The question “how do I stay relevant?” is honest.

Form follows function? The inductive prejudice and the stories that help us see it.

On how a phrase coined by an architect in 1896 describes, backwards, what happens today in computing systems trained on data.

A note on vocabulary. In this text I do not use the phrase "artificial intelligence" or its abbreviation (AI), and I avoid expressions that attribute states of mind or intentions to a machine: "thinks", "understands", "knows", "wants", "sees", "chooses", "advises", "keeps company". I speak of computing systems, statistical models and conversational interfaces, and I describe what they produce, calculate or reproduce, not what they "feel" or "understand". I keep only the established technical names ("machine learning", "neural network", "training") and the titles of the works cited, so as not to falsify sources. The characters in fairy tales think and feel; analogies with them concern only the structure of a mechanism, not a mind of the machine.

A workshop, a phrase and a reversal

My father repaired engines. His workshop smelled of oil, had a lamp hanging above the workbench, a radio always on, and tools laid out in an order that, to me as a child, seemed incomprehensible. It was not an order by labels or alphabet; it was his. And yet, when he needed a spanner, his hand knew where it was, without looking.

If I asked him why he kept them that way, I got no answer, but a phrase he said often, in front of anything broken: "It can still be mended." The workshop's form had not been designed to a plan. It had settled, over years, after his hands had done the same things thousands of times.

I often think of that workshop when I read about computing systems trained on data, because it poses, on a small scale, a much larger question: who follows whom — does form follow function, or function form?

The formula that put the question into circulation belongs to the American architect Louis Sullivan. In 1896, in an essay about the tall office building, he wrote that form ever follows function.

The formula became the slogan of modernism: first you know what a thing is for, then you give it the appearance that suits it. A chair has four legs because it must stand, not because four legs look nice.

In systems trained on data, this relationship turns upside down. And that is precisely why a whole series is worth writing about it.

Let me explain why. When you build a system by training, you cannot know in advance, in the strong sense of the word, what function it will end up calculating. You know what function you want to obtain: to classify pictures of cats, to produce a translation, to estimate the weather. But the function it actually calculates — the one it reproduces on data encountered only after training — depends on a series of choices of form: the network's architecture, the training procedure, the data used, the constraints imposed. Engineers have a precise name for this set of choices. They call it inductive bias, that is, inductive prejudice.

The word "function" has, in addition, two senses that meet here. One is mathematical: a machine-learning model is, in essence, a function mapping inputs to outputs, and the training procedure selects, from among countless possible functions, a suitable one. The other is social: what do people, in fact, do with this technology? The two senses do not overlap, but they influence one another, and much of this text tries to hold them in the same frame.

So that the discussion does not remain abstract, I did something I often do in my texts: I called upon the Romanian fairy tale for help. Not as decoration, but as a mnemonic. A reader of seventeen or twenty has almost certainly met, at school, Harap-Alb (the hero of Ion Creangă's classic Moldavian fairy tale), the two daughters in Creangă's tale, or Făt-Frumos — the Romanian "Prince Charming" — who sets out after "youth without old age". These images are already in their head. They can be used as hooks on which to hang ideas that would otherwise slip away.

One warning, though, so as to be fair to the fairy tale. The analogies in this text are interpretations, not technical equivalences. Harap-Alb is not a neural network and was never thought of as one. Where the analogy holds, I use it; where it breaks, I say it breaks. One further point of precision: Harap-Alb and the tale of the daughters are literary fairy tales, written by Ion Creangă on popular motifs, while Master Manole is a ballad and legend, not a fairy tale proper. I gather them under the same roof anyway, because they work as shared images of a culture.

The text follows eight steps. It begins with what the inductive prejudice concretely is and where the idea comes from. It continues with the way modern architectures build in hypotheses about the world. Then it passes through the field's most important dispute, Rich Sutton's "bitter lesson" and the replies to it. It stops at what we know, publicly and at this moment, about what functions these systems in fact fulfil in current use. And it closes with the limits and the costs of form.

The fence in the park.

On freedom, limits and speed, from the fairy tale to automated computing systems.

