Nobody records the morning you complied. The ledger has no column for it — and that missing column, rather than the obedience itself, is what the arrangement is built on.

Sharp’s contribution was not a moral argument. It was an inventory.

The standard picture has power sitting inside the institution, the way water sits inside a tank. It is a property the ruler owns, and it is spent on you. On that picture your obedience is a consequence of power, arriving after it, and your only question is whether the thing spent on you is being spent fairly.

Sharp turned the tank upside down. In the first volume of The Politics of Nonviolent Action — three volumes, Porter Sargent, 1973 — he argued that political power has no reservoir of its own. It is assembled every morning out of things that other people supply, and it lasts exactly as long as the supply does. The interesting question is therefore not whether power is legitimate. It is what, specifically, is being delivered, by whom, and how quickly it could be replaced.

He listed six sources. They are worth reading slowly, because five of the six arrive from below.

Read the list as an accounting document and the conclusion is uncomfortable for both sides. Nothing on it is mystical. Nothing on it is permanent. Every item is a flow rather than a stock, and a flow has a source that can be closed. The second volume of the work is nothing but a catalogue of ways it has been closed — 198 methods, sorted, described, and footnoted to the cases they were drawn from.

This is what separates Sharp from exhortation. He is not asking whether you ought to obey. He is asking you to find the line item.

Somebody had to write the rule that turns merely being here into having agreed. Locke wrote it, and set the bar on the floor.

The doctrine has a name and a date. Locke needed an account of how a government could bind people who had never signed anything, and he reached for a distinction between express consent and tacit consent. Express consent is straightforward: you join, you are a member. The difficulty, as he put it, is “what ought to be looked upon as a tacit consent, and how far it binds.”

Look at what the sentence actually does. It converts presence into agreement, without a moment at which anyone is asked. It requires no act on your part — it requires the absence of one. And once the conversion is made, the obligation it produces is not partial or provisional: you are bound as far forth … as any one under it, exactly as bound as a person who swore an oath.

Fifty-nine years later, Hume took it apart in a single paragraph, and did it by asking what the alternative to staying actually costs.

That is the flaw, stated once and never adequately answered: a consent read off from non-departure is only meaningful if departure is available. Where the exit is expensive, dangerous or imaginary, staying carries no information at all. The inference is not weak. It is empty.

And it is still in daily use. By continuing to use this site, you agree to our terms. That is §119 with a server behind it — the same conversion of presence into agreement, performed by a banner, over a document the drafter knows will not be read. Locke was at least trying to solve a real problem in political obligation. The modern version is a contract of adhesion that has borrowed his sentence and dropped his subject matter.

Whatever the merits of the doctrine, notice its function. It supplies the institution with a standing answer to the only dangerous question — who said you could? — and the answer is: you did, by being here, and you do it again every day you are still here.

There are two columns. Almost everything you might do lands in the same one.

Think of the last arrangement you disliked and did not change — a policy at work, a charge you did not agree to, a rule you thought was wrong. Pick what you actually did. The book will post it.

The list is not a trick, and the options are not strawmen. Private complaint, resentment, careful argument and a vote at the next opportunity are all real responses, and several of them are how a decent person handles a disagreement. The point is narrower and harder: from the institution’s side of the desk, none of them changes the delivery. The shift arrived. The payment cleared. The form came back signed.

This is why the framing matters so much more than the sentiment. An institution that measures approval by surveys can be told it is disliked and carry on unchanged, because the number it actually runs on is the one in the left column. Sharp’s six sources are all left-column quantities. So is every one of the 198 methods, read backwards.

If obedience were obviously a choice, it would have to be won every morning. So a great deal of work goes into making it not look like one.

None of these requires a conspiracy, and most are not even deliberate. They are the ordinary shapes an institution settles into when it needs a great many people to keep doing something without being persuaded each time. Tap any card for what it does and how to catch it.

The eight are not independent. They reinforce one another in a fairly reliable order: naturalisation removes the question, procedure disposes of it if it is raised anyway, asymmetry prices the refusal, diffusion ensures nobody has to answer for it, and futility closes the file. By the time a person says they have no choice, they are usually describing the accumulated output of all five rather than any single obstacle.

The argument is testable, and it has been tested. Filter by what was actually taken off the table.

Six occasions on which a specific supply stopped. Each card names what was withheld, what it broke, and what it cost the people who withheld it — because the last of those is the part that usually goes missing.

Look at what the cases have in common, and it is not courage. It is specificity. In every one, the thing withheld was something the institution actually needed and could not quickly buy elsewhere: a shipyard’s labour, a bus system’s fare box, a jury’s verdict, the discretionary judgement that makes a rulebook workable. Nobody withheld their disapproval, because disapproval is not a source of power on anyone’s list.

The work-to-rule case is the most instructive of the six precisely because it is the least dramatic. It establishes something the other five only imply: the system does not run on the rules. It runs on the unpaid judgement people add to the rules in order to make them function — and that judgement is a gift, given daily, which has never appeared in any budget and can be taken back without breaking a single one.

Withholding is not free, and a page that pretends otherwise is asking you to pay for its rhetoric. The cost is not fixed. It is a function of two numbers.

Refusal is priced by how easily you can be replaced and by how many people are refusing with you. Both matter, and the second one matters more than most people expect, because the institution’s capacity to punish is itself one of Sharp’s six sources — finite, staffed, and spread across however many refusals it has to process at once.

The shape of that curve is not invented for this page. It is the central empirical finding in the field. Erica Chenoweth and Maria Stephan built a dataset of 323 violent and nonviolent campaigns between 1900 and 2006, all of them making maximalist demands — unseating a government, expelling an occupation, winning independence — and published the comparison as Why Civil Resistance Works (Columbia University Press, 2011). Nonviolent campaigns succeeded in about 53 per cent of cases; violent ones in about 26. Chenoweth’s later and more famous observation was that nearly every campaign in the set that mobilised at least 3.5 per cent of the population at its peak achieved its aim.

