The Tailor Who Blamed the Customer

A tailor makes two jackets that don't fit and concludes that the customer is the problem. Institutions, economies, and AI systems repeat his mistake whenever they treat dissatisfaction as a defect in people rather than evidence about the design.

The Customer Who Would Not Fit

A tailor makes a jacket for a customer. It is too large. The customer complains, so the tailor makes another. This one is too small.

After the second complaint, the tailor reaches a conclusion: the customer is impossible to satisfy.

It does not occur to him to take better measurements. Instead, he begins to wonder what is wrong with the person standing in front of him. Perhaps the customer is ungrateful. Perhaps they do not know what they want. Perhaps they should simply make more effort to fit a perfectly reasonable jacket.

So the tailor stops trying to make a jacket that fits the customer. He starts trying to make the customer fit the jacket.

The absurdity is easy to see. The tailor has reversed the purpose of his work. The jacket exists for the person, not the person for the jacket.

The same reversal is much harder to see when the jacket is an institution, an economic arrangement, or a technology. A system delivers what its designers believe people should want. People remain dissatisfied. Their dissatisfaction is then treated as a flaw in them: they are lazy, irrational, ungrateful, uneducated, or resistant to progress. Rarely does anyone begin with the simpler possibility that the system misunderstands the people it was built to serve.

Not every complaint is justified. People can be wrong about what they need. But a system that can explain every failure by blaming its users has made itself incapable of learning.

A Paradise Built on the Wrong Measurements

In The Matrix, Agent Smith tells a captured Morpheus that the first Matrix was designed as a perfect human world without suffering. It was a disaster. Humans would not accept it. Smith concludes that human beings define their reality through suffering and misery.

It is a bleak verdict, and the film lets him deliver it with total confidence. But Smith also mentions, almost in passing, that another explanation was available. Some, he says, “believed we lacked the programming language to describe your perfect world.”

That is the tailor’s choice in miniature. When the design fails, either the designers misunderstood what they were designing for, or the people are defective. Smith chooses the second explanation. It is the more flattering one for the machines.

But consider what a designed paradise might actually contain.

Every desire is satisfied immediately. Every obstacle disappears. Every uncertainty is removed. Every experience is pleasant.

Someone wants to climb a mountain. The system places them at the summit.

Someone wants to build a beautiful house. The system presents the finished house.

Someone wants to become an excellent musician. The system gives them the experience of performing flawlessly.

Each request has been fulfilled. Yet the climber never climbed, the builder never built, and the musician never learned to play.

The philosopher Robert Nozick made a similar point with his “experience machine,” a device that could give you any experience you wanted while you floated in a tank. Most people, he argued, would not plug in for life, because we want to do things, not merely have the experience of doing them. We want to be a certain kind of person, not a body in a tank that feels like one.

The designed paradise does not fail because it makes life too good. It fails because its definition of a good life leaves too much of life out.

The Summit and the Climb

None of this means people need suffering.

Hunger does not make food meaningful. Preventable illness does not become valuable because recovering from it takes courage. A great deal of human misery is simply misery, and a system that removes it has done something good.

The distinction that matters is between hardship imposed on someone and difficulty they choose in pursuit of something they value.

“I want to reach the summit” can mean “I want to be up there.” It can also mean “I want to climb.” The destination and the activity are connected, but they are not interchangeable. A system that cannot tell them apart will deliver the wrong thing while believing it delivered exactly what was asked.

A humane system removes imposed hardship without assuming that every chosen difficulty should be removed too. It does not decide on someone’s behalf which effort was the point.

Care Is Not Control

Consider a child who asks for chocolate.

A good parent does not have to give the child every chocolate in the world. The parent can give some, save some for later, and explain why eating chocolate all day is not a good idea. The parent can set limits without treating the child’s desire as a defect.

If the child is still unhappy, the parent can try to understand why. Perhaps the child is tired. Perhaps they wanted to share the chocolate with a friend. Perhaps they are asking for something else and do not yet have the words for it.

Now imagine a different parent. This one refuses the chocolate entirely and insists that the child does not need it. When the child protests, the parent hands over a mountain of chocolate to end the noise. When that creates new problems, the parent takes everything away and decides the child is impossible to satisfy.

This parent swings between deprivation and excess without ever asking what would actually serve the child. The child’s desires are not the problem. The problem is that the child is being managed rather than understood.

The analogy has a limit, and it matters. Children really are entrusted to their parents’ judgment for a time. Adults are not entrusted to a designer’s judgment. Institutions and technologies routinely borrow the language of care—we know what is good for you—to claim an authority they were never given. Even good parenting aims at a child who will one day decide for themselves. A system that treats adults as permanent children has mistaken its purpose.

Good care requires judgment, observation, adaptation, and respect. It does not license unlimited control.

Help That Leaves You Able

Artificial intelligence makes these questions practical rather than philosophical.

We are approaching a world in which a person can describe a goal and have an AI system perform much of the work: building software, designing a house, running research, starting a company. Much of this is a genuine expansion of freedom. People should not have to master every discipline before they are allowed to pursue an idea.