In brief Freedom is not the absence of the fence, but the bounded space in which danger has been excluded in advance. From this follows a simple and uncomfortable thesis: the speed at which we can move, in a story or in a technology, depends on the quality of the limits we have set. Not on their number, but on how visible, proportionate, easy to revise and well understood they are. I start from a dog park, pass through the prohibition in fairy tales, the sacrifice in the legend of Master Manole and the rites of initiation, arrive at the mechanisms by which automated computing systems are constrained today, and stop at the cost of the fence, which nobody gets for free.

"I take my little dog to the dog park because I want him to have freedom, but I don't want him to run into the street. The dog sees the fence and does not go past it. He can play freely, and I know he is safe. The key to moving fast [in AI development] is having the right guardrails."

(The text from which this essay sets out, reproduced verbatim. Keith Fulton, Chief Data Officer and Chief AI Officer at the fintech company Jack Henry)

The cursor that blinked

In 1997 I spent six months in a room in Botoșani where, strictly speaking, I had no business being. I had enrolled in a two-week course, "IBM PC Operator". The course ended, but the room stayed open. There were two computers that worked and one taken apart, an instructor who chose not to notice that I came in every day, and a black screen bearing a single line, C:\>, followed by a cursor that appeared and disappeared every half second.

Nothing appeared on that screen of its own accord. If I did not type, nothing changed. If I typed an exact command, the command was executed exactly. If I got a letter wrong, a terse message appeared, without explanation and without tone, and I was left to work it out. It was, as I would say now, a pure tool: everything that happened began with me and stopped where I stopped.

And yet, even there, on so passive a machine, there was a fence.

If I gave the command that deletes all the files in a directory, the program did not execute it on the spot. It displayed a question, in English, whether I was sure, and waited until I pressed a letter: yes or no. If I tried to prepare the disk for fresh use — which meant erasing everything on it — the message was more emphatic still: all data on the disk will be lost. None of this stopped me doing as I pleased. The fence was not a prohibition. It was a second gate, placed exactly in front of the one gesture that can no longer be undone, and it asked one small thing: to repeat, consciously, what a moment earlier I had done in haste.

Today I believe it was precisely that question which let me experiment. On a computer that everyone in the room depended on, a beginner like me would have been a danger if every mistyped command could have wiped everything with no brake. As it was, I could type fast, I could try commands I did not yet understand, I could read, err, start again. The speed at which I learned in that half-year did not come from an absence of limits. It came from the fact that the one place where a mistake cost a great deal was guarded.

Years later, when I began working with websites and far more complicated machines, I recognised the idea under other names: backup copies, access rights, staging environments. All are variants of the 1997 question. They do not say "don't do it"; they say "do it, but not where you can no longer turn back."

Everything that follows is about this simple structure.

Between Gomorrah and Ninive: why a broken world sometimes receives one more year

Ninive, not Gomorrah — Part I of III

In brief Gomorrah did not burn because it had sinned; Ninive had sinned too. It burned because ten just men could not be found in it, and no one turned around. A second chance is not an amnesty, but one extra year, with the spade in your hands. A class, a block of flats, a city or an internet are held up, in that year, by a few. The web will not be saved by everyone. It will be saved by those who keep writing, checking and repairing.

It can still be repaired

The table has been cleared of plates, but nobody gets up. At one end, a friend of my son's turns his phone face-up on the table: a white web of cracks across the whole of the glass. "It doesn't matter, I'm replacing it anyway." He says it without sadness, the way you say it's raining outside.

At the other end, a parent I know from the class meetings takes up the question we all went home with last time: at what age do you get your child their first phone? An hour was spent on it at the meeting. Another week was spent on it on the class WhatsApp group. No conclusion was reached.

Between them, the talk slides, as it usually does at tables of this kind, towards homework. Someone tells, laughing, that last week's essay "came out" in ten minutes, with the assistant. Nobody is surprised. The parent at the other end is quiet for a moment. His child is ten, and in two or three years he will have the same tool in his pocket.

I sit between those three sentences and think of my father's workshop.

My father was a mechanic. He has not been for a long time, but I can still see him bent over an engine everyone else had written off. He was in no hurry to take it apart. First he listened to it and thought, waited a day, two, said something to me about the fault, opened a discussion with other mechanics about a model. Then he repaired it, and the engine started. His sentence, the one I carry with me, was simple: "It can still be repaired."

