Greene's image for this law is a snake you stepped on and did not finish: "A viper crushed beneath your foot but left alive, will rear up and bite you with a double dose of venom. An enemy that is left around is like a half-dead viper that you nurse back to health. Time makes the venom grow stronger."1
Two claims are packed in there, and only one of them is obvious.
The obvious one: a surviving enemy remains an enemy. Fine.
The other one is the load-bearing claim and it is much stranger. The venom grows. Not persists — grows. The half-crushed enemy is more dangerous than the uncrushed one, because you have now added humiliation to whatever they already had against you. On this account, wounding is worse than never having touched them at all.
Greene states the law's actual content in one sentence, and it is not advice about aggression. It is a claim about a category of relationship: "The law governing fatal antagonisms reads: Reconciliation is out of the question. Only one side can win, and it must win totally."2
Read that as a diagnosis rather than a strategy and the chapter changes shape. He is not saying you should refuse to reconcile. He is saying that in certain relationships reconciliation is not on the menu, and that believing otherwise is what kills people.
The mechanism he gives is emotional rather than strategic: "This is the fate that faces all of us when we sympathize with our enemies, when pity, or the hope of reconciliation, makes us pull back… We only strengthen their fear and hatred of us. We have beaten them, and they are humiliated; yet we nurture these resentful vipers."3
And note the clause that does the sorting: "This is all the truer with a former friend who has become an enemy."4 The law is aimed most precisely at people you used to like — which is exactly the population where the pity that disables you is strongest.
The strategic half of the argument is Clausewitz's and Greene reports it directly: partial victory is surrendered again at the table. "After war come negotiation and the division of territory. If you have only won a partial victory, you will inevitably lose in negotiation what you have gained by war."5
Hence the instruction: "Allow your enemies no options… give your enemies nothing to negotiate, no hope, no room to maneuver."6
This is a genuinely different argument from the emotional one and Greene runs them together. The emotional argument says a wounded enemy will hate you more. The strategic argument says a wounded enemy will bargain — and bargaining is where your battlefield gains get spent. One is about what they feel; the other is about what a negotiating table structurally does to an incomplete result.
They point the same way here, which is why the chapter never has to separate them. But they come apart the moment the enemy has no capacity to hate and every capacity to negotiate, or the reverse, and the chapter has no vocabulary for either case.
In 1934 Mao and some 75,000 poorly equipped soldiers fled into the mountains of western China. Chiang Kai-shek pursued, and by a few years later Mao had fewer than 10,000 soldiers left. Then in 1937 Japan invaded, and Chiang "calculated that the Communists were no longer a threat." He gave up the chase and turned to the Japanese. Ten years later the Communists routed his army; he and what remained of it fled to Taiwan.7
Greene's verdict: "Chiang had forgotten the ancient wisdom of crushing the enemy; Mao had not."8
Now look at what the case actually shows, because Greene's reading requires ignoring the middle of it.
Chiang did not stop out of pity, hope of reconciliation, or sympathy for a former friend — the three mechanisms the law's whole emotional argument runs on. He stopped because a foreign army invaded his country. He made a resource-allocation decision between two enemies under a constraint he did not choose.
That is the case's real content and it is a much better lesson than the one drawn: the instruction crush your enemy totally is silent on what to do when you have two enemies and enough force for one. The chapter presents an unlimited-resources rule and then illustrates it with the one case in which resources were the binding constraint.
There is a threshold buried in the case that Greene walks past. Chiang reduced Mao from 75,000 to under 10,000 — an 87% destruction. The law's claim is that this was worth nothing, that "more is lost through stopping halfway than through total annihilation."9
If that is right, the value curve of destroying an enemy is almost entirely back-loaded. The last few percent carries nearly all the value, and 87% is not most of the way there but essentially nowhere.
That is a real and testable claim. The chapter neither states it nor argues for it.
One instruction in the chapter deserves separating out, because it is the psychological engine the rest depends on:
"There will be people you cannot win over, who will remain your enemies no matter what. But whatever wound you inflicted on them, deliberately or not, do not take their hatred personally."10
This looks like equanimity and it is not. It is an anaesthetic. Taking their hatred personally would require you to hold the fact that you hurt someone — deliberately or not, Greene says, waving both away in one clause — and holding that fact is precisely what generates the pity that the chapter identifies as fatal.
