The colossus is not held up by its base. It is held up by everyone watching everyone else hold it up.

La Boétie’s whole argument fits in one move: he asks where the tyrant keeps his strength.

Every other writer on tyranny asks how to fight it. La Boétie asks a stranger question, and a much harder one to answer: why does anyone have to fight it at all? One man rules millions. He has one body, one lifetime, two hands. He cannot physically compel a million people, because he does not possess the force required — the force is theirs. It follows that the millions are not being overpowered. They are supplying, every morning, the power that holds them down.

He was around eighteen when he wrote that down. The Discours is the earliest text in this series by three centuries, and the most structurally radical thing in it, which is why it spent most of its life circulating in manuscript, in fear, and under other men’s names.

That variant is the whole book in one word. In the printed tradition the tyrant harms you as long as you have the power to endure it, which makes servitude a matter of capacity. In the manuscript Payen recovered, he harms you as long as you have the will to endure it, which makes servitude a decision you are re-making right now. The title is not a metaphor. La Boétie means the servitude is voluntary, and he means it as an accusation.

“Stop holding it up” is a mood until somebody builds the means of letting go.

The weakness of the Discours as an instruction is that it asks for a negative act and supplies no equipment. Four places where the equipment is actually being built.

Further reading — from the source and after it

• Étienne de La Boétie — Discours de la servitude volontaire, quoted here from Payen’s 1853 edition, the first printing of the de Mesmes manuscript and still the base text.
• David Hume — Of the First Principles of Government (1741). The same observation about opinion and force, turned into an argument for leaving the foundation alone.
• Timur Kuran — Private Truths, Public Lies (1995), and Now Out of Never (World Politics, 1991) on why 1989 was unforecastable.
• Mark Granovetter — Threshold Models of Collective Behavior (American Journal of Sociology, 1978). Why the same crowd with the same grievance either moves or does not.
• The Federation, File №01 — The Usurper. Proudhon names the hand laid on you; La Boétie asks why you are holding it.

It reads as an attack on the tyrant. It is really a claim about the people carrying him.

Strip out the sixteenth-century fury and four load-bearing claims remain.

Look at what the colossus is actually standing on.

Not bronze, and not an army. Take the base apart and every piece of it turns out to be a person doing what the person beside them appears to be doing.

Each item is the same trick performed at a different scale. Nothing in the list contains any force of its own; each one borrows its authority from the apparent agreement of everyone else in the frame. Which means the base is not a structure at all. It is a reading — a few million people looking at each other, concluding that this is evidently how things are, and then becoming the evidence.

La Boétie saw this in 1548 and could describe it exactly. What he could not do was say why the reading is so stable, or what it would take to change it. That took another four hundred years, and it is the subject of the 2026 section below.

Proudhon, in File №01, names the hand laid on you to govern you and calls it usurpation. Kropotkin, in File №02, names the commons that hand encloses. La Boétie comes three hundred years before both of them and asks the question neither has to: not who is governing you, and not what is being taken, but why you are still carrying him.

It is worth being honest about how ugly this move is. Told without care, “the people enslave themselves” is victim-blaming with a Renaissance accent, and it has been used that way. The text earns it only because of what comes next: he immediately shows that the weight is not carried by the people at all. It is carried by a specific, nameable, and much smaller group.

Not the crowd, and not the army. Four or five men, and then a chain.

This is the most concrete passage in the Discours, and the most useful. La Boétie dismisses the guards outright — Roman emperors, he notes, were more often killed by their own archers than saved by them — and then counts the real structure.

Two things in this are worth more than the arithmetic. The first is the mechanism of the sixth thousand: they are made complicit deliberately, given enough to lose and enough to answer for that leaving becomes more dangerous than staying. Servitude is secured by implicating people, not by frightening them.

The second is what this does to the instruction. If the load sits on four or five, then six hundred, then six thousand, the question “who should stop holding it up?” has a specific answer, and it is not the crowd. The people best placed to drop the weight are the ones already being paid to carry it — and they are precisely the ones with the most to lose by letting go. That is not a detail. It is the reason the Discours has never worked as simply as it reads.

La Boétie could describe the base but not explain it. Four bodies of modern work can.

The Discours leaves a hole exactly where it should be strongest. If servitude is voluntary and withdrawal is free, why does anybody comply for more than an afternoon? La Boétie’s answer is habit, which is true and nowhere near sufficient. The research that closes the gap is recent, empirical, and almost eerily well matched to the text.

So why doesn’t it fall? Because the cascade is contingent, and nobody can see the variable it turns on.

Mark Granovetter’s Threshold Models of Collective Behavior (1978) is the coldest water available on the Discours. Each person has a threshold: the number of others who must go first before they will go. What happens next depends on the distribution of those thresholds across the crowd — not on how angry anyone is, not on how unjust the situation is, not on how persuasive the argument was.

His demonstration is one sentence long and it ruins your afternoon. Take a hundred people with thresholds of 0, 1, 2, and so on up to 99: the first needs no one, the second needs one, and the chain runs to the end. All hundred act. Now move a single person’s threshold from 1 to 2. The chain breaks at the second link. One person acts, nobody follows, and the incident is written up as an isolated crank. Identical grievance, identical crowd, opposite outcome, invisible cause.

