Prison guards organise a relay race. Guards against prisoners.
The point is obvious to everyone: the guards want to demonstrate physical superiority, and the prisoners are expected to lose.
So the prisoners lose. Enthusiastically.
"Miming exaggerated exertion while barely moving, running a few yards and collapsing, limping, jogging ever so slowly while the guards raced ahead at full speed."1
They did precisely what was wanted. And it ruined the event.
Greene's statement of the mechanism is the cleanest sentence in the chapter:
"Both by joining the race and by losing it, they had obliged the guards obediently; but their 'overobedience' had mocked the event to the point of ruining it… they could not rightly punish the prisoners, who had only done what they asked."1
Every instruction was followed. Race — they raced. Lose — they lost.
To punish them, a guard would have to articulate the actual expectation, and the actual expectation is unsayable: you were supposed to lose while appearing to try, so that our winning would mean something.
Saying that out loud concedes the whole thing — that the race was staged, that the guards needed the performance, that the superiority being demonstrated required the prisoners' cooperation to exist.
So the attack is protected by the rule it obeys. The only defence against it would be an admission that destroys the thing being defended.
Not the guards. Not the rules. The meaning of the event.
A relay race between guards and prisoners is supposed to produce a fact: we are stronger. That fact requires the losing side to be genuinely trying, because a victory over people not trying establishes nothing.
By complying visibly and uselessly, the prisoners removed the event's ability to generate the fact it existed to generate — without breaking a single rule, without refusing anything, and without giving anyone a grievance to act on.
Greene names what the alternative would have cost:
"Resistance would have engaged the prisoners in the cycle of violence, lowering them to the guards' level."1
Refusing to race would have been a reason. A reason is what a guard needs. Overobedience supplies none.
The chapter's marginal on the facing column is the same mechanism in one line:
"When the great lord passes, the wise peasant bows deeply and silently farts."2
The bow is real. It is deep — deeper than required, in fact, which is the joke. Nothing in the peasant's visible conduct is deficient, and the lord receives everything he came for.
What the peasant keeps is unshared and unpunishable, and its existence does not depend on the lord ever knowing.
That last part matters and separates the proverb from the relay race. The prisoners' mockery had to be perceived to work — the event's ruin was the point. The peasant's does not. It is entirely private, and would be identical if nobody ever knew.
Two different things are being called the same thing here: sabotage that must land, and dignity that needs no audience.
Worth naming plainly, because the chapter does not.
The relay race is from Milan Kundera's The Joke, described by Greene as "based on the author's experiences in a penal camp in Czechoslovakia."1
It is fiction. Greene's other cases in this chapter are historical claims; this one is a scene from a novel, presented in the same register and doing the heaviest analytical work in the chapter.
🚩 The provenance chain — a novelist's fictionalisation of his own camp experience, summarised by Greene, uncited — is the weakest in Law 22, and the insight is the strongest. Both things are true and the chapter marks neither.
This tactic has a requirement that Greene never names, and it decides whether the move is available at all.
It only works where the powerful party needs your participation to produce what they want.
The guards did not want to run. They wanted to beat somebody, and beating somebody requires a somebody. That dependency is the entire opening.
Test it against cases where the dependency is absent. A guard who wants a floor mopped does not need enthusiasm — mopping slowly and badly gets you punished, because the output is the point and the output is inspectable. A regime that wants a signature does not need sincerity.
So overobedience attacks ceremonies, not tasks. It works on anything whose value is symbolic — a demonstration, a display of loyalty, a ritual of deference, a performance of consensus — because symbolic value is the only kind that can be destroyed by being supplied incorrectly while being technically supplied.
That is why Brecht works. A congressional hearing is a ceremony, and its product is a demonstration of authority that requires a witness who is visibly submitting. Give it too much and it produces nothing.
You are being asked to participate in something whose real purpose is to demonstrate somebody's authority over you.
Check the dependency first, because it decides everything. Does this need you specifically to behave a certain way, or does it just need an output?
If it needs an output — a report, a number, a piece of work — this tactic does not apply, and attempting it gets you fired for poor performance. Do the work.
But if what is wanted is a performance — visible enthusiasm, a public agreement, a display of being managed — then the dependency is real and you have an option.
Comply completely and specifically. Not slowly, not grudgingly, not with visible reluctance — reluctance is a grievance and grievances are actionable. Ask more clarifying questions than necessary. Follow the stated process exactly. Produce precisely what was asked for, in the form requested, on time.
And keep the temperature warm, because that is the load-bearing part and the easiest to lose. The prisoners were not sullen; they were enthusiastic. Overobedience delivered with an edge is just insubordination with extra steps, and it will be read and punished as such.