I inherited the gesture, though not the tool. I look after old websites that anyone would rewrite from zero, and before I touch them I listen to them too, the way he listened to the engine: what still works, what has broken, what holds to what. I can listen to them because they are open: their code can be read line by line, like an engine without a bonnet. Sometimes it is harder than building anew. It is almost always more honest, though it is not paid the same: a rewritten site is invoiced more easily than a listened-to one.

The cracked phone, the phone not yet bought and the essay written by an assistant belong to the same culture: the culture of replacement. The broken thing is thrown away, the tired application is abandoned for a new one, the platform is left for the next one. Last year's predictive model is declared obsolete by the one launched this month, trained anew on everything we have written in the meantime. And the effort of understanding a subject is replaced by a ready-made answer.

The virtues, in the reverse order of the damage: from gentleness to humility

Ninive, not Gomorrah — Part II of III

In brief In "The Mechanical Bird", man was damaged in order, from pride to anger. Repair goes the other way: it starts with the gesture closest to hand, the pause before anger, and descends to the root, humility. Each virtue is a small daily gesture that no tool can make in your place. And each has a counterpart in the way we build the internet, train the models, and share what we know.

The engine is repaired backwards

My father — your friend's grandfather — had a rule: to repair something, you walk back along the road by which it broke. He said it one evening, relaxed, after coming back from a friend who was a mechanic. An engine that stops did not break all at once. The last thing to give way is the first thing you see, but it is rarely the one that started it all. If you change only the burnt part, the engine stops again a week later. You must go back, part by part, until the first one.

I say it as a household story. You who are parents can take it home as such: when something breaks between your child and a friend, on the class group chat, the road back is the same. Not the last message, but the first.

In my trade the rule has a drier name: the version history and the error log. When a site goes down, I don't rewrite it. I read backwards, change by change, until the one that broke it. I can do that because I work with open code, whose history is kept and can be read by anyone. A closed advanced predictive model — what is usually called AI, after the definition in "The Mechanical Bird" — does not give you this chance. You don't know what changed inside it, nor when, nor why. And a hurried economy has no patience for the road back: it is cheaper to launch the next version.

In "The Mechanical Bird" I followed the damage done to man along its road: from pride to avarice, to greed, to envy, to lust, to accidie, and, at the end, to anger. Here I go back, as my father would have gone. I start with anger, because it is closest to the surface and is repaired with the smallest gesture: a pause. Then I descend, chapter by chapter, to pride, the root of them all. To each sin answers a virtue: gentleness, zeal, fidelity, joy at another's good, temperance, generosity, humility.

To you who are young, I say from the start: virtue is not a conduct grade. It is a habit, like a muscle. You don't have it or lack it; you exercise it and it grows, you neglect it and it weakens. No application can exercise it in your place, just as nobody can run in your place.

And to you who are parents, this order gives a practical piece of advice too. Don't start at the root. Don't ask a ten-year-old for humility; ask for a pause before he slams the door. It is the smallest gesture, it shows at once, and it opens the road to the others. No mechanic starts with the engine block.

This reverse order is also a criticism of the economy we live in. The economy sees what is produced and what is sold. Repair is not seen: it has no launch, no advertisement, it appears in no growth statistic. All seven gestures in this article are, in this sense, invisible. And that is precisely why they hold things up.

In brief, where we come from. In "Ten Just Men" we read four old parables. A city is held up by a few just people, and the web by a few people who still write on their own. Sodom's sin was comfort, not lust. Ninive turned around from the bottom up, and its prophet had to learn mercy. Looking back petrifies you; dor — longing, the Romanian word that cannot quite be translated — takes you home. And a second chance is not an amnesty, but one extra year, with the spade in your hands. I ended with a question: what exactly in man needs repairing? The answer has seven parts.

The remnant: how a society that has forgotten its values carries on, and what I can leave to you

Nineveh, not Gomorrah — Part III of III

In brief. Civilisations do not carry on through masses, nor through heroes, but through remnants: a few people who do not lie, a few places that preserve, a few tools kept in their place, and children who begin again. A second chance does not erase the fall. It makes it visible, like the gold with which the Japanese mend a broken bowl. On the web, the remnant is the archives, open source, and the people who still write. At home, it is the people at the table.