This is the hard news. “Stop holding it up” is not a plan, because whether your withdrawal spreads or strands you depends on a distribution that nobody has ever measured, that nobody chose, and that no one in the crowd — including you — can see from the inside. La Boétie promises that refusal is free. Granovetter shows that the price is set by other people’s thresholds, and the bill arrives afterwards.

Which is why the serious modern counter-move is not censorship. It is flooding.

Gary King, Jennifer Pan and Margaret Roberts put Chinese censorship on an empirical footing in 2013, and the result reads like a commentary on this text. The censor permits criticism of the state and its leaders — often sharp criticism — and removes posts with collective-action potential. The apparatus is not defending the leader’s reputation. It is defending the base. Set that finding beside the Discours and the effect is unsettling: the censor and La Boétie agree completely about how power works, and disagree only about which side to be on.

Roberts’ Censored (2018) sorts the toolkit into fear, friction and flooding, and flooding is the cheapest of the three by a wide margin. In 2017 King, Pan and Roberts estimated the scale from roughly 44,000 leaked posts in a single county: about 448 million fabricated social-media posts a year, overwhelmingly cheerleading and distraction rather than argument. Nobody is being persuaded of anything. That is the point. Argument would concede that there is something to argue about; flooding simply ensures that no one can hear the count.

Read through Kuran, this is precision engineering. A regime that cannot stop people from privately defecting can still survive indefinitely by keeping the rate of defection unknowable. You do not have to hold the colossus up. You only have to keep everyone from noticing who has let go.

The colossus comes down. The Discours never says what lands.

This file does not resolve the following objections, because the text does not resolve them and pretending otherwise would be the easiest way to make it useless. Hume’s is first because it is the one that does not go away.

1. Withdrawal has no direction

   It is a negation with no vector. “Stop holding it up” specifies precisely what to stop and nothing whatsoever about what to start, and a base that dissolves does not get to choose its successor. The falling colossus does not leave a vacuum; it leaves rubble, and rubble is cleared by whoever arrived with equipment. Hume’s objection is not that La Boétie is wrong about the mechanism. It is that being right about the mechanism tells you nothing about where the next opinion settles — and the people best organised to settle it are rarely the people who withdrew.

2. A cascade is a mechanism, not a politics

   Every dynamic in the 2026 section above runs in any direction at all. Preference falsification collapses just as fast toward a new orthodoxy as away from an old one; Chwe’s rational ritual works exactly as well for a junta consolidating power as for a crowd dispersing it. The cascade runs toward whatever is most salient, and manufacturing salience is the one thing a small organised faction is genuinely good at. 1989 and 1979 are the same machine.

3. The costs are not shared, and the text pretends they are

   La Boétie’s most seductive claim is that liberty is free: you need only stop. That is true only for the person whose withdrawal is invisible. For the first to go — the named organiser, the clerk who refuses in writing, the one whose face is on the footage — the cost is total and the benefit is collective. Granovetter’s low-threshold actors are not braver than everyone else; frequently they are just cheaper to be. Any honest use of this text has to say out loud who is being asked to go first, and what happens to them if the chain breaks at the second link.

4. Habit is not only the disease

   The same custom that keeps a tyrant upright keeps a contract enforceable, a court approximately honest, and a road maintained. There is no surgical withdrawal that dissolves the bad half of la coustume and leaves the rest standing, and the Discours never confronts this because it has no after. The text ends at the moment of collapse, which is the last moment at which its argument is still simple.

So the file stands, with its limit marked. The Discours is the best account ever written of why power is weaker than it looks. It is not an account of what to build, and it should never be asked to be one. For that, this series has File №04 on capacities people already have, File №02 on what is actually being enclosed, and the federative argument running underneath all of them: coordination without a centre is a design problem, not a mood.

A magistrate who wrote the most dangerous text of the century before he was twenty, then spent his short life serving the crown.

Étienne de La Boétie was a career judge. He took his seat on the Bordeaux parlement at twenty-two, young enough that the appointment required a dispensation and an examination before the assembled chambers, which found him idoine et suffisant and swore him in on 17 May 1553. He is remembered today for two things he never published: an essay arguing that all political power is a loan from the governed, and a friendship.

The friendship was with Montaigne, who met him around 1558 and never really recovered from his death five years later. De l’amitié, the essay Montaigne built around him, was meant to carry the Discours as its centrepiece. It does not. By the time the Essais went to press, Protestant polemicists had printed the text as anti-royalist propaganda in the wake of the St Bartholomew’s Day massacre, and Montaigne — furious at what had been done with his dead friend’s schoolwork — pulled it and dropped twenty-nine sonnets into the gap instead.

The gap between the two halves of the man is the honest part of the story. The author of the Discours also wrote, in 1562, a sober policy memorandum on the January edict proposing a moderate settlement of the religious wars under royal authority. He was not a revolutionary. He was a young lawyer who followed an argument to its end on paper and then went to work. The text spent three centuries circulating in manuscript, in fragments, and under other men’s purposes, until Payen printed the de Mesmes manuscript in 1853 and gave it back its own words.