So the instruction is functional: the law cannot be executed by a person who is still able to feel the thing that stops people executing it. Same structure logged at Law 10, where the prohibition on pity turned out to be protection for the size of the intervention rather than a claim about the mechanism. Whenever advice requires you to stop feeling something in order to work, the thing to examine is not the feeling.
After four pages of annihilation language — exterminate, crush, no mercy, Moses having "every last offender slaughtered"11 — Greene turns and says: "It is not, of course, a question of murder, it is a question of banishment. Sufficiently weakened and then exiled from your court forever, your enemies are rendered harmless."12
Flag this in place: the qualification does not fit the chapter it closes. Every case in the law is a killing. Hsiang Yu cuts his own throat. Empress Wu smothers a newborn and executes an empress. The Faliscans and the Canaanites are destroyed. Ramón María Narváez, in the margin, answers a priest asking whether he forgives his enemies with "I do not have to forgive my enemies, I have had them all shot."13
And the "of course" is doing a lot of work. It presents the qualification as something the reader obviously already understood, when nothing in the preceding four pages supports it.
Then a further concession undercuts even the qualification: "if they cannot be banished, at least understand that they are plotting against you, and pay no heed to whatever friendliness they feign. Your only weapon in such a situation is your own wariness."14 So the law's actual fallback, when total victory is unavailable, is vigilance — the very half-measure the chapter spends four pages calling fatal.
You have won something and the other person is still there.
Not destroyed, not gone. Demoted, or moved sideways, or quietly outmanoeuvred in a way everybody noticed. And you feel the thing this chapter is built to remove: a small pull toward making it up to them. You want to be generous now, because you can afford to be.
Stop and separate two questions that feel like one.
The first is are they still able to hurt me — a question about capability, and answerable. What do they still control, who still takes their calls, what would it cost them to move against you.
The second is do they now want to — a question about the wound, and this is the one Greene is actually right about. The half-measure did not leave their hostility where it was. It added being beaten in public to it.
Here is where the chapter is no use and you have to do the work it refuses. Greene's answer is finish it. But look at Chiang: the real question was never is this enemy dealt with, it was what else is coming, and can I afford this. You almost never have unlimited force, and a rule written as though you do will spend everything you have on the wrong opponent.
So the usable move is the one the chapter's own qualification gestures at and then abandons. Not annihilation, and not generosity either — separation. Remove the contact surface. No shared projects, no shared reports, no forum where they have standing over you. You are not trying to make them harmless in general. You are trying to make yourself unreachable by them specifically, which is achievable, cheap, and does not require you to become the kind of person who cannot feel what he has done.
The strongest evidence for the emotional mechanism is Hsiang Yu, whose case is genuinely clean: four separate opportunities, four hesitations, each one explicitly attributed to respect for a former friend, and a reversal the moment the tables turned. The pattern is internal to one relationship rather than assembled from unrelated cases.
Tension: the chapter's two arguments are not the same argument. The emotional one (a wounded enemy hates you more) and the strategic one (a partial victory is surrendered at the negotiating table) point the same direction in these cases and would diverge in others. Greene never separates them, so the law has no way to handle an enemy who cannot negotiate, or one with no capacity for resentment.
Tension: Law 15 against Law 47. Do Not Go Past the Mark You Aimed For; in Victory, Learn When to Stop is the same author, the same book, thirty-two laws later, and it is this law's direct negation. Greene supplies no criterion for which applies. Preserved, not resolved — this is one of the build's matched tension-pairs and is not to be merged.
Tension: the Reversal contradicts the law and then argues past its own counter-example. "It is sometimes better… to leave them an escape route"15 — because a cornered army fights hardest. Then Greene names Versailles as a case where crushing produced decades of revenge, concedes "some would argue that in the long run it would be better to show some leniency," and answers that leniency "may embolden the enemy" — concluding "it is almost always wiser to crush your enemy. If they plot revenge years later, do not let your guard down, but simply crush them again."16 The answer to this produced a catastrophic backlash is do it again. The Reversal raises the strongest objection in the chapter and disposes of it by repetition.
🚩 SINGLE SOURCE · 🚩 SECONDARY WITHOUT PRIMARY — the Hsiang Yu and Empress Wu material is narrated verbatim with no source named; the Moses material is a paraphrase of scripture used as strategic evidence; the Clausewitz passage is quoted without work or page. The one named, dated citation in the chapter is marginal (Ivan Cloulas, The Borgias, 1989, for Sinigaglia) — sixth instance of this build's inverse-typography finding. [POPULAR SOURCE]
Open questions. If Chiang's 87% destruction was worth nothing, the value of destroying an enemy is almost entirely in the last few percent — is that true, and is there any evidence for it beyond the cases selected because it was? And what is the actual base rate: how many partially-defeated enemies return, against how many simply absorb the loss and go elsewhere? The chapter's cases are selected on the outcome it argues for.