The Broken Bowl

In Japan there is an art with a short name: kintsugi, "joining with gold". When a ceramic bowl breaks, the craftsman does not throw it away or mend it in secret. He gathers the shards, joins them with a lacquer mixed with gold dust, and leaves the seam in plain sight. The repaired bowl carries upon it a glittering map of the break. It is not less beautiful than before. To many, it is more beautiful.

One story says the art was born in the fifteenth century, when a shogun sent his favourite bowl away to be mended and received it back held together with ugly metal staples. Japanese craftsmen then looked for a better way. Whether legend or not, the idea has remained: the trace of the repair is not a shame. It is the object's history.

The economy we live in does exactly the opposite. It discards. The phone that receives no more updates after a few years, though it works perfectly. An institution's website deleted at the change of management, archive and all. The post that vanished without trace when the platform closed. Everything that breaks is replaced, and everything replaced is forgotten.

Open source works like the kintsugi craftsman. When someone repairs a mistake, the repair stays in plain sight, in the version history: what was broken, who mended it, when, and why. And some companies building predictive models have begun to publish, equally in plain sight, reports on their models' errors and limits: where they go wrong, what they do not know, what has been corrected from one version to the next. It is little. But it is gold placed in the crack, not hidden adhesive. A model that shows its errors is more trustworthy than one that claims to have none.

To you, the young, this art is closer than you think. Second-hand clothes, the phone repaired instead of replaced, the bicycle rebuilt from parts: all carry the traces of a life before, and do not hide them. The same goes for your scars, the ones on your skin and the ones within. They need no longer be hidden.

And you, parents, have at home the most important kintsugi craftsman of all: the child who makes mistakes. He can be punished, and that is all. Or he can be helped to repair: to apologise, to glue back what he broke, to redo the homework. Punishment hides the crack. Repair fills it with gold.

Next time something breaks in the house — a mug, a flowerpot — glue it back together with your ten-year-old and tell him the story while you work. In Japan, when a beautiful bowl breaks, the craftsmen do not throw it away. They mend it with gold, so that you can see where it broke, because that is part of its story. Then ask him which of his own scars he likes: the knee grazed on the bicycle, the chipped tooth, a quarrel after which he became better friends with someone.

In short, where we come from. In "Ten Just Men" we saw that a city is held up by a few, and a second chance is one more year, with the hoe in hand. In "Seven Repairs" we walked back down the road of damage, from gentleness to humility, with one small gesture for each virtue. One question remained: a repaired person is not enough. How does a society that has forgotten its values carry on? History has an unexpected answer. Not through crowds, and not through heroes. Through remnants.

What my dead left me, and what the dead of the internet left us all

Three laws left "with a dying breath" (cu limbă de moarte) by my forebears, set beside the last testaments of those who invented the Web.

The law changes, it is postponed, it is circumvented. What was left with the death language is not a paragraph in the code — it is a philosophy of life: don't sell the land, keep the stick, don't exaggerate. It doesn't regulate. It guides.
That's why the parallel with the founders of the Internet is sharper than it seems. Barlow, Postel, Berners-Lee, Swartz didn't write laws. They left guidelines: the network belongs to everyone, be conservative in what you send and liberal in what you accept, this is for everyone, liberate knowledge. When the platforms treated them as archival documents — cited on anniversaries, forgotten the next day — they didn't violate a regulation. They betrayed a philosophy of life.
And the philosophy of life, unlike the law, has no police. It only has conscience, or it doesn't have it, without the intention of the durable but of the ephemeral.

At the Deathbed

Where I come from, the dying do not leave until they have said their piece. It is not a rule written anywhere, but I heard it throughout my childhood in various forms. "He left word with his dying breath" was a sentence that ended every discussion. There was nothing left to comment on. You could have disagreed with the dead man while he lived, you could have quarrelled with him over a garden boundary or a debt of two lei, but what he said at the end became law. Not state law — something weightier than that. Law of conscience.

I grew up with three such laws. I never asked for them, never signed for them, but I carry them. My grandfather said we must not sell the land. My grandmother, in her final years, spoke of almost nothing but the commemoration of the dead: who must be remembered, when, and with what; who makes the coliva (the boiled-wheat dish shared at Orthodox memorial rites); who takes the loaves to church. And an uncle of mine — a man the village remembered for the way he laughed — used to say that the joy of living lies in excess of nothing: not in food, not in drink, not in work, not in sorrow.

