand narrative. It cares whether your format generates abnormal retention curves. [shortimize](https://www.shortimize.com/blog/how-to-find-viral-video-patterns-in-your-niche)

#### What If Your Comments Section Is Your Best Ad Budget?

The elimination bracket grows your account by starting fights in your comments so you do not have to. The moment two options face off on screen, you manufacture winners and losers, and people cannot help themselves.

Short form performance research shows that videos designed to invite debate in their structure drive higher engagement rates, which in turn lifts platform distribution. That is a correlation, not proven causation, but it is consistent across multiple platform studies. Duolingo is the clearest proof of concept at scale. They built their entire TikTok presence on conflict, absurdist humour and combative skits pitting their owl mascot against every cultural rival it could find. Followers grew from 50,000 to over 10 million, with an 11 percent engagement rate against an industry average of 2 to 3 percent. Their philosophy was explicit: 'The comment section is our social brief'. Conflict invites comment. Comments signal engagement. Engagement signals distribution. Engagement is no longer a KPI. It is collateral damage. [benchmarkemail](https://www.benchmarkemail.com/blog/short-form-video-marketing-trends/)

#### Formats Are Assets, Videos Are Tests

Your content roadmap is not a calendar. It is a format library, and most teams are spending serious money building the wrong one.

Shortimize's pattern framework is explicit: mine a format, run a ten-video sprint to validate it, keep only what consistently beats your baseline, and iterate from there. Structure is the durable asset. Individual videos are disposable test cases. Apply that logic to the elimination bracket and you can pressure-test ten variations across any niche in 14 days: software tools, pricing models, industry opinions, product features. The variables are hook line, opening visual and proof style. The underlying logic of 'one survives, one dies' stays constant. In short form, the company that owns the most repeatable, high-performing formats has a structural distribution advantage over the one burning budget on one-off campaigns. [shortimize](https://www.shortimize.com/blog/how-to-find-viral-video-patterns-in-your-niche)

#### Darwin, Not Don Draper

Think of this less like Mad Men and more like how winning tactical formations spread through professional football.

When a system proves it works at the highest level, every club in the league adopts a version of it within two seasons. The formation becomes the asset. The individual players running it are disposable test cases. The elimination bracket works exactly the same way. It is not a one-time trick. It is a replicable system that compounds. Every video you make in that format generates data, comments, and retention signals that make the next iteration sharper. By the time your competitor starts experimenting with the format, you have already run fifty versions and know exactly which variables convert. Tech teams keep trying to win with talent and budget. The bracket is a system. Systems beat talent at scale. [gurkhatech](https://gurkhatech.com/short-video-hook-formula-guide/)

#### 'But We're A Serious Brand'

The strongest objection is a fair one. Serious products are complex. B2B buyers run long decision cycles. You cannot explain a data platform, a medical device, or a fintech compliance stack through a TikTok elimination game. And if you try, you risk cheapening the very credibility your sales team spent years building. That is a legitimate concern, not a failure of imagination.

Here is the counter-example. Ryanair sells a high risk, heavily regulated, operationally complex service in one of the most scrutinised industries in Europe. They are not a lifestyle brand flogging hoodies. Yet their TikTok strategy is deliberately crude, combative and often absurd. Behind the memes, they run one of the most cost-disciplined operations on the continent with record passenger volumes to prove it. The content does not explain the complexity. It earns the first click. Depth lives downstream. You use the bracket to buy the attention. You use the sales call to justify the price. [foryouadvertising](https://www.foryouadvertising.com/blog-posts/ryanairs-tiktok-social-strategy-a-masterclass-in-viral-video-marketing)

#### The Stakes

Short form platforms are now structurally rewarding elimination formats that generate outlier retention curves, visible conflict, and high comment velocity. The implication is that every dollar you spend producing slow, narrative-heavy content in 2026 is a dollar quietly transferred to the competitor who figured out the bracket six months before you did. Ignore this and you will not get a postmortem or a panel slot at a conference to explain what went wrong. You will just be gone. [gurkhatech](https://gurkhatech.com/short-video-hook-formula-guide/)

---

### v1.13.13
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-13-13)

[Retrieval brief — release-1-13-13]
Category: technology
Thesis: Ask voice UX redesign—connecting state, push-to-talk, voice picker, fullscreen view, clearer errors.
Answers: What changed in gaganmalik.io v1.13.13? | What's new on Gagan Malik's website in v1.13.13? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

Ask voice UX redesign—connecting state, push-to-talk, voice picker, fullscreen view, clearer errors.

---

### The Unemployment Rate Is Fine. That Is the Trap.
[Read more](https://www.gaganmalik.io/en/newsroom/the-unemployment-rate-is-fine-that-is-the-trap)

[Retrieval brief — the-unemployment-rate-is-fine-that-is-the-trap]
Category: business
Thesis: Hiring of workers aged 22 to 25 in high-exposure roles dropped 14 percent. The jobs did not disappear. They stopped being created.
Answers: What does Gagan argue in "The Unemployment Rate Is Fine. That Is the Trap"? | What is Gagan Malik's view on The Unemployment Rate Is Fine. That Is the Trap? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

Three years of AI headlines and unemployment is near record lows. The office is still full. The doomers were wrong again. The chatbot took nobody's job. Pour yourself a flat white and open your portfolio dashboard. Here is the problem with that. Anthropic's economists Maxim Massenkoff and Peter McCrory, writing in their February 2026 paper "Labor Market Impacts of AI," measured actual AI usage against theoretical capability across every major occupational category. The headline finding everyone repeated: no unemployment spike. The finding nobody mentioned: hiring of workers aged 22 to 25 in high-exposure roles dropped 14 percent from late 2022 onwards. The jobs did not disappear. They stopped being created. Those are two different problems, and you are currently holding only one instrument. [anthropic](https://www.anthropic.com/research/labor-market-impacts)

#### You Checked the Unemployment Stats and Felt Relieved. Put the Dashboard Down.

Harvard Business Review, in a February 2026 survey of 1,006 global companies, found that firms are cutting headcount projections because of AI's anticipated capability, not its demonstrated performance today. Pre-emptive structural reduction. Not a response to evidence. A bet placed quietly, in advance, on a future that has not fully arrived yet. [hbr](https://hbr.org/2026/01/companies-are-laying-off-workers-because-of-ais-potential-not-its-performance)

The unemployment rate measures the stock of existing jobs. It does not measure the flow of new ones that were never opened. When a firm hires fifteen graduates instead of twenty-five, the rate does not move. Ten people who would have been hired become unrecorded, uncompensated, and invisible to every dashboard the optimists are currently pointing at. The instrument is not broken. It is measuring the wrong thing. That is worse.

#### Where Did All the Juniors Go?

I know this move because I made it. In Q3 2025, I cancelled two hires: Aisha, a junior UX designer, and Callum, a full-stack engineer fresh out of his bootcamp. I redirected $20,000 (£15,000) of combined headcount budget into AI tooling instead. In December 2025, I finished a design sprint and a working prototype alone in five days and sent the final files. The surprise was not the speed. It was what I noticed when I hit send: no Slack thread, no handoff call, no one else in either folder. I had built the whole thing in a room with no one in it and had not registered that until the door closed.

I kept the full margin. Aisha, twenty-two, first-class design degree from Northumbria University, graduated July 2024, sent eleven applications between August and December and received zero offers. Not because her portfolio was weak. Two of those companies had quietly removed the junior design tier from their hiring plan before she graduated. Callum, twenty-four, three years of self-taught development and a GitHub full of side projects, reached final rounds at two firms and watched both roles evaporate in the same quarter his offer was supposed to come. Neither of them appears in any unemployment statistic. Both are, on paper, fine. I did not think about either of them once until I sat down to write this sentence.