But a single request can contain different ambitions.

If I ask an AI to build a software engine, I may simply need the engine. I may have no interest in programming, just as someone who commissions a house need not want to become an architect.

Or I may want something more. I may want to understand the architecture, test my own judgment, discover solutions, and become someone who can build things I could not build before.

A finished product serves the first purpose. On its own, it does little for the second.

So the design question is not whether AI should do the work or make me do it. It is whether the system can tell the difference between work I want to hand off and participation I want to keep.

That distinction has to remain mine. A system that turns every request into a lesson is as presumptuous as one that completes everything without asking. People sometimes want instruction, sometimes collaboration, and sometimes relief from a job they would happily never do again.

Good assistance responds to those differences. It can deliver a result, explain its reasoning, invite participation, or step aside. It makes competence available without making competence the price of help.

AI should make it easier to become capable, not merely easier to remain dependent. But it should never insist on capability when what a person wanted was help.

The Prince Who Could Only Command

Imagine a prince who can obtain almost anything by giving an order. Houses are built, meals prepared, disputes settled, accounts managed. His wishes have enormous consequences, yet he understands little of how they become real.

Now imagine a second prince with the same resources who knows enough to question advice, evaluate results, and notice when the people serving him are pursuing interests of their own.

Both can command. They do not have the same control.

The second prince’s advantage is not that he can do everything himself. Nobody can. Human capability has always depended on other people, shared knowledge, tools, and institutions. Total self-sufficiency is neither realistic nor desirable.

What matters is the character of the dependence.

A patient does not need to become a surgeon to keep their agency. But informed consent, a second opinion, and professional accountability make dependence on a surgeon very different from submission to an authority that cannot be questioned.

The same holds for any powerful system. Its users do not need complete technical mastery. They do need to understand enough to make real decisions, to challenge an answer, to seek an alternative, and to obtain redress when something goes wrong.

The first prince has the freedom to request things. If the mechanisms he relies on fail, change their terms, or quietly begin serving someone else, he will discover that the freedom to request was never the same as the freedom to direct his own life.

When Complaints Become Evidence

The tailor’s deeper failure is not bad craftsmanship. It is his refusal to let the customer’s experience correct his understanding.

His model says the jacket is right. The customer says it does not fit. Rather than revise the model, he discredits the evidence.

Institutions develop the same habit. A service is declared accessible, so anyone who cannot use it must be careless. A procedure is declared fair, so anyone disadvantaged by it must have failed to follow instructions. A product is declared intuitive, so confusion becomes a defect in the user.

The habit is dangerous because it protects the system from learning. If every dissatisfied customer is unreasonable, the tailor never improves. If every person who rejects a manufactured paradise is defective, the designers never question their idea of paradise. The system loses the ability to distinguish among three very different people:

  • someone who needs help;
  • someone whose needs differ from what the designers assumed;
  • someone responding rationally to a genuinely bad arrangement.

None of this means every complaint is conclusive. People misjudge their own interests. Their requests conflict with each other and with other people’s. Some limits are necessary, and necessary limits disappoint.

But those possibilities are reasons to investigate, not reasons to dismiss. Dissatisfaction should begin an inquiry, not end one.

What Listening Requires

Goodwill is not enough. A system can sincerely claim to listen while leaving its designers free to ignore everything they hear.

Listening that changes anything needs practical mechanisms:

  • Explanation. People can understand what the system decided and why.
  • Appeal. Decisions can be contested by someone with the authority to reverse them.
  • Alternatives. People can leave, or choose a different arrangement, without losing the conditions of a decent life.
  • Independent scrutiny. Someone other than the system’s beneficiaries checks whether it works.
  • Influence. The people affected have a real say in how the system changes.

Sometimes listening improves the service. Sometimes it reveals that the designers should hold less control than they do. A system that only ever concludes the first has not really listened.

The Real Measure of a System

We usually judge systems by what they produce: goods delivered, time saved, wealth generated, suffering prevented. These measures matter. A system that promises empowerment while failing to meet basic needs deserves no praise.

But outputs do not tell the whole story. We also have to ask what happens to the people living inside the system.

Do they become more capable or more dependent? Do they gain meaningful choices, or merely more options chosen by someone else? Can they understand consequential decisions, pursue purposes the designers never anticipated, challenge the rules, and leave if they must?

Abundance and agency can reinforce each other. They can also come apart. A system can give people more while leaving them less say over how they live.

The answer is not to romanticize struggle or to make people earn assistance through personal mastery. It is to recognize that being well served includes having a voice in what service means.

The tailor does not have to satisfy every imaginable demand. He has limits of material, cost, and skill. But he must remain willing to take new measurements, to explain his limits honestly, and to admit when his work does not fit.

What he cannot do is declare the customer’s body a mistake.

When a system repeatedly fails the people it was built to serve, the first question is not why they are so hard to satisfy. It is whether anyone ever took their measurements.

For a fuller discussion: Read the extended version →

Back to all essays