All three have long been dead. So long that I sometimes have to stop and count the years. And yet what they said settled in me more deeply than almost anything I have ever read.

In recent weeks I have been working on a structure about a different kind of testament: what the people who conceived the internet left behind. Licklider, Postel, Engelbart, Barlow, Swartz, Berners-Lee. Some are dead; others are still writing their memoirs. And reading their words, I had a strange sensation. They resembled my own dead. Not in style — in direction. Some of them said, in their own way, "do not sell the land". Others spoke of memory, and of how those who have gone are to be remembered. And almost all of them, the most clear-sighted, arrived in the end at a form of moderation, at the idea that a good tool is one that does not consume you.

This text is my attempt to lay the two deathbeds side by side. On the one hand, the bed in the old house, with the icon in the corner and the votive lamp that flickered whenever the door opened. On the other, the deathbed of a network I watched being born, earned my living from, and now see — in many respects — dying in the form I loved it.

I am not certain the parallel holds to the end. But I think it is worth attempting, because otherwise we risk treating the founders' testaments as archival documents, quoted at anniversaries and forgotten the next day. Where I come from, words spoken with a dying breath are not quoted. They are either carried out, or betrayed. There is no third way.

Or at least that is what I believed when I began to write.

Measure twice. On Ayn Rand, a ruler, and the state we grow up in.

A Question That Came from a Phone

It all started with an Instagram post. My son held his phone out across the table, with the gesture he usually uses to show me a meme I won't understand. This time it wasn't a meme. On a black background, in elegant white letters, it said: "I advocate the separation of state and economics, in the same way as we had the separation of state and church. — Ayn Rand."

Well? he asked me. Is it true? Would things be better that way?

I told him I didn't know. That wasn't evasion. I genuinely didn't know what to answer in thirty seconds, between two sips of tea, to a question that very clever people have been arguing about for a hundred years. But I realised something else, too. The quotation had reached him exactly the way most ideas reach people his age these days: without context, without the author explained, without a counter-argument—beautifully packaged and ready to share. An idea in a jewellery box. It looked precious, but you had no idea what was inside.

This text is the long answer I couldn't give him then. I'm writing it for him—but also for anyone between fourteen and twenty-something who has stumbled across a sentence like this one and wondered whether it is wise or dangerous.

I'll tell you from the start that I won't tell you whether Ayn Rand was right. I'll propose something else instead, which I find more useful: let us not treat her idea as a political programme you vote for or against, but as a measuring instrument. A ruler. And I'll show you that, if you use it honestly, this ruler measures in two directions, not just one.

But to get there, I have to begin with a workshop.

Man, the measure of the screen.

A Letter to Those Who Have Never Known a World Without Wi-Fi

"You shall choose your horse by placing a tray full of glowing embers in the middle of the herd." — Ion Creangă, The Story of Harap-Alb

Prologue: C:> and the Cursor That Waited

In 1997, I had a problem none of you will ever have: I had no access to a computer.

I'm not talking about a good computer, with a decent graphics card, or one that doesn't freeze when you open too many tabs. I'm talking about any computer. In Botoșani, in that year, a PC was an object you saw on television, in badly dubbed American films, or in a few offices where you had no business being. And yet I enrolled in a two-week course, "IBM PC Operator", at an institution that ran training courses. It had two working computers and one taken-apart broken one, a patient instructor, and a smell of heated dust that I still recognise today, anywhere I encounter it.

The course ended after two weeks. I stayed on for six months.

No one ever gave me written permission. The instructor chose not to notice that I kept coming back day after day, like a dog that had found its master. I would sit in front of the green screen, on which a single thing blinked:

C:\>

And after it, the cursor. That little line, white, appearing and disappearing every half second.

I want to describe what I felt then, because this entire essay begins there. That cursor waited. It didn't call me, didn't lure me, sent me no notifications, didn't know my name and didn't want to learn it. If I wrote nothing, nothing happened. The machine was completely, absolutely, reassuringly passive. Everything it did began with me. If I wanted to see what files were on the disk, I typed dir. If I got a letter wrong, the machine answered tersely, Bad command or file name—without getting annoyed, without judging me, but also without helping me. I had to learn its language, but once learned, it did exactly what I asked of it. No more, no less.

Three decades ago, the machine waited for the man. Today, the man is the one waiting for the machine: the vibration in your pocket, the blue tick, the feed loading, the next video starting on its own before you've had time to decide whether you want it.