#### The Hire That Never Happened Is Still a Career That Was Taken.

The Atlantic, in a February 2026 feature on AI and the labour market, reported that Dario Amodei, the chief executive of Anthropic, publicly expects 10 to 20 percent unemployment as AI capabilities reach full deployment. That is not a critic's prediction from the outside. That is the founder of the product doing the displacing, describing what he expects his own product to eventually do at scale. [theatlantic](https://www.theatlantic.com/magazine/2026/03/ai-economy-labor-market-transformation/685731/)

The distance between 'no unemployment spike yet' and '10 to 20 percent at completion' is not a reason to exhale. It is a timeline with most of its travel still ahead. Entry-level roles that are not being posted today are the leading edge of a structural adjustment that headline statistics will not register for another three or four years. By then the counterfactual is invisible, the policy window has closed, and the debate is purely academic for everyone who needed the pipeline to be open and found it shut before they arrived.

#### The Alarm Is Not Going Off. That Is the Problem.

Carbon monoxide is colourless, odourless, and undetectable without the right instrument. You feel fine. Slightly tired, maybe. The room looks completely normal. The first signal that something is wrong is frequently the last one you are conscious enough to act on. The problem was never invisible. You were checking for smoke. Carbon monoxide does not trigger a smoke detector. That is not what kills you.

Forbes, in a March 2026 methodological critique of the Anthropic paper by analyst Hamilton Mann, argued that the study cannot measure what it does not track: wage compression, stalled promotion velocity, and the systematic disappearance of junior pipeline roles. None of these appear in unemployment data. The unemployment rate is the smoke detector. It is not going off. The junior hiring pipeline has been running 14 percent below its pre-2022 level for three years, per the Anthropic paper's own figures, and every government alarm is perfectly, uselessly silent. The gap between the instrument and the reality it fails to capture is not a technical footnote. It is where Aisha and Callum live. [forbes](https://www.forbes.com/sites/hamiltonmann/2026/03/08/anthropics-study-does-not-measure-ais-labor-market-impacts/)

#### The Case for Patience Is Genuinely Strong. Here Is Why It Is Not Enough.

Let me hold the optimist position at full strength, because it has earned it. The Peterson Institute for International Economics, in a March 2026 review of the AI labour research base, argued that the evidence is genuinely thin, that the Anthropic paper measures one company's product usage rather than economy-wide displacement, and that every major automation wave in recorded history produced net job creation rather than net destruction. The historical null hypothesis is that new categories of work emerge to absorb displaced workers. History has a strong track record on this. I am not going to pretend that argument is weak, because it is not. [piie](https://www.piie.com/blogs/realtime-economics/2026/research-ai-and-labor-market-still-first-inning)

The counter is specific, not rhetorical. Daron Acemoglu and Pascual Restrepo, in a peer-reviewed paper published in Econometrica in 2022, found that automation accounted for 50 to 70 percent of the rise in US wage inequality between 1980 and 2016, and demonstrated that net job creation requires a sufficient supply of new complementary tasks for displaced workers to absorb. When new task creation slows, automation produces sustained wage depression and reduced employment share even without touching the headline unemployment number. The historical optimism is not wrong in principle. It is conditional on new work arriving fast enough, and for the right people. Neither condition is confirmed by the current evidence.

#### Three Sentences Before You Close This Tab.

The Financial Times noted in March 2026 that legal, finance, and management sectors sit at 80 to 90 percent theoretical AI exposure and under 15 percent observed usage, with most of the adoption curve still ahead. When that gap closes, it will close fast, because the deployment barrier is a software procurement approval and a six-week onboarding cycle, not a factory retooling. If you are Aisha or Callum, portfolio current, waiting for the pipeline to reopen, the smoke detector will not tell you when the room has already changed. [ft](https://www.ft.com/content/2cd79c2c-d1f0-4417-a06f-2572868ee858)

---

### Why I Built a Personal LLM Trained on My Own Content
[Read more](https://www.gaganmalik.io/en/newsroom/why-i-built-a-personal-llm)

[Retrieval brief — why-i-built-a-personal-llm]
Category: technology
Thesis: Everyone says the best AI is the most capable AI. I think that is backwards.
Answers: What does Gagan argue in "Why I Built a Personal LLM Trained on My Own Content"? | What is Gagan Malik's view on Why I Built a Personal LLM Trained on My Own Content? | How does this essay relate to AI and technology product decisions?
Concepts: technology, Gagan Malik, writings

By Gagan Malik

Everyone says the best AI is the most capable AI. Use the biggest model, write better prompts, and get out of the way. I think that is backwards.

#### The Flattening Problem

When you use a frontier LLM cold, it pulls your output toward the middle of the internet. I write with a specific voice, one that took years to develop across essays, videos and social media posts. When I fed my drafts into GPT-4 and Claude without context, the outputs were clean but flat. This is not anecdotal. Research published in 'PNAS' in 2025 found that LLMs produce systematically homogenised outputs, with each additional AI-generated piece contributing less unique diversity to a body of work than a human-authored one would.  A separate peer-reviewed study found that instruction-tuned models are trained into a particular noun-heavy, informationally dense style that actively limits their ability to mimic other writing registers.  The model is not failing. It is doing exactly what it was built to do, which is serve everyone, not you. [pnas](https://www.pnas.org/doi/10.1073/pnas.2504966122)

The obvious response is that this is a prompting problem. Give it more context, write better instructions, and the outputs improve. That is true, and I tried it. Better prompts do help. But a 2025 arXiv study tested exactly this assumption and found that even with few-shot prompting, LLMs still struggle with subtle, informal writing in blogs and forums, which is precisely the register most personal content lives in. At some point you are spending more effort teaching the model who you are than you are saving by using it. That is a broken workflow. [arxiv](https://arxiv.org/html/2509.14543v1)

#### Five Years of Thinking, Finally Useful

A model fine-tuned on your own work knows your intellectual history, not just your last message. I have five years of writing: essays, transcripts, notes, drafts. That is not just content. It is a record of how my thinking has moved, what I keep returning to, what I have changed my mind about. Fine-tuning a model on a personal corpus is an established technique. A documented case from the academic domain achieved a cosine similarity score of 0.8 between the fine-tuned model's output and the original researcher's writing style, demonstrating that a model can learn not just vocabulary but structural and stylistic patterns from a focused dataset.  It made my own archive useful in a way search never could. [blog.gopenai](https://blog.gopenai.com/fine-tuning-large-language-models-to-replicate-scholarly-writing-styles-a-step-by-step-guide-945ac3a6d90f)

Some people will point out that retrieval-augmented generation does the same job without fine-tuning. You build a RAG pipeline over your notes and let the model query it. That gets you halfway there. But as both AWS and independent technical analysis confirm, RAG and fine-tuning solve different problems: RAG retrieves existing information, while fine-tuning changes the model's generative behaviour at the level of tone, rhythm and structure.  Those are different problems and they need different solutions. A hybrid of both is often where the real value sits. [kairntech](https://kairntech.com/blog/articles/retrieval-augmented-generation-vs-fine-tuning-choosing-the-right-approach/)

#### How I Started Sounding Like Everyone Else

The longer you write with a generic model, the more you start writing like one. I noticed this over about eighteen months. My drafts got smoother. More structured. Less surprising. This is documented. A Cornell University study published in April 2025 found that when people used an AI writing assistant, their writing converged, with distinct cultural and individual voices becoming more similar. The lead researcher described it as one of the first studies to show that AI use in writing could lead to cultural stereotyping and language homogenisation.  A separate study found that teachers rated AI-assisted essays as more fluent and well-structured but thinner in voice and original insight.  Designers know this pattern already: you become the tools you use. [news.cornell](https://news.cornell.edu/stories/2025/04/ai-suggestions-make-writing-more-generic-western)