Then know which of the two versions you are running, because they have different exit conditions. The relay race version needs an audience — if nobody notices the event was hollow, nothing has happened. The peasant's version needs nobody, and is the one available when the cost of being noticed is too high to bear.
The strongest material is the punishment problem, which is a genuine structural insight: an attack conducted entirely through compliance cannot be charged without conceding the expectation it violated.
Tension: the chapter's best analytical case is fiction, presented as evidence. No other case in Law 22 is a novel and the register is identical.
Tension: two incompatible things are gathered under one heading. The relay race requires perception to succeed; the Ethiopian proverb requires none. Sabotage and private dignity are different objectives with different risk profiles.
Tension: the precondition is unstated and it is the whole of the tactic's scope. Greene presents overobedience as generally available; it works only where the powerful party's aim is symbolic and depends on your manner.
🚩 SINGLE SOURCE — Kundera is named as an author and the novel as The Joke, with no edition, translator or page; the Ethiopian proverb is given with no collection or source. [POPULAR SOURCE] ⚠ Fiction used as evidence, unmarked.
Open questions. How many times can a group run this before the ceremony is simply abandoned — the tactic ruins one event, and an authority that stops holding relay races has removed the opening. And: is there a version that survives being named? Once overobedience is a recognised category, does the deniability that makes it work survive?
Against the law it sits inside, this is the one case where surrender is not a holding action. Everywhere else in Law 22 yielding buys time — Goujian's three years, Brecht's exit, Hotta's generation. Here the yielding is itself the blow, and there is no later strike to wait for. Greene files it as a variety of surrender; it is closer to an attack conducted in surrender's grammar.
Set beside Law 21 the two share a structure the corpus keeps rediscovering. Both work by giving the powerful party exactly what their vanity requests, in a quantity that spoils it. Law 21's mark gets to feel clever and is robbed; the guards get to win and are humiliated. Oversupply as attack appears twice in consecutive chapters and is never named as a principle.
And against the chapter's Reversal, the relay race is the counter-example that makes the Reversal's case. The Reversal argues martyrdom is "messy, inexact" and unpredictable in its returns. The prisoners achieved a complete, immediate, cost-free victory over the same authority — and they achieved it by declining every opportunity for a principled stand.
Psychology — The Puppy in the Other Room
That page works the Milgram replications and what they actually changed — obedience studied as a graded, situational response rather than a fixed trait, with compliance shifting sharply on the details of how authority is presented.
The relay race is the same apparatus inspected from the subject's side.
What the pair produces: obedience is a variable the obedient party can manipulate, and the Milgram frame only ever treats it as an outcome. The experiments measure how much people comply. They do not ask what happens when a subject complies with deliberate, ruinous precision — because the paradigm assumes compliance is the thing authority wants, and the prisoners demonstrate that authority wants a specific quality of compliance it cannot request.
So there is a hidden term in every obedience situation: the unstated standard of manner. The guards' instruction was race, and their requirement was try. The gap between the instruction and the requirement is the whole exploitable space, and it exists in every hierarchy because no authority can fully specify the spirit in which it wants to be obeyed. Which suggests obedience research has been measuring the easy half — the part that can be instructed — and the interesting half is what subjects do with the part that cannot.
History — The Tatekai: Preservation as Resistance in the Ryukyus
That page describes cultural preservation conducted where the culture is illegal — forms taught mouth to ear, nothing written down, a secret federation carrying a kingdom's martial memory through occupation and prohibition.
Both pages are about resistance that refuses confrontation. The mechanisms are opposite in a useful way.
What the pair produces: unpunishable resistance comes in two forms — too visible to charge, and too invisible to find — and the choice between them is set by how long you need to last. The prisoners' method is public and immediate: it works in one afternoon, and it works because everyone can see it.
The Tatekai's method is the inverse — it survives four centuries precisely because nothing is written and nobody outside knows. And the relay-race tactic could not have done that job, because a ceremony ruined is a ceremony an authority stops holding: the opening closes the moment it is used.
So overobedience is a tactic for a bounded encounter and useless as a strategy for surviving an era. Brecht's hour, the prisoners' race, the peasant's bow — all single events. What outlasts an occupation is not clever compliance but a transmission channel the occupier never learns about, and Greene's chapter, which is entirely built from single encounters, has no category for it.
Sharpest implication. The prisoners could not be punished because punishing them would have required saying the quiet part: you were supposed to lose while appearing to try.
An attack made entirely of compliance forces the powerful to either accept it or articulate an expectation that destroys their own claim — and most will accept it, which is why it works.
Generative questions.