During an important battle in the War of the Three Kingdoms, in the third century A.D., advisers to the commander Tsʻao Tsʻao discovered documents showing that certain of his generals had conspired with the enemy, and urged him to arrest and execute them. Instead he ordered the documents burned and the matter forgotten. At this critical moment in the battle, to get upset or demand justice would have reverberated against him: An angry action would have called attention to the generals' disloyalty, which would have harmed the troops' morale. Justice could wait—he would deal with the generals in time.1
⚠ The chapter's phrase is "the matter forgotten" and its own next sentence contradicts it: justice could wait—he would deal with the generals in time.
He does not pardon anyone. ⚠ He destroys the evidence and defers the reckoning, which are two separate decisions, and the corpus should keep them apart because only one of them is about anger.
Deferring is the timing decision. ⚠ Burning is a decision about what other people can know — and it is the more interesting one.
⚠ This is the finding and it inverts the ordinary reading of evidence.
Possessing proof of a conspiracy is normally an asset. ⚠ Mid-battle it is not, because acting on it requires making it public, and the public fact is not these men are traitors — it is this army contains traitors and the commander cannot tell which.
And that fact damages Tsʻao Tsʻao more than the conspiracy does. ⚠ An angry action would have called attention to the generals' disloyalty, which would have harmed the troops' morale.
⚠ So the documents were dangerous to hold, not because they might be lost, but because holding them created a standing temptation to use them — and the only moment when using them was cheap had not arrived.
Burning removes the temptation along with the proof. ⚠ It is a commitment device: he cannot act on evidence he no longer has.
⚠ He gave up the ability to prove it later.
He would deal with the generals in time — with what? ⚠ The documents are ash. Whatever he does subsequently will look like a commander removing men he distrusts, and there will be nothing to show anyone.
Which the corpus should record as a genuine trade rather than a clever escape. ⚠ He exchanged provable justice later for no morale damage now, and the exchange only makes sense if the later action does not need to be justified.
⚠ Which it did not, because he was the commander. A person who must account for their decisions cannot burn the documents, and the chapter's advice is silently addressed to people who do not have to.
⚠ To get upset or demand justice would have reverberated against him.
Greene's verb is doing real work and is worth unpacking, because it names a specific mechanism rather than a general risk.
An accusation is a statement about the accuser's situation. ⚠ A commander who announces that his generals are conspiring has announced that his generals conspire — and the troops do not hear these two are traitors, they hear the command is compromised.
The information cannot be delivered selectively. ⚠ You cannot tell an army that a specific conspiracy has been discovered without telling it that conspiracies happen here.
⚠ Which is the same structure the corpus recorded at Roosevelt and Fala: an example does not illustrate an argument, it replaces it. Here a prosecution does not demonstrate control, it demonstrates that control was in question.
⚠ 🚩 SECONDARY WITHOUT PRIMARY · 🚩 SINGLE SOURCE — no battle named, no date beyond "the third century A.D.", no source, no historian. ⚠ The Three Kingdoms period is extensively documented and extensively fictionalised, and the chapter does not say which it is drawing on.
🚩 ⚠ The anecdote closely resembles a famous episode set after Guandu (200 CE), in which captured correspondence between Tsʻao Tsʻao's own people and the enemy is burned unread. That episode circulates in both the historical record and in Romance of the Three Kingdoms, a fourteenth-century novel, ⚠ and the two are routinely conflated. [CONTESTED] — the corpus should not carry this as history without a source.
🚩 ⚠ Greene's version has a detail the famous episode does not: advisers… urged him to arrest and execute them. The pressure-from-advisers element sharpens the lesson considerably, and it is exactly the sort of thing a retelling adds.
🚩 "Tsʻao Tsʻao kept his head and made the right decision" — ⚠ no outcome is reported. The battle's result, the generals' subsequent fate, and whether he ever did deal with them are all absent.
You have discovered that someone on your side has been disloyal, and you can prove it.
Separate the two decisions. ⚠ When to act, and what to let people know. They feel like one decision and they are not, and conflating them is how a private problem becomes a public one.
Ask what the accusation announces about you. ⚠ Not whether it is true — whether the fact of your making it damages your position more than the conduct does. A commander announcing a conspiracy has announced that his command is conspirable. This is the calculation Tsʻao Tsʻao made and it is available in five seconds.