That figure has been repeated so often, and so badly, that it now does more harm than good. It is not a threshold, a law, or a target. Chenoweth has been explicit that it is a rule of thumb, descriptive of a particular dataset rather than prescriptive of anything, and that it says nothing about organisation, momentum or strategy — the factors that actually decide whether the participation is reachable in the first place. What survives the caveats is the modest and useful claim the slider is built on: individual exposure falls as participation rises, and it falls faster than linearly.

An institution that has never had to explain itself has not passed a test. It has avoided one.

Here is the test. Name something you comply with — an employer, a landlord, a platform, an agency, a professional body — and answer five questions about it honestly. The card at the bottom is not a verdict on whether the institution is good. It is an estimate of how much of its position depends on your side of the ledger, which is a different and more actionable thing.

The reason the questions are about supply rather than about fairness is that fairness has no lever attached to it. You can be certain an arrangement is unjust and have no idea what to do on Monday. The five questions are designed to end somewhere specific: a named thing you provide, a rough estimate of how replaceable it is, and the number of people who would have to be alongside you for withholding it to be survivable.

The consent theory of power is a real theory with real critics, and some of them are right. Four places it breaks.

It is voluntarist about things that are not voluntary. Brian Martin’s critique in the Journal of Peace Research (vol. 26, no. 2, 1989, pp. 213–22) is the sharpest version. Sharp’s ruler-and-subject frame, Martin argues, has nothing to say about power that is structural rather than commanded. Capitalism divides the people who would have to act together by status, skill, wages, gender and ethnicity, and pays them while it constrains them. In a bureaucracy the dichotomy dissolves entirely, because nearly everyone has both superiors and subordinates. Patriarchy is maintained by upbringing, discrimination and violence across many institutions at once, and cannot be switched off by withdrawal of consent from any of them. Sharp, Martin concludes, supplies no conceptual tools for any of this. He is not wrong about that.

“Consent” is the wrong word for compliance under a gun. A theory that describes the behaviour of someone who obeys because they will otherwise be shot as a supply they are choosing to provide has stretched a word until it no longer carries information. Sanctions are the sixth source for a reason. Where they are severe, immediate and certain, the other five stop being the live variables, and telling a person in that position that their obedience is a choice is both technically defensible and practically obscene. The theory earns its keep in the enormous space between that case and a routine morning — not by flattening the difference.

The empirical record has got worse, not better. The 53-per-cent figure comes from 1900–2006, and the world since has not cooperated. Success rates for nonviolent campaigns peaked around 65 per cent in the 1990s and have fallen to below 34 per cent since 2010. Average peak participation has dropped to roughly 1.3 per cent — well short of the threshold the popular version of the rule advertises. Chenoweth attributes the decline to movements getting smaller and to regimes getting better: blaming foreign hands, censoring and surveilling, spinning, excluding independent journalists, training security forces specifically against defection, and sharing suppression playbooks with each other across borders. The counter-tactics are now the mature technology.

The best-loved case may not be a case. In February and March 1943, after some 1,800 Jewish men married to non-Jewish women were held at Rosenstraße 2–4 in Berlin, their wives and relatives — by some accounts up to 6,000 people, overwhelmingly women — stood in the street until the men were released. It is the most cited example of unarmed refusal stopping a machine at its worst. Nathan Stoltzfus’s Resistance of the Heart (1996) makes the case that the protest was decisive, citing Goebbels’s own note that deportations were discontinued because “large throngs” had gathered. Wolf Gruner has argued since the late 1990s that this is wrong — that directives at the time barred the Gestapo from deporting intermarried German Jews, that release was always the plan, and that the women therefore changed nothing. Historians including Wolfgang Benz, Claudia Schoppmann and Beate Meyer have sided with him; a majority still find the protest the most plausible explanation. The honest position is that it is contested, and that a page which cites Rosenstraße without saying so is selling something.

What survives all four is narrower than the slogan and more useful than it. Withholding obedience is a lever, not a solvent. It works where an institution needs a particular supply from identifiable people who cannot be replaced fast enough — and where enough of those people move at once that the sanction budget runs out before the refusal does. Outside those conditions it is a gesture, and gestures have costs too.

Everything above is checkable. Here is where to check it.

• Gene Sharp — The Politics of Nonviolent Action (Porter Sargent, Boston, 1973). Volume 1, Power and Struggle, carries the six sources and the consent theory; volume 2 is the catalogue of 198 methods; volume 3 is what happens after you use them.
• John Locke — Second Treatise of Government (1689), §§119–122. Tacit consent, in the author’s own words, including the lodging and the highway.
• David Hume — “Of the Original Contract” (1748) for the peasant and the ship, and “Of the First Principles of Government” (1741) for the sentence at the foot of this page.
• Erica Chenoweth and Maria J. Stephan — Why Civil Resistance Works (Columbia University Press, 2011). The 323-campaign dataset, and the source of every number in the price column.
• Brian Martin — “Gene Sharp’s Theory of Power,” Journal of Peace Research 26:2 (1989). The structural critique, made by someone broadly sympathetic to the project.
• Nathan Stoltzfus — Resistance of the Heart (1996), against Wolf Gruner — Widerstand in der Rosenstraße (Fischer, 2005). Read both, in that order, before citing the case.
• The Federation, File №05 — The Colossus. La Boétie asked why the usurper is obeyed at all, three centuries before anyone had a dataset. This file is the accounting department of that question.