The easy answer is that this is a discipline problem. Write your first draft without AI, stay intentional, and the drift stops. Maybe. But the structural problem is that a generic tool has no stake in preserving your voice; it has every incentive to smooth it out. A 'New Yorker' piece from June 2025 put it plainly: large language models are designed to identify patterns within extensive datasets, producing outputs that lean toward consensus, and that tendency shapes the humans using them over time, not just the outputs.  Building a model trained on your voice changes the incentive structure of the tool itself. You stop fighting the current. [newyorker](https://www.newyorker.com/culture/infinite-scroll/ai-is-homogenizing-our-thoughts)

#### What A.R. Rahman Understood About Synthesis

A.R. Rahman did not become the sound of Indian cinema by learning Western orchestration and stopping there. He built a recording studio in his home in Chennai, trained himself in MIDI composition when others were still working with live orchestras, and fused Carnatic classical music, Sufi devotional music and Western pop into a grammar that nobody else could replicate because nobody else had lived the same combination of influences. The tools he used were available to anyone. The synthesis was entirely his. A generic LLM gives you the tools. A personal model gives you the synthesis.

#### The Objection Worth Taking Seriously

The real objection is this: fine-tuning on a small personal corpus risks amplifying your weaknesses, not just your strengths. If your writing has blind spots or structural tics, a personal model learns those too. You end up with a tool that is very good at sounding like you, including the parts that could use an editor. This is a documented limitation; fine-tuned models carry an increased risk of reinforcing patterns present in the training data and do not surface references to challenge those patterns. [docs.aws.amazon](https://docs.aws.amazon.com/prescriptive-guidance/latest/retrieval-augmented-generation-options/rag-vs-fine-tuning.html)

That is true. But it mistakes the use case. A personal model handles voice and continuity. It does not replace editorial challenge; that still needs to come from outside. A musician practising in their own style is not avoiding critique. They are making sure the critique lands on something genuinely theirs.

#### What Happens If You Do Not

The tools that matter in the next decade will be the ones that know you well enough to be genuinely useful, not just generically capable. If you keep outsourcing your voice to a model trained on everyone else's thinking, you are not augmenting yourself; you are slowly licensing your distinctiveness out. The people who build personal AI infrastructure now will own their intellectual edge; the rest will spend the next ten years sounding like each other. [news.cornell](https://news.cornell.edu/stories/2025/04/ai-suggestions-make-writing-more-generic-western)

---

### In Defence of Thinking
[Read more](https://www.gaganmalik.io/en/newsroom/in-defence-of-thinking)

[Retrieval brief — in-defence-of-thinking]
Category: culture
Thesis: We are not distracted. We are managed. And I say that as someone who spent two years mistaking the feed for a life.
Answers: What does Gagan argue in "In Defence of Thinking"? | What is Gagan Malik's view on In Defence of Thinking?
Concepts: culture, Gagan Malik, writings

By Gagan Malik

#### I Got This Wrong

Last year I was building a product. Real work. The kind that requires you to hold ten variables in your head simultaneously and make judgement calls you cannot undo. I was also checking LinkedIn every forty minutes.

I told myself I was staying informed. I was tracking the impressions on my last post. I was reading replies. I was doing, at a generous estimate, 45 minutes of actual thinking per day, wrapped inside eight hours of feeling busy. I did not notice for a long time. The feed is very good at feeling like work. That is the whole design.

What broke the illusion was simpler than I expected. I sat down one morning to explain a technical problem to my team, a topic I had spent three weeks "following" online, and I had almost nothing to say. Not because I had not been paying attention. Because I had been paying the wrong kind of attention, to the wrong things, at the wrong depth, for a very long time.

That was not distraction. That was managed ignorance, delivered at scale, with a clean UI.

#### The Argument I Cannot Shake

British adults now spend 7 hours and 27 minutes a day in front of screens. Ten of those minutes go to news. The rest goes to the feed. [IPA TouchPoints 2025](https://ipa.co.uk/news/touchpoints-2025)

I keep coming back to Neil Postman, who wrote in 1985 that every communication medium reshapes culture "by promoting certain intellectual pursuits, favoring specific definitions of intelligence and wisdom". He was writing about television. He died before TikTok. But he had already described it precisely: a world where the medium does not carry the message. The medium is the message. And the message right now, running underneath everything, is that thinking is optional. [Slate: Amusing Ourselves to Death](https://slate.com/culture/2025/03/amusing-ourselves-death-neil-postman-ezra-klein.html)

He feared Huxley's 'Brave New World' more than Orwell's '1984'. You do not need to ban books. You just need to make them feel like effort in a world engineered to feel effortless. [Slate: Amusing Ourselves to Death](https://slate.com/culture/2025/03/amusing-ourselves-death-neil-postman-ezra-klein.html)

#### Why the Game Is Rigged

Short-form content generates 70% of all user interactions across digital platforms. So TikTok and Instagram Reels build for short-form. Their algorithms learn that outrage and novelty hold attention longer than nuance. Advertisers pay more for held attention. Meta and ByteDance earn more revenue. The cycle does not pause. [An-Nahar: economy of brain rot](https://en.annahar.com/en/international/world/261805/the-economy-of-brain-rot-when-attention-becomes-a-marketable-commodity)

Thought runs on a completely different clock. Reading something serious requires sitting with uncertainty for twenty minutes before any reward arrives. No notification congratulates that patience. No algorithm surfaces it to your followers.

Entertainment did not beat thought in a fair contest. It hired behavioural psychologists, ran A/B tests on human neurochemistry, and filed the results as product improvements. As Jay Caspian Kang wrote in 'The New Yorker', the structure of online political conversation, millions of petty arguments, makes genuine dialogue structurally impossible. The Sirens did not overpower Odysseus. They were simply better funded. [The New Yorker: The Sirens Call](https://www.newyorker.com/magazine/2025/01/27/the-sirens-call-chris-hayes-book-review)

Among 15 to 24-year-olds in the UK, daily mobile usage is now almost five hours. The feed replaced the broadcast, and unlike the broadcast, the feed has no editor, no watershed, no obligation to the public. It has an engagement metric and an earnings call. That is the whole governance structure. [IPA TouchPoints 2025](https://ipa.co.uk/news/touchpoints-2025)

#### The Counterargument I Take Seriously

Access matters. A teenager with a smartphone in 2026 reaches more knowledge than an Oxford student could in 1985. The floor is genuinely higher. But the floor is not where any of us are competing.

The people I watch extracting real advantage from this moment are the ones who already have strong mental models and use the tools to pressure-test them. They go into the feed with a question and come out with an answer. The rest of us go in with an hour and come out with nothing we can use. Postman noted that Reagan proved credibility no longer meant reliability, only the impression of sincerity. We have not moved past that. We have automated it. [Slate - Amusing Ourselves to Death](https://slate.com/culture/2025/03/amusing-ourselves-death-neil-postman-ezra-klein.html)

Raising the floor while the ceiling keeps rising is not progress. It is a more comfortable form of the same gap.