Consider whether holding the evidence is safe. ⚠ This is the case's genuine contribution and it is counterintuitive: proof you cannot currently use is a standing temptation, not an asset. If you know you will be tempted at a bad moment, removing the option is a stronger move than resolving not to take it — and the corpus recorded the mechanical version of that at Shigemune's tea mill.
⚠ But price what you are giving up. Burning the documents means the later action will be unjustifiable to anyone who asks. Only do it if nobody will be asking — and if somebody will, the technique is not available to you and the chapter does not say so.
Strongest evidence. A clean strategic logic — the public fact of an accusation damages the accuser more than the concealed conduct does — with a stated mechanism (morale) and an explicit deferral rather than a pardon.
Tension — the chapter says "the matter forgotten" and its own next clause says the opposite.
Tension — burning the evidence costs the ability to justify the later action, which the chapter does not mention.
Tension — the technique requires an actor who answers to nobody.
🚩 [POPULAR SOURCE] · 🚩 SINGLE SOURCE · 🚩 SECONDARY WITHOUT PRIMARY — no battle, date, source or historian. 🚩 [CONTESTED] — the episode resembles a famous post-Guandu account that circulates in both history and a fourteenth-century novel, and Greene does not distinguish them. ⚠ No outcome is reported.
Open questions. Did he ever deal with them? ⚠ The chapter asserts the deferral and never returns to it, which leaves the technique's second half entirely unevidenced. A deferred reckoning that never happens is not patience — it is a pardon by attrition, and the corpus recorded the same open structure at Law 35's end time: a staged strategy with no marker for the second stage frequently has no second stage.
⚠ Greene explicitly frames this against the Transgression — "Compare this to Napoleon's response to Talleyrand" — which makes it one of the few places in the build where the book runs its own controlled comparison. Same situation: two senior subordinates conspiring, discovered. ⚠ Napoleon made it public and lost standing; Tsʻao Tsʻao made it disappear and kept his.
And the comparison is fair, which is worth saying because most of the book's are not.
⚠ Against Law 36 the convergence is exact and unremarked. That law's finding was that the real content is do not respond visibly, not ignore, and that Greene's own escape clause is conspire in secret. ⚠ Tsʻao Tsʻao is that clause executed — he does not ignore the conspiracy, he declines to respond where it can be seen. Three laws apart, and the corpus can now say Law 36's repaired version and Law 39's Keys are the same instruction.
Organisational behaviour — the investigation that damages the institution. Authority and Institutional Override examines how invoking a formal process changes what an event is, because the process creates a record, a category and a set of actors with their own momentum.
An angry action would have called attention to the generals' disloyalty.
The insight neither produces alone: the chapter's reasoning is about morale — the troops would be dismayed. ⚠ The institutional frame identifies the more durable cost: a prosecution creates a category that did not previously exist in the organisation. Before the arrest there are two disloyal generals; after it there is a thing called disloyalty among the generals, with a procedure, a precedent and a name. ⚠ And categories persist after the individuals are gone — which is why Tsʻao Tsʻao's burn is stronger than a quiet dismissal would have been: a quiet dismissal still leaves a question, and ash leaves nothing to attach a category to. The general rule: before opening a case, ask what the case will still be called after it closes.
Behavioural economics — removing the option rather than resisting it. Loss Aversion establishes that decisions under pressure are evaluated against a shifted reference point, and that the shift is not corrected by knowing about it.
Burning the documents is a commitment device.
What the pairing produces: the chapter reads the burn as composure — he kept his head. ⚠ The commitment frame says something less flattering and more useful: he may have burned them precisely because he did not trust himself to keep his head. Proof of betrayal, retained, is a standing invitation to act at the worst moment — and the moments when the temptation peaks are exactly the moments when the reference point is most distorted. ⚠ So the strongest reading is that Tsʻao Tsʻao did not resist the temptation, he abolished it, which is the same structure as setting the trigger while cold and as Shigemune's mill. Three instances now in one law: the reliable interventions are all things you install in advance, and none of them is self-control.
Sharpest implication. ⚠ Proof of a conspiracy was a liability, not an asset — because acting on it requires making it public, and the public fact is not these men are traitors but this army contains traitors and the commander cannot tell which. And burning is a commitment device, not forgiveness: the chapter's own next clause says justice could wait. ⚠ The strongest reading is that he did not resist the temptation, he abolished it. But the trade is real and unpriced: he gave up the ability to justify the later action to anyone who asks — which is only affordable if nobody will be asking.
Generative questions.