I'm not writing this to tell you that "things were better in my day". They weren't. In my day, most people had access to none of what you have, and information was distributed at a trickle—through connections, through luck, through being tolerated for six months in a room where you had no business being. You carry in your pocket more computing power than the entire Botoșani Chamber of Commerce had in 1997—perhaps more than the whole county. That's an enormous gain, and I don't scorn it.

But something got inverted along the way. And I think it's worth us looking together at exactly what.

At the Market, Everyone Knows You

Author’s Note: This article uses a declared narrative fiction—a walk through the agro-food market in Botoșani—as a metaphor for browsing the internet without protection. The market scenes are literary constructs, not journalistic accounts. The technical guide within the article is real and functional. The article was written with assistance, in accordance with the transparency policy declared on this site.

It’s a Saturday morning.

I leave the house at half past eight.

The main market in Botoșani is a twenty-minute walk away—past the town hall, down the back of the cultural centre. I’ve known the route since I was a child. I could walk it blindfolded—the puddles after the rain, the occasional whiff of diesel from a passing cart, the vendors’ voices calling out prices to no one in particular and to everyone at once.

But this morning, something is different.

At the first street corner, a man I don’t know looks up from his phone and says, “Good morning. Off to buy vegetables?”

I stop. I stare at him. I don’t recognise him.

“How do you know that?” I ask.

“You go every Saturday,” he says simply. “You usually buy tomatoes, cucumbers if they’re good, sometimes cheese from the woman at the second corner. You buy less when it’s the end of the month. It’s the end of the month now.”

I swallow hard and walk on without answering. At the next corner, a woman with a notebook in her hand greets me before I can even look at her: “We know you have high blood pressure, Mr. Petru. The woman at the cheese stall said you take it unsalted.”

At the market entrance, a man in an orange vest stops me politely: “We have a special offer on spine supplements. We know you spend a lot of time at your desk.”

I turn around and go home. I haven’t bought a thing.

This is fiction. There are no people with notebooks at the street corners of Botoșani—only a complex web of interests, but within the bounds of common sense.

But there are browsers.

And your browser, in its default configuration, does exactly what I described above—with one key difference: it does it invisibly, without warning, and sells the collected information before you even get home.

The Last Days of Social Media. Or: What I Saw From My Window in Botoșani

My father died without ever having had a Facebook account.

Not on principle. Not out of some considered conviction about the nature of the digital age. He simply never got round to it, or never felt the need — he was the sort of man who sorted things out face to face, with a hand on a spanner, a direct look, and a word given that stood in place of a written contract. Mechanics of his generation had no use for an "online presence." They had something more durable: a neighbourhood reputation, built decade by decade, from work done properly and from word of mouth.

I think of him now when I unlock my phone and see what has become of the social networks in which I once, for a time, believed there might be something genuine.

There isn't. Not any longer.

The Feed as Visual Desert

In the morning, before my son wakes and before I make coffee, I pick up the phone out of an inertia I no longer manage well. It is more muscular reflex than intention. Thumb on glass, scroll, scroll. What I see resembles a conveyor belt transporting identical goods repackaged in slightly different wrapping: ten posts featuring the same stock portrait and the same breathless "click here for something free," three replies from filter-smoothed avatars with identically inflated lips directing me towards "exclusive content," an advertisement for a cashback crypto card, a football clip in which the players move awkwardly, like puppets whose strings have been poorly calibrated.

Between them, rarely, something human. A friend from school who has posted a photograph of his garden. A neighbour announcing he has changed his car. An acquaintance from the industry writing something about a project. These fragments of actual life are drowning in an ocean of content produced by algorithms, bots, and people who have adapted so thoroughly to the logic of machines that you can no longer tell where the person ends and the automation begins.

This is what we are now: content among contents.

Social networks were built on a romantic promise — that your friends who couldn't attend would see your wedding, that grandparents would know their granddaughter had taken her first steps into nursery, that people whom distance would otherwise have separated could remain in touch. It worked, for a while. I have my own memories from 2009 and 2010, when Facebook felt like a provincial market — crowded, noisy, but full of familiar faces, real conversations, a certain communal warmth that you could not have manufactured.

That's gone.

What remains is a visual desert with false oases.