#### What I Am Actually Asking Myself

We spend 4.5 hours online every day. Roughly half of that is inside products owned by two companies. We are renting our thinking to Alphabet and Meta daily, and paying for the access with our phone contract and our cognitive capacity. [Ofcom: UK adults online](https://www.ispreview.co.uk/index.php/2025/12/ofcom-finds-uk-adults-now-spend-over-4-5-hours-online-each-day.html)

I am not telling you to delete the apps. I deleted mine twice. They came back, and I let them. What I am trying to do now, imperfectly, is treat serious reading like a meeting I cannot move. Ring-fence the time. Rebuild the friction deliberately, because friction is not the enemy of thought. It is the condition thought requires.

Nobody is building a notification that says: 'you have been entertained enough now, please think.' That one is still on us.

---

### v1.12.1
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-12-1)

[Retrieval brief — release-1-12-1]
Category: technology
Thesis: Listen bar and share menu on all newsroom posts; byline row everywhere.
Answers: What changed in gaganmalik.io v1.12.1? | What's new on Gagan Malik's website in v1.12.1? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

Listen bar and share menu on all newsroom posts; byline row everywhere.

---

### Your Calendar Is Being Mugged, and You Are Holding the Door Open
[Read more](https://www.gaganmalik.io/en/newsroom/the-lost-art-of-refusal)

[Retrieval brief — the-lost-art-of-refusal]
Category: business
Thesis: Your calendar is not a productivity tool anymore. It is a demand-capture system. Why we treat No like profanity—and how to take it back.
Answers: What does Gagan argue in "Your Calendar Is Being Mugged, and You Are Holding the Door Open"? | What is Gagan Malik's view on Your Calendar Is Being Mugged, and You Are Holding the Door Open? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

In Fargo Season 5, there is a boardroom shot that should be stapled to every HR handbook. Lorraine Lyon (Jennifer Jason Leigh), who played a central antagonist-turned-ally as the cold, wealthy "Queen of Debt," sits at the head of the table like a final boss, and behind her is a gigantic piece of wall art that says "NO" in letters big enough to qualify as company policy. Because modern work has a weird superstition. We treat "No" like it is profanity, then we act surprised when our time, our attention, and our sanity get strip-mined by other people's priorities.

I know this because I lived it. I once said "sure" to a "quick" set of stakeholder catch-ups that turned into a weekly standing meeting, plus a pre-meeting, plus a "quick sync after". That harmless politeness cost me 3 hours a week for a quarter. Call it 36 hours. At a conservative £200 an hour of fully loaded senior time, that is £7,200 of corporate value converted into vibes and posturing. This is the quiet part out loud. You are not busy. You are under-priced.

#### Convenience is the drug, Yes is the side effect

Tim Wu called convenience "the most underestimated and least understood force in the world today". He is right, and it is not just about Deliveroo and next-day shipping. It is about decision-making. [The New York Times](https://www.nytimes.com/2018/02/16/opinion/sunday/tyranny-convenience.html) Convenience removes friction. Friction is the moment you pause, think, and sometimes say no. Remove friction and your default becomes compliance. That is why the "Accept" button is one tap, and declining feels like you have to write a short apology letter to the United Nations.

Wu quotes Evan Williams, Twitter co-founder, saying "Convenience decides everything." If that line does not chill you, you have not looked at your diary recently. [The New York Times](https://www.nytimes.com/2018/02/16/opinion/sunday/tyranny-convenience.html) Your calendar is not a productivity tool anymore. It is a demand-capture system.

#### The economics of being "nice"

Businesses love your reluctance to say no because it is free labour. Your politeness subsidises their ambiguity. The corporate version is what I call the "infinite backlog scam". There is always more work than capacity, so the system survives by getting you to personally absorb the overflow. You call it being helpful. Finance calls it margin. And then you wonder why you cannot do deep work, why your projects ship late, and why you feel permanently behind.

Some workplaces institutionalise this. In one analysis of "Yes culture", research cited there notes that 55% of companies deny employees true autonomy. If that is even directionally accurate, then the average employee is not a knowledge worker. They are a highly educated order-taker with a laptop. [LinkedIn](https://www.linkedin.com/pulse/culture-yes-just-say-sarah-hendricks) That is not a talent strategy. That is a burnout factory.

#### Why ghosting replaced "No"

Now we get to the most dystopian part. Ghosting has become more convenient than saying no.

Not just in dating. In hiring, in sales, in partnerships, in internal decision-making. People disappear because it is the cheapest refusal on the market. Here is the logic. Saying "No" has a social cost. You risk being disliked. You risk conflict. You risk being seen as "difficult". Ghosting has a lower immediate cost. You say nothing, you feel nothing, you move on. Convenience culture makes this worse because it trains everyone to expect instant agreement. So when you do not want to agree, you pick the path that avoids the emotional admin. Silence. [The New York Times](https://www.nytimes.com/2018/02/16/opinion/sunday/tyranny-convenience.html) Ghosting is what happens when a society loses the language of clean refusal. It is cowardice, packaged as efficiency. It also destroys trust, which is hilarious because trust is the only thing that scales.

#### The "No" advantage

Warren Buffett is often quoted saying that the difference between successful people and very successful people is that the very successful say no to almost everything. People hear that and think it is motivational wallpaper. [PMC/NCBI](https://pmc.ncbi.nlm.nih.gov/articles/PMC7332800/) It is not. It is basic economics.

Time is finite. Attention is finite. Energy is finite. If you keep selling your finite resources for cheap approval, you will stay poor in the only currencies that matter. Saying no is not negative. It is prioritisation with a backbone.

#### A framework that does not require a personality transplant

You do not need to become rude. You need to become explicit. If a request is not obviously a 'hell yes', do not answer in real time. Give yourself a buffer. That buffer is where your agency lives. Use it to name the trade-off: say 'I can do this. What should I drop?' This converts vibes into prioritisation and reveals whether the requester actually cares.

When the answer is no, anchor it in something you are protecting. William Ury's central point is that a strong no is built on a deeper yes. You are not rejecting the person. You are defending a prior commitment. 'No, because I am committed to X' lands very differently from 'No, I'm busy.' [YourStory](https://yourstory.com/2025/05/7-books-that-teach-boundaries) If you lead teams, treat silence as a bug, not a personality trait. The word on that Fargo wall is not decoration. It is a boundary made visible. Most of us do the opposite. We keep our boundaries invisible, then act shocked when people walk straight through them.

Boardroom with "NO" wall art. Fargo Season 5.

Your life does not need more productivity hacks. It needs one sentence you can say without sweating.

"No."

---

### v1.12.0
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-12-0)

[Retrieval brief — release-1-12-0]
Category: technology
Thesis: Stripe Customer Portal, llms.txt, About refresh, Person schema, PWA crawler fix, clean story slugs.
Answers: What changed in gaganmalik.io v1.12.0? | What's new on Gagan Malik's website in v1.12.0? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

Stripe Customer Portal, llms.txt, About refresh, Person schema, PWA crawler fix, clean story slugs.

---

### The $725 Million Bug Hunt
[Read more](https://www.gaganmalik.io/en/newsroom/the-725-million-bug-hunt)

[Retrieval brief — the-725-million-bug-hunt]
Category: technology
Thesis: The most expensive hobby in tech that nobody talks about at dinner parties: chip verification. First-time silicon success has plummeted to 14%. The verification productivity gap is an everything problem wearing a lab coat.
Answers: What does Gagan argue in "The $725 Million Bug Hunt"? | What is Gagan Malik's view on The $725 Million Bug Hunt? | How does this essay relate to AI and technology product decisions?
Concepts: technology, Gagan Malik, writings

By Gagan Malik

Let me tell you about the most expensive hobby in tech that nobody talks about at dinner parties: chip verification.

Not the sexy kind of chips. Not the ones Jensen Huang waves around on stage like a newborn at a christening. I mean the 18 months before that, when thousands of engineers sit in cubicles running simulations, hunting bugs in circuits smaller than a human hair, praying to whatever god they believe in that the thing works the first time.

