What Franz Kafka, You, and I Have in Common

There is a particular kind of confidence possessed by the clerk behind the counter when the clerk knows there is a rule and you do not.

The clerk may not have written the rule. The clerk may not understand the rule. The clerk may privately consider the rule idiotic. None of this weakens the clerk's position. The rule exists somewhere, perhaps in a binder, perhaps in a system, perhaps in the mind of a supervisor who is at lunch, and for the moment the clerk stands on the side of it while you stand on the other.

You have brought the wrong form.

Or the right form with the wrong date.

Or the right form with the right date, submitted through a door intended only for people carrying the wrong form.

You ask what you should do next.

The clerk looks at you with the mild disappointment of someone watching a squirrel attempt a mortgage application.

This is Kafka's territory.

His characters live in worlds where the machinery of authority is everywhere and its explanation is nowhere. In The Trial, Josef K. is arrested, though no one will tell him what he has done. In The Castle, a land surveyor named K. is summoned by an authority that becomes harder to reach each time he tries to approach it. Offices proliferate. Messages arrive. Procedures are followed. Someone always seems to know what is happening, though that person is invariably in another room.

Kafka is often described as a writer of the absurd, which is true in the same way that describing the ocean as damp is true. The absurdity matters, but what makes his work endure is the orderliness of it. His worlds are not chaotic. Chaos would almost be a relief. They are administered.

The doors have labels. The officials have titles. The papers are stamped.

The system is functioning.

It is simply not functioning for you.

This is what Franz Kafka, you, and I have in common.

We spend our lives inside systems that reveal their decisions more readily than their reasons.

A job application disappears into a portal. A loan is denied. A post is shown to twelve people, three of whom appear to sell fractional ownership in laundromats. A search engine elevates one page and buries another. An AI system produces a polished answer with the serene tone of a person who has never once been required to show his work.

The result arrives.

The explanation does not.

This used to be an occasional inconvenience, the sort of thing encountered at the motor vehicle office or during a dispute over an insurance claim. Now it is a basic condition of modern life. More of our decisions are mediated by systems whose rules are hidden, shifting, proprietary, automated, or all four. They decide what we see, what reaches us, what is recommended to us, and increasingly what is done on our behalf.

We are told these systems are sophisticated.

This is usually true.

We are also told they are personalized.

This may be true in the same sense that a hotel room is personalized because the television says your name.

What we are rarely told is why this result appeared instead of another.

Imagine my surprise over the last twenty-four months as AEO began supplanting SEO in our leads at Acadia, and I found myself feeling a bit like Gregor Samsa. I had not awakened as an insect, exactly, but I had awakened inside a profession whose name, borders, and apparent purpose were changing while I was still in it. The work I had spent decades learning had not disappeared. It had simply been redescribed overnight by people eager to announce that the old thing was dead and the new thing had arrived, often before anyone could agree on what the new thing was. Clients wanted answers about systems none of us fully understood. Vendors arrived with dashboards that made uncertainty look attractively sortable. Every week produced a new framework, usually with a diagram and a proprietary noun. My job remained what it had always been: study the results, infer the causes, and resist confusing confidence with knowledge.

This is not unique to search. It is what happens whenever a system changes faster than its explanations.

The output is visible. The mechanism is hidden. Someone succeeds, something surfaces, a recommendation appears, a customer chooses, and the rest of us gather around the result looking for the rule.

When the rule is unavailable, imitation rushes in.

A page ranks with two thousand words, so two thousand words becomes the requirement. A successful post begins with a confession, so the internet fills with executives confessing that they once cared too much about efficiency. An AI answer cites Reddit, and suddenly every company wants a Reddit strategy, preferably one that does not require becoming part of Reddit.

We copy the visible feature because the cause is hidden.

This is not foolish. It is what people do with incomplete evidence. The mistake comes later, when the observation hardens into law.

The page may be long because the subject requires length. It may rank because the domain has been trusted for fifteen years. The hired candidate may have used a certain resume format, or she may have worked with the hiring manager's former boss. The proposal may have won because of its forty slides, though it is also possible the client had decided before slide one.

The visible thing is real. It just may not be the important thing.

This is how superstition enters business.

A pattern appears. Someone names it. Someone else turns it into a process. The process acquires a dashboard. The dashboard survives long after the behavior it was built to measure has moved somewhere else.

By then, the original mystery has been replaced by a cleaner one: why the numbers look healthy while the business feels ill.

We like rules because rules spare us from judgment.

Judgment is slower. It asks us to hold more than one explanation in mind. It asks what else might have produced the result. It asks whether the pattern survives across different cases, whether it breaks under pressure, whether we are looking at the cause or merely at what the cause happened to leave behind.

The fisherman who catches three fish with a red lure would be foolish not to notice the lure. He would be equally foolish to conclude that red is now the permanent color of fish.

The distinction is small and enormous.

A visible feature may be part of the cause. It may be a byproduct of the cause. It may have nothing to do with the cause. It may have worked only because of ten conditions that no longer exist. Yet the feature is easy to copy, and the conditions are difficult to see, so the feature becomes the tactic and the tactic becomes the explanation.

This is the trap.

The work of finding is not the work of discovering a permanent rule. There may not be one. It is the work of building a provisional explanation from incomplete evidence, then trying to break it before the world does.

That requires a temperament less like the obedient clerk and more like the detective.

Not because the detective always knows the answer, but because the detective understands that an answer is only as good as the alternatives it has survived.

A result is evidence.

It is not revelation.

This matters when we are trying to find something, and it matters equally when we are trying to be found. In one direction, we look at the world as it is and reason backward toward the forces that produced it. In the other, we decide what effect we hope to create and work backward toward the conditions that might make it possible.

The investigator asks, "What caused this?"

The builder asks, "What would cause this?"

Between those questions sits almost everything worth knowing about discovery.

A search engine leaves patterns in what it rewards and abandons. A market leaves clues in what people compare before buying. A customer leaves hints in the questions she asks twice. A failed proposal leaves small disturbances in the room: the objection that arrived too early, the detail everyone praised too warmly, the silence after the price.

An AI answer leaves citations, omissions, repetitions, and peculiar preferences. A recommendation system leaves a trail through what it keeps showing and what it seems unable to see. An organization leaves evidence in the work it funds, the work it praises, and the work it quietly makes impossible.

None of these clues is the rule.

Together, they give us somewhere to begin.

The danger is not merely that systems hide their rules. It is that hidden rules invite false certainty. They tempt us to treat the first plausible explanation as the final one, to mistake a tactic for a principle, a dashboard for understanding, a polished answer for a true one.

That temptation is growing stronger.

The systems are becoming faster, more fluent, and more willing to act before we have finished asking what they know. They will produce answers, recommendations, rankings, summaries, and decisions with increasing confidence. They will shorten the distance between the question and the result until the path itself disappears.

When that happens, judgment becomes more important, not less.

We will need to notice what the system selected, but also what it excluded. We will need to ask which signal mattered, which one merely traveled beside it, and whose interests were encoded in the choice. We will need to resist the pleasure of a clean explanation when the evidence remains untidy.

Kafka imagined the horror of a system that would not explain itself. Our systems are usually less complete, more commercial, and more badly organized, which is both less elegant and more hopeful. They contradict themselves. They reveal preferences through repetition, omissions, exceptions, and mistakes. They leave fingerprints on every result they produce.

The system will not hand us its rulebook.

It will offer something more dangerous: a result polished enough to make us stop looking.

Our work is to keep looking anyway.