
By Gagan Malik
If you run a product team, you can say by lunchtime how many people used your product yesterday. You probably cannot say why any one of them came. In 1988 Russell Ackoff gave his presidential address to the International Society for General Systems Research. bernstein-dikw In it he described the content of the mind as a hierarchy of five levels, which I will call floors. "Descending from wisdom," he wrote, "there are understanding, knowledge, information, and, at the bottom, data". ackoff-data-wisdom fricke-dikw
Most of the pyramids drawn since have four floors. fricke-dikw bernstein-dikw The one that went missing is understanding, the floor that answers why. I am a product designer, and my working belief is that people hire products and services to get a job done. This essay argues that the job lives on the missing floor. It also argues something less comfortable. Understanding is what a team's slow work was buying. The slow work is the easiest thing to cut, because losing it does not show on the numbers.
Ackoff gave the floors questions. Data are symbols that record what happened. Information answers who, what, where, when and how many. fricke-dikw In another paper he put the next two plainly: knowledge is contained in instructions, answers to how, and understanding in explanations, answers to why. ackoff-systems-understand Wisdom sits above them, and each floor contains the ones beneath it. fricke-dikw Bernstein reads Ackoff's understanding as the ability to assess and correct errors. bernstein-dikw The two readings meet: you can only correct a mistake once you know why it happened.
Most later writers simply never picked the floor up. bernstein-dikw A few removed it on purpose. Bellinger, Castro and Mills argued that understanding is not a level of its own but what carries you from one level to the next. bellinger-dikw I think they are right, and it is the best reason to keep the floor in the drawing. Moving between floors is work, and work takes time. Ackoff closed with percentages nobody had measured, among them ten for understanding and "virtually no wisdom". bernstein-dikw A precise-looking number that nobody measured is the commonest object in my trade, and it usually sits on a slide.
A product analytics tool lives on the second floor, and it is very good there. It tells you who signed up, what they tapped, when they left and how many came back. An experiment climbs higher and can tell you why a number moved. None of it tells you why the person came at all.
Christensen and his co-authors define a job as "the progress that a person is trying to make in a particular circumstance". christensen-luck People hire a product to help with it. hbr-jobs-to-be-done In June I argued that usage metrics miss whether the job got done, and Ackoff's model shows where the miss happens. A job is a claim about a person's reasons, which is different from a cause in a funnel. It sits two floors above the dashboard.
Ezra Klein visited Nvidia's headquarters in Santa Clara to interview Jensen Huang, for an episode published on 23 September. klein-huang Huang's vision was that soon we will "know everything and do anything". On Ackoff's ladder, that is a promise about the third floor. Then Huang made a distinction about work: every job has a purpose, and there is the task you do as the job. klein-huang A radiologist's task is studying scans in a dark room. Her purpose is to diagnose disease and help patients find out what is wrong.
He went further with engineers. Engineering, he said, exists to solve problems and "connect the social need with the technology that exists". klein-huang That is as good a description of product design as I have read, and he offered it about a different profession. I should be plain about what I am doing with it. Huang's job is the worker's, and Christensen's is the customer's. They are different jobs, and I am putting them on the same floor on purpose.
What follows is my reading, not Huang's and not Ackoff's. The customer hires a product for a reason. Somebody on the team has to understand that reason, and that understanding is the purpose behind the screens, variants, copy and test scripts. Those tasks are the parts a team can now speed up. Huang was honest about the other side. Where the job and the task are one, as with phone customer service, the job itself can go. klein-huang
I am not a neutral witness. Researchers, product managers, analysts and salespeople all claim this floor, and some hold it better than designers do. It belongs to whoever does the slow work, and time alone does not earn it: plenty of slow research ends in a deck nobody reads.
Klein brought a study to the interview. klein-huang Its authors followed 26,811 students in grades seven to twelve in one Chinese county for 30 months. cepr-learning-penalty demombynes-learning-penalty They compared students who had started using AI with those who had not yet. AI raised homework scores by 18% and cut completion time by 30%. Within six months, monthly closed-book exam scores fell by 20%. High-stakes entrance exam scores fell by 18% and 24%, and the full penalty took about two years to appear. cepr-learning-penalty
The detail that matters is underneath. About 80% of the students using AI showed signs of handing the homework over. Those who kept spending about as long on it lost little. cepr-learning-penalty The paper is not yet peer-reviewed, and some students may have turned to AI because something else in their lives went wrong. demombynes-learning-penalty Exams test the third floor, not the fourth, so this shows what happens when the result holds and the time goes. In July I spent a week at a dental practice's empty reception desk, reading call transcripts before I would recommend a tool. The owner had asked me to stop the double-bookings, not to sit there, so that week she did not get the switch she wanted. I got three gaps the voice agent had missed, and she later said finding them cost less than the agent. Product teams make the same trade when they cut that week. In a 2024 survey of people who do research, 22% said their company had laid off researchers that year. user-interviews-2024 The report's authors saw teams trying to "do more with limited headcounts". The survey's publisher sells research recruitment, so it has a stake in the finding.
A broken wrist goes into plaster until the bone heals. The cast does exactly what it promises. It holds the bone still, and the bone knits.
Then the cast is cut off, and you cannot turn a key in the door. The bone has healed, the wrist is stiff, and the muscles have gone weak from weeks out of use. It happened a little each day, because the cast was doing their work. Nothing hurt, and nothing looked wrong until the support came away.
A controlled experiment answers why better than any interview, and analytics tools now write their own explanations of a drop. Interviews are small samples of what people say, and people misremember, perform and explain themselves after the fact. Synthetic users, AI models answering as customers, promise the interview in minutes. Huang would add that losing a skill is not losing the job: we will lose some finer intellectual dexterity, he said, and become better systems thinkers. klein-huang
Much of that is right. An experiment answers why a number moved better than I can. The students who kept their time lost little, so the tool is not the villain. But an experiment tests a reason somebody has already named. On 21 June 2024, Maria Rosala and Kate Moran, writing for Nielsen Norman Group, compared synthetic users with three real studies. nng-synthetic-users The synthetic learners said they had finished their online courses. Many real learners had not. Huang also sells the chips, so he has reason to say machines take tasks rather than jobs. Klein put the limit plainly to Huang: "It can't be the case that everything can be traded off". klein-huang The students' most useful instrument was not the score. It was the clock.
Ackoff's fourth floor holds two jobs: the customer's reason for hiring the product, and the work of understanding it. Speeding up the floors beneath has a price: people whose job was the task pay it now, and teams that saved the slow work pay later. Among the hardest hit were the high achievers, and nobody saw it until the entrance exam took the support away.
Plans देखें, या सवाल पूछें।
नए लेख और अपडेट अपने आप पाने के लिए इस लिंक को अपने RSS रीडर में जोड़ें।