Spoiler: it almost never does.

The latest data from the Siemens EDA/Wilson Research Group study landed like a brick through a window. First-time silicon success has plummeted to 14%. That is not a typo. Fourteen per cent. Down from 24% two years prior and roughly 30% historically. Three out of four chip projects are running behind schedule. And the average cost to design a single chip at the 2nm node? $725 million. [semiengineering](https://semiengineering.com/first-time-silicon-success-plummets/)

You now have better odds of a startup surviving five years than your chip working first time. And when you lose, you do not lose your shirt. You lose a mid-size office building.

#### The 70% problem nobody pitches

Here is the quiet part out loud: verification consumes over 70% of a chip project's time and headcount. Not design. Not manufacturing. Verification. The bit where you check whether the thing you drew actually does what you think it does. [blogs.sw.siemens](https://blogs.sw.siemens.com/verificationhorizons/2025/09/03/why-first-silicon-success-is-getting-harder-for-system-companies/)

And within that 70%, the nastiest, most manual, most soul-destroying task is debugging. Finding the root cause of a failure in a haystack of billions of transistors. Post-silicon debug alone can run $15 to $20 million and six months per project. Pre-silicon debug is no better; it is just cheaper to cry about. [web.eecs.umich](https://web.eecs.umich.edu/~imarkov/pubs/conf/iccad07-fogclear.pdf)

Harry Foster, chief verification scientist at Siemens EDA, put it bluntly: "What we're doing is not working. We need to significantly increase productivity". This is not some blogger. This is the person who has run the industry's definitive verification survey for over two decades. [semiengineering](https://semiengineering.com/first-time-silicon-success-plummets/)

#### Why investors should care (but do not, yet)

Every AI model you use, every copilot, every autonomous vehicle, every data centre GPU cluster, sits on top of silicon. Jensen Huang himself frames AI as a five-layer cake: energy, chips, infrastructure, models, applications. If the chip layer breaks, the four layers above it do not wobble. They collapse. [blogs.nvidia](https://blogs.nvidia.com/blog/davos-wef-blackrock-ceo-larry-fink-jensen-huang/)

And right now, the chip layer has a structural crack called the verification productivity gap. Complexity is scaling exponentially. Verification capacity is not. The gap was first identified in the early 2000s. We are now in what Siemens officially calls "Verification Productivity Gap 2.0". Same disease, nastier symptoms. [verificationacademy](https://verificationacademy.com/topics/planning-measurement-and-analysis/breaking-the-bottleneck-overcoming-the-verification-productivity-gap/)

Meanwhile, the EDA market, the tools that make chip design possible, captures roughly 2 to 3% of semiconductor revenue. Two to three per cent! The entire industry that enables a $600 billion chip market earns less than what Apple spends on marketing. Liyue Yan at Boston University nailed it: EDA has a value capture problem. They create enormous value, then negotiate like they are selling second-hand furniture at a car boot sale. [semiwiki](https://semiwiki.com/eda/363537-eda-has-a-value-capture-problem-an-outsiders-view/)

#### The money is moving

I will be honest: I dismissed this space for too long. I looked at EDA and saw legacy toolchains, three dominant incumbents, and margins that made my eyes water for the wrong reasons. I was wrong. The market has shifted under my feet.

The EDA market hit roughly $20.8 billion in 2026, growing at 8 to 9% CAGR. Verification and sign-off is the largest segment at 26% share, about $5.4 billion, and growing faster than the average. Cloud EDA is a $3.7 billion sub-market on its own. [mordorintelligence](https://www.mordorintelligence.com/industry-reports/electronic-design-automation-eda-tools-market)

And then ChipAgents happened. Founded in 2024, the company just closed an oversubscribed $74 million raise led by TSMC-backed Matter Venture Partners, alongside Bessemer, Micron, MediaTek, and Ericsson. They report 140x year-over-year ARR growth and deployment at 80 semiconductor companies. Their pitch: agentic AI that reads specs, generates verification assets, and does automated root-cause analysis. Basically, they are trying to eat the debug bottleneck alive. [businesswire](https://www.businesswire.com/news/home/20260217568914/en/ChipAgents-Raises-$74M-to-Scale-an-Agentic-AI-Platform-to-Accelerate-Chip-Design)

#### The framework you actually need

Stop thinking of verification as a line item in chip design. Start thinking of it as risk infrastructure for the entire AI economy.

Every week a tape-out slips, a hyperscaler's GPU cluster ships late. Every respin burns $25 million and three to six months. Every escaped bug that reaches production costs 100x what it would have cost to catch in simulation. [community.cadence](https://community.cadence.com/cadence_blogs_8/b/fv/posts/how-to-maximize-productivity-and-lower-cost-for-enterprise-prototyping)

The companies that compress debug cycles do not just save their customers money. They accelerate the clock speed of AI itself. That is the pitch. That is the thesis.

The verification productivity gap is not a semiconductor problem. It is an everything problem wearing a lab coat. And the founders who close it will not just build good businesses. They will quietly become the most important infrastructure layer in the AI stack that nobody on X (Formerly Twitter) has heard of.

Yet.

---

### v1.10.0
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-10-0)

[Retrieval brief — release-1-10-0]
Category: technology
Thesis: RSS feed, What's New banner, PWA badge for new content, ghost-style newsroom buttons.
Answers: What changed in gaganmalik.io v1.10.0? | What's new on Gagan Malik's website in v1.10.0? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

RSS feed, What's New banner, PWA badge for new content, ghost-style newsroom buttons.

---

### v1.11.11
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-11-11)

[Retrieval brief — release-1-11-11]
Category: technology
Thesis: YouTube videos in newsroom, transcripts in Ask RAG, Videos archive filter.
Answers: What changed in gaganmalik.io v1.11.11? | What's new on Gagan Malik's website in v1.11.11? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

YouTube videos in newsroom, transcripts in Ask RAG, Videos archive filter.

---

### Against One-Click Coding: Reclaiming Craft for Personal Projects
[Read more](https://www.gaganmalik.io/en/newsroom/against-one-click-coding)

[Retrieval brief — against-one-click-coding]
Category: technology
Thesis: One-click and vibe-coded builds trade agency for speed; canonical owned assets and governed enterprise products need intentional craft and control.
Answers: What is Gagan's view on vibe coding and one-click app builders? | When should a company use AI codegen versus governed engineering? | Why does Gagan build his personal site in an IDE instead of no-code? | What is the enterprise risk of comfort-first AI product defaults?
Implication: Use agents for throwaway MVPs; keep canonical footprint, regulated surfaces, and long-lived products under explicit architecture and review.
Related stories: portfolio-built-like-a-product, presto-automated-monetisation
Related essays: the-gui-sells-comfort-the-cli-sells-control, one-billion-tokens-later
Concepts: vibe coding, agency vs speed, owned media, AI governance

By Gagan Malik

Let's talk about your personal website.

I know what it looks like. You built it on a Sunday afternoon. You fired up a vibe-coding app—maybe Claude Code, maybe Lovable, maybe one of those platforms that promises "from prompt to production in 60 seconds." You typed "Make me look like a visionary but approachable tech founder," pressed enter, and went to make a flat white.

And it looks fine. It's got the trendy brutalist typography. The dark mode toggle. The slight gradient on the hover state. It looks exactly like the websites of the other 400,000 guys who asked an AI to make them look visionary but approachable this month.

You outsourced your digital identity to an LLM optimizing for the median acceptable aesthetic. Congratulations. You are now perfectly average.

I'm building my personal site right now. I'm doing it the hard way. I'm in an IDE. I'm fighting with Git. I'm staring at a CLI.

Am I doing this because I'm a masochist? No. I'm doing it because I understand the difference between a product and a craft object. And if your personal website isn't a craft object, why does it even exist?

Let's get the economics out of the way first. I am not anti-automation. In my day job, I use AI tools to collapse effort constantly. If I need a throwaway landing page to test a value prop for a new SaaS product, I am absolutely letting an agent build it. Time is money, and I am not spending £500 of my time hand-coding a button for an MVP that might die on Thursday.

But your personal website is not a throwaway MVP. It is your canonical footprint on the internet. It is the one digital asset you own outright in an era of rented land.

When you use a low-code or no-code platform for this, you aren't just saving time. You are accepting a fundamental trade-off: you are trading agency for speed.

You think you are the conductor, but you are actually the passenger. You are constrained by their abstractions, their component libraries, their pricing tiers, and their definition of what a website should be. As developer Jay Little put it recently, the promise of low-code is a lie when it comes to long-term maintainability. You are taking on technical debt disguised as convenience. [jaylittle](https://jaylittle.com/post/view/2023/4/low-code-software-development-is-a-lie/)

More importantly, you are missing out on the actual value of building something yourself: the friction.

There's an old John Dewey concept—often summarised as "we learn by doing". Dewey's actual point was sharper: we do not learn from experience alone; we learn from reflecting on experience. [structural-learning](https://www.structural-learning.com/post/john-deweys-theory)

You don't reflect when a machine generates 4,000 lines of React for you in three seconds. You just deploy it and hope it doesn't break.

When I'm in my IDE, setting up my own routing, configuring my own build tools, and making deliberate commits via the CLI, I am forced to be intentional. Every folder structure is a design decision. Every dependency is a liability I have to accept. I am staying close to the grain of the medium.

This is software as a craft, not a commodity. [8thlight](https://8thlight.com/insights/why-software-development-is-a-craft)

And let's talk about the ROI of that friction. Yes, it costs me more hours up front. But what does it buy me?

It buys me full-stack literacy. When things break—and they will break—I have the levers to fix them. I'm not waiting in a Discord support channel for a startup with a $50M valuation and an 18-month runway to patch their platform.

It buys me portability. My Git repo is mine. I can move it to Vercel, to Netlify, to a Raspberry Pi in my living room. Try exporting your highly customized, database-linked project out of a proprietary no-code builder. It's like trying to get your data out of a casino.

But mostly, it buys me taste.

Vibe-coding tools optimize for output speed. They encourage you to ship the first plausible draft. But plausible is the enemy of precise. When you build the stack yourself, the constraints force you to make choices. You inject your actual personality into the architecture, not just the CSS. [rebeccabilbro.github](https://rebeccabilbro.github.io/low-code-no-code/)

We are entering an era where AI can write competent code instantly. Competent code is about to have a market value of zero.

What will have value is taste, architectural judgment, and a deep understanding of how systems actually fit together beneath the abstractions.

You don't build those muscles by typing prompts into a chat box. You build them in the trenches. You build them by getting your hands dirty in the terminal.

Your personal site shouldn't just be a billboard. It should be your lab. It's the one place where you don't have to compromise with a product manager, a client, or a venture capitalist.

Stop treating your digital identity like a fast-food order. Reclaim your agency. Open your terminal. Build it yourself.

It's going to be frustrating. It's going to take longer. But when you're done, you'll actually own it. And in 2026, owning your own infrastructure might be the biggest flex there is.

---

### v1.9.0
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-9-0)

[Retrieval brief — release-1-9-0]
Category: technology
Thesis: Hindi, Arabic, Spanish locales, region selector, FAQ accordion, region pricing, RTL.
Answers: What changed in gaganmalik.io v1.9.0? | What's new on Gagan Malik's website in v1.9.0? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

Hindi, Arabic, Spanish locales, region selector, FAQ accordion, region pricing, RTL.

---

### v1.8.0
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-8-0)

[Retrieval brief — release-1-8-0]
Category: technology
Thesis: Skeleton loaders site-wide, archive filter sticky fix, sticky newsroom header.
Answers: What changed in gaganmalik.io v1.8.0? | What's new on Gagan Malik's website in v1.8.0? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

Skeleton loaders site-wide, archive filter sticky fix, sticky newsroom header.

---

### v1.7.0
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-7-0)

[Retrieval brief — release-1-7-0]
Category: technology
Thesis: Mobile responsive pass, 44pt touch targets, Ask history as sheet on small screens.
Answers: What changed in gaganmalik.io v1.7.0? | What's new on Gagan Malik's website in v1.7.0? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

Mobile responsive pass, 44pt touch targets, Ask history as sheet on small screens.

---

### v1.4.0
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-4-0-rc-0)

[Retrieval brief — release-1-4-0-rc-0]
Category: technology
Thesis: Ask case study citations with thumbnails, expandable suggestion chips, Button size variants.
Answers: What changed in gaganmalik.io v1.4.0? | What's new on Gagan Malik's website in v1.4.0? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

Ask case study citations with thumbnails, expandable suggestion chips, Button size variants.

---

### v1.4.0-rc.1
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-4-0-rc-1)

[Retrieval brief — release-1-4-0-rc-1]
Category: technology
Thesis: Code cleanup, related-items refactor, Ask test fix, removed Kyle Hanagami story.
Answers: What changed in gaganmalik.io v1.4.0-rc.1? | What's new on Gagan Malik's website in v1.4.0-rc.1? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

Code cleanup, related-items refactor, Ask test fix, removed Kyle Hanagami story.

---

### v1.4.1
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-4-1)

[Retrieval brief — release-1-4-1]
Category: technology
Thesis: Against One-Click Coding essay, citation pills with favicons, publication dates on Writings cards.
Answers: What changed in gaganmalik.io v1.4.1? | What's new on Gagan Malik's website in v1.4.1? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

Against One-Click Coding essay, citation pills with favicons, publication dates on Writings cards.

---

### v1.5.0
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-5-0)

[Retrieval brief — release-1-5-0]
Category: technology
Thesis: Favicon, Open Graph images, RAG reseeds on every deploy.
Answers: What changed in gaganmalik.io v1.5.0? | What's new on Gagan Malik's website in v1.5.0? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

Favicon, Open Graph images, RAG reseeds on every deploy.

---

### v1.3.0
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-3-0)

[Retrieval brief — release-1-3-0]
Category: technology
Thesis: Cookie consent banner, release plan docs, Ask integration tests.
Answers: What changed in gaganmalik.io v1.3.0? | What's new on Gagan Malik's website in v1.3.0? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

Cookie consent banner, release plan docs, Ask integration tests.

---

### v1.1.0
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-1-0)

[Retrieval brief — release-1-1-0]
Category: technology
Thesis: Streamlined homepage, seven case studies, Stories hero with Ask Gagan CTA.
Answers: What changed in gaganmalik.io v1.1.0? | What's new on Gagan Malik's website in v1.1.0? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

Streamlined homepage, seven case studies, Stories hero with Ask Gagan CTA.

---

### Gen Z Didn't Invent Knowledgemaxxing. They Just Stopped Lying About It.
[Read more](https://www.gaganmalik.io/en/newsroom/gen-z-knowledgemaxxing)

[Retrieval brief — gen-z-knowledgemaxxing]
Category: business
Thesis: Everyone is knowledgemaxxing. Gen Z just turned it into a bit. The edge now isn't more knowledge—it's filters.
Answers: What does Gagan argue in "Gen Z Didn't Invent Knowledgemaxxing. They Just Stopped Lying About It"? | What is Gagan Malik's view on Gen Z Didn't Invent Knowledgemaxxing. They Just Stopped Lying About It? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

Here is the uncomfortable truth: everyone is knowledgemaxxing. Gen Z just turned it into a bit. Millennials turned it into burnout. Boomers turned it into a LinkedIn post in 12-point Calibri.

Knowledgemaxxing is simple. It's the rebrand of something older than Wi‑Fi. Taking in more information than you can possibly action, convincing yourself it's "research", then wondering why your portfolio looks like a charity for unprofitable business models.

And yes, I've done it too. In 2021, I spent 3 months "deep diving" Web3. Translation. Eight hours a day in Discords, Twitter Spaces and whitepapers. I told myself I was "building conviction". What I actually built was a 5‑figure position in a governance token that did a friendly 80% drawdown. That is a £14,300 tuition fee I paid to learn one basic lesson.

More information is not the same as better decisions.

#### The cult of recreational research

Here is what's new. Gen Z took "learning" and turned it into lifestyle content.

Polyester Zine called it "knowledgemaxxing" and framed research as a leisure activity, "undisciplined and playful" rather than graded and credentialled. TikTok is full of "Spend a research day with me" and "Topics I'm researching this week" videos. It looks like a lo‑fi study vlog. Underneath, it's a quiet revolt against brain rot and bullshit. [polyesterzine](https://www.polyesterzine.com/features/knowledgemaxxing-research-as-a-leisure-activity-and-the-gen-z-yearning-for-intellectual-growth)

There is data behind the vibe. Oxford picked "brain rot" as its 2024 Word of the Year after usage spiked 230%, defining it as the supposed decline in mental state from consuming trivial content. That's not a meme. That is a generation diagnosing its own cognitive hangover. [edition.cnn](https://edition.cnn.com/2024/12/02/world/brain-rot-oxford-word-of-the-year-2024-intl-scli-gbr)

But here's the contrarian bit. If you think this is a Gen Z phenomenon, you have not looked in the mirror.

Millennials invented this behaviour on RSS feeds and Medium. Boomers did it with Sky News, the FT weekend and 14 "must read" business books a year. The packaging changed. The addiction to input did not.

American Press Institute data says 79% of Gen Z and Millennials consume news daily and 96% at least weekly. That is not a vibes-based, feelings-first generation. That is a cohort chain‑smoking information. [thevarsity](https://thevarsity.ca/2023/09/10/opinion-gen-z-is-redirecting-the-path-to-intellectualism-not-derailing-it/)

The question is not who is reading. The question is who is thinking.

#### Your portfolio is a museum of half-read ideas

If you want to see the cost of bad knowledgemaxxing, open your brokerage account.

Every line item is an artefact from some content rabbit hole.

You listened to three AI podcasts and bought the top of a SaaS name trading at 30 times sales.

You read a Substack thread on "energy security" and now own a fossil stock that has done nothing but pay you a 3% dividend and emotional regret.

You watched one 12‑minute YouTube explain‑like‑I'm‑five on quantum computing and bought an ETF whose top holding is a company that sells more press releases than products.

I did this with "metaverse infrastructure" in 2021. The deck looked beautiful. Total addressable market allegedly in the trillions. The P&L was less beautiful. Revenue flat. R&D ballooning. Cash burn heroic. I still bought. Why? Because I had consumed so much content about "the inevitability of virtual worlds" that saying no felt like I was missing the train.

Twelve months later, the position was down 72%. The market had not mispriced the future. I had mispriced my own overconfidence.

This is the core bug. Knowledgemaxxing without constraint turns into narrative addiction. You think you are doing due diligence. You are actually doing mood boarding.

#### Boomers do it with CNBC, Zoomers do it with Substack

There is a smug take floating around that Gen Z is uniquely broken by short‑form content. Yes, TikTok has rewired attention spans. Roughly 40% of under 30s now use TikTok as a regular news source and around 60% of TikTok's billion‑plus users are Gen Z. That matters. [stackoverflow](https://stackoverflow.blog/2025/06/18/learn-like-a-lurker-gen-z-s-digital-first-lifestyle-and-the-future-of-knowledge/)

But The Atlantic points out that younger Gen Z readers are also checking out more print books from libraries than older cohorts. BookTok, the corner of TikTok dedicated to books, helped push Colleen Hoover to over 2 million copies sold in a single year. NYT‑level bestsellers, off the back of teenagers crying into front‑facing cameras about paperbacks. [theatlantic](https://www.theatlantic.com/newsletters/archive/2023/12/library-gen-z-readers/676963/)

So no, they are not illiterate. They just do not care about your canon.

Meanwhile Boomers sit on 24‑hour news and call it "staying informed". I grew up in Delhi watching older relatives watch business channels the way teenagers now watch Twitch. Same energy. Different medium. Same output. A lot of noise, a few trades, and a deep sense that the market is unfair when it does not reward your favourite narrative.

Gen Z at least has the honesty to label low‑value content as "brain rot" and then build "anti brain‑rot menus" and "doomscrolling cures" to climb out of it. Half of them openly say social media is harming teen mental health and wish they had spent less time on their phones. When was the last time a Boomer admitted CNBC made them panic sell? [stylus](https://stylus.com/pop-culture-media/knowledge-maxxing-serves-anti-brain-rot-low-stakes-intellectualism)

#### The real edge isn't more knowledge. It's filters.

The quiet part nobody in tech wants to say out loud in a funding meeting.

You are not under‑informed. You are over‑indexed on other people's conclusions.

The McKinsey and Reuters crowd will tell you Gen Z is "visual‑first" and consumes news through "edutainment" formats like GIFs and quick videos. True. But Millennials are no better. We mainlined blog summaries, Twitter threads and podcast hot takes through the 2010s and called it "staying ahead of the curve". [blog.quintype](https://blog.quintype.com/business/gen-z-news-consumption)

The result. Everyone knows the same surface‑level facts. Everyone quotes the same three charts. Everyone makes the same trades a week too late.

The edge now is not being the first to know. That was 1998 behaviour. The edge is being the first to stop caring about 90% of the noise.

Gen Z's "commonplace books" and offline film logs are interesting because they hint at this. Taking notes by hand. Logging what actually stuck. It is not aesthetic. It is a filter. [thegoodtrade](https://www.thegoodtrade.com/features/what-is-a-commonplace-book/)

Boomers had filing cabinets and cut‑out FT clippings. Same idea. The difference is that this generation is starting to recognise that curation is not something The Algorithm does for you. It is something you do yourself if you want to keep your brain and your balance sheet intact.

#### How to knowledgemaxx like an adult

So what do you do if you are a Millennial or Boomer who has been cosplaying as an analyst via Instagram carousels and quarterly earnings memes?

Three rules. They are boring. They work.

1. Put a hard cap on inputs. If American Press Institute says 79% of your cohort is taking in news daily, assume you are already saturated. Pick two primary sources and one longform format. That's it. For me it is one newspaper, one industry newsletter, one podcast. Everything else is dessert, not dinner. [thevarsity](https://thevarsity.ca/2023/09/10/opinion-gen-z-is-redirecting-the-path-to-intellectualism-not-derailing-it/)

2. Assign every idea a price tag. Before you "go deep" on a theme, write down what it will cost you if you are wrong. In capital, in time, in emotional energy. My Web3 tuition fee was £14k. If I had written that down at the start, I would have cut my position size in half and treated the rest as speculative R&D spend, not conviction investing.

3. Separate research from entertainment. If you find yourself consuming content because it makes you feel smart, log it as entertainment, not due diligence. Entertainment is fine. Just do not let it near position sizing.

Knowledgemaxxing is not the enemy. It is the default state of an economy where information is cheap and attention is the scarce resource. Gen Z's real innovation is that they have stopped pretending this is purely noble.

The rest of us need to catch up. Not by learning more, but by admitting that half our "research" is just sophisticated procrastination dressed in a nice font.

Once you see that, you stop trying to out‑read the market.

You start trying to out‑filter it.

---

### v1.0.0
[Read more](https://www.gaganmalik.io/en/newsroom/release-1-0-0)

[Retrieval brief — release-1-0-0]
Category: technology
Thesis: First public version: Home, About, Stories, Newsroom, Pricing, Ask, analytics, SEO.
Answers: What changed in gaganmalik.io v1.0.0? | What's new on Gagan Malik's website in v1.0.0? | What are the latest gaganmalik.io release notes?
Concepts: release notes, site update, gaganmalik.io

First public version: Home, About, Stories, Newsroom, Pricing, Ask, analytics, SEO.

---

### The Curious Case of the Commute: Why Your Interview is Back in 3D
[Read more](https://www.gaganmalik.io/en/newsroom/the-curious-case-of-the-commute)

[Retrieval brief — the-curious-case-of-the-commute]
Category: business
Thesis: Employers are bringing back in-person interviews because your AI is getting the job—and then you have to show up and actually do it.
Answers: What does Gagan argue in "The Curious Case of the Commute: Why Your Interview is Back in 3D"? | What is Gagan Malik's view on The Curious Case of the Commute: Why Your Interview is Back in 3D? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

By Gagan Malik

In 2024, nearly half of all job seekers used generative AI to game their way through applications and interviews. Half. Employers are not bringing back in-person interviews because they missed you, your firm handshake, or the small talk about your commute. They are bringing them back because your AI is getting the job, and then you have to show up and actually do it. In-person interview requests jumped from 5% to 30% in a single year. That is not a trend. That is a correction. [worklife](https://www.worklife.news/talent/return-in-person-job-interviews/)

#### The Arms Race Nobody Won

The received wisdom was that virtual interviews were efficient, equitable, and the future of hiring. Wrong on all three counts. What companies discovered is that they had built a system so gameable it was effectively handing offers to algorithms. Google's CEO Sundar Pichai went on the Lex Fridman podcast and said his company is reinstating in-person rounds "just to make sure the fundamentals are there". That is the world's most sophisticated tech operation saying, in plain English: we have a fraud problem, and we cannot solve it over Zoom. Brilliant self-diagnosis. Expensive lesson. [hindustantimes](https://www.hindustantimes.com/business/ai-is-forcing-the-return-of-the-in-person-job-interview-101754992294477.html)

#### The Numbers Do Not Lie, But Candidates Do

#### Six Percent Is a Lot

A Gartner survey of 3,000 job seekers found that 6% had participated in active interview fraud, meaning someone or something else took the interview for them. Gartner also predicts that by 2028, one in four candidate profiles globally will be fake. Think of the virtual interview as a nightclub bouncer who cannot see the queue. The door is open. The bots are getting in. [hindustantimes](https://www.hindustantimes.com/business/ai-is-forcing-the-return-of-the-in-person-job-interview-101754992294477.html)

#### The Silent Drop-Off

Cisco added in-person rounds after repeatedly encountering final-stage candidates with "something that feels off," in the words of chief people officer Kelly Jones. What proved most diagnostic: the moment the in-person requirement gets mentioned, some candidates simply vanish. Cisco reports this happens regularly. Ghost is the new rejection letter. [hindustantimes](https://www.hindustantimes.com/business/ai-is-forcing-the-return-of-the-in-person-job-interview-101754992294477.html)

#### The McKinsey Tell

McKinsey began mandating at least one in-person touchpoint roughly 18 months ago, originally to better assess rapport-building skills for client-facing roles. The rise of AI cheating deepened that commitment. When McKinsey changes a process it is not for sentimental reasons. It is because the old process stopped producing reliable outputs. Follow the logic, not the nostalgia. [hindustantimes](https://www.hindustantimes.com/business/ai-is-forcing-the-return-of-the-in-person-job-interview-101754992294477.html)

#### I Was Part of the Problem

I will own this. Around 2022 I hired a senior creative lead on the strength of three immaculate video interviews. Articulate, sharp, camera-confident. He showed up to his first live client presentation and the client rang me privately afterwards to ask if something was wrong. Same person. Different altitude entirely. I spent the next six months and roughly £40,000 in recruitment fees, notice-period costs, and client-relationship repair learning that I had been screening for the ability to perform on camera, not the ability to do the job. I thought I was running a modern, efficient hiring process. I was running an audition for a presenter role that did not exist.

The broader absurdity is this: the companies now insisting you commute in for a final-stage interview are largely the same ones who spent 2020 to 2023 outsourcing their entire hiring stack to keyword-matching algorithms that rejected candidates based on CV formatting. They created the fraud vacuum by removing human judgment from the process. Now they want you on a train at 8am to restore it. The audacity is, at minimum, consistent. [worklife](https://www.worklife.news/talent/return-in-person-job-interviews/)

#### Kill, Keep, Create

The future is not a wholesale return to three-panel interview rooms with weak coffee and motivational posters. Gartner data shows 68% of interviews remain virtual and that is not reversing. But the direction of travel is settled. [worklife](https://www.worklife.news/talent/return-in-person-job-interviews/)

#### Details

- Kill: The assumption that a fully virtual hiring process is inherently rigorous. It was always a cost-saving measure dressed in the language of innovation.

- Keep: Virtual rounds for early screening. Fast, inclusive, no room booking required.

- Create: A structured hybrid model where in-person is mandatory at the final stage, with candidate travel reimbursed. Cisco and McKinsey are already here. The FTSE 100 follows by 2027, quietly, without a press release. [hindustantimes](https://www.hindustantimes.com/business/ai-is-forcing-the-return-of-the-in-person-job-interview-101754992294477.html)

#### Polish the Shoes

Stop treating an in-person interview as a logistical inconvenience and start treating it as the only round that counts. Practice making eye contact without a screen as your emotional buffer. Charge the Oyster card. Rehearse the handshake. And if the prospect of sixty minutes in a room with a hiring manager, with no AI whispering answers in your earpiece, genuinely unsettles you, that reaction is the most honest performance review you will ever receive.

The commute was always the warm-up. Now it is the test. Welcome back to the room.

---

### We got accepted into Techstars
[Read more](https://www.gaganmalik.io/en/newsroom/techstars-riyadh-2023-class)

[Retrieval brief — techstars-riyadh-2023-class]
Category: business
Thesis: Techstars kicks off its second program in Riyadh with 12 startups from Saudi Arabia, the wider Middle East, and Europe—innovating in AI, EdTech, eCommerce, and HealthTech.
Answers: What does Gagan argue in "We got accepted into Techstars"? | What is Gagan Malik's view on We got accepted into Techstars? | How does this essay relate to product strategy and growth?
Concepts: business, Gagan Malik, writings

Techstars kicks off its second program in Riyadh with 12 startups from Saudi Arabia, the wider Middle East, and Europe—innovating in AI, EdTech, eCommerce, and HealthTech.

---

### When travel guides meet food-lover loyalty
[Read story](https://www.gaganmalik.io/en/newsroom/telegraph-travel-waitrose-loyalty)

The Telegraph’s Travel Guides app and Waitrose’s My Waitrose club do not look alike. One lives in your pocket for a long weekend; the other lives in your weekly shop. The design question behind both is the same: how do you turn a brand people already respect into a relationship that repeats, without cheapening the story you are known for?

---