The sharpest disagreement in the vault is with R. G. H. Siu, and it is structural rather than a difference of emphasis.
Siu opens Op#67 of The Craft of Power (1979) with a Taoist couplet: "When the sun reaches the meridian, it falls. When the moon becomes full, it wanes."siu1 Then the mechanism:
"As the aggressive corporation approaches progressively closer to its objective of eliminating all competition, the progression reverses itself with the emergence of a new kind of opposition. John Kenneth Galbraith calls it a 'countervailing power.' … 'private economic power is held in check by the countervailing power of those who are subject to it. The first begets the second.'"siu1
Watch what that does to Greene's instruction to make examples. Visible destruction as deterrent is supposed to prevent future challenges. Siu's claim is that it manufactures them — watchers internalise the example and organise against you with whatever they have left. Lao Tzu's compression: "Stretch a bow to the very limit, and you'd wish you had stopped in time."siu2
Both are practitioner traditions citing historical evidence, and the disagreement does not dissolve. What it turns on is system topology. In a closed system with no substrate for countervailing power — a small group, a deep asymmetry, an isolated organisation — Greene's total crushing may hold. In an open system where opposition can self-organise, Siu's reversal applies and completeness breeds the thing it was meant to prevent. Neither writer names the variable, and the chapter's own evidence sits on both sides of it: Empress Wu, inside a sealed court, crushed everyone and ruled forty years; the Treaty of Versailles, in an open field of nations, produced exactly Siu's countervailing power and Greene mentions it and argues past it. See Mortgage of Power for Siu's account of how the bill arrives.
Law 47's Reversal restates this principle and then bounds it — which is the qualification the Law 15 material never receives:
"As Machiavelli says, either destroy a man or leave him alone entirely. Inflicting half punishment or mild injury will only create an enemy whose bitterness will grow with time, and who will take revenge. When you beat an enemy, then, make your victory complete. Crush him into nonexistence. In the moment of victory, you do not restrain yourself from crushing the enemy you have defeated, but rather from needlessly advancing against others. Be merciless with your enemy, but do not create new enemies by overreaching."E1
⚠ That last sentence is the boundary, and it resolves an apparent contradiction between two laws of this book. Law 15 says crush completely; Law 47 says stop at the mark you aimed for. They are not in tension once the object is specified: completeness applies to the enemy you already have, and restraint applies to acquiring further ones.
⚠ Which makes Law 47's Reversal the clearest boundary-clarification in the corpus — after a long run of Reversals that substituted a different technique, denied a limit existed, renamed the subject, or contradicted the law outright, this one simply states what the law meant and where it stops.
And it identifies the failure mode this page describes from the other side. ⚠ Half-measures fail because the injured party survives to retaliate; overreach fails because it manufactures parties who had no quarrel with you. Both are errors about scope, in opposite directions, and Law 47 is where the book finally says so. See Do Not Go Past the Mark You Aimed For.
Psychology — Cold Hatred vs Hot Hatred
That page turns on a paradox: in 1934 the Chancellor of Germany had a minister find Dr. Eduard Bloch, the Jewish physician who had cared for his mother, and ensure he got out safely — while systematising the machinery that would kill six million. The resolution is that cold hatred is directed at a category, not a person, so the individual case is simply irrelevant to it. Hot hatred is episodic, personal, and burns out.
Now put Greene's instruction beside it: "do not take their hatred personally."10
The insight the pair produces, which neither page reaches alone: Greene is not asking you to endure hot hatred, he is instructing you to convert your own into cold. Depersonalising an enemy is exactly the operation that moves hostility from the episodic register to the categorical one — and the categorical register is the one that does not burn out, cannot be reached by an apology, and is indifferent to who the person actually turns out to be. Which supplies the mechanism for the chapter's own central image. The venom grows because the half-measure converts the enemy's hatred from hot to cold too. A hot grudge decays; a person who has been publicly beaten and left alive reclassifies you from someone I am angry at to a category of thing that must be removed, and that version has no half-life. Greene observed the escalation and had no account of why. This is the account, and it also means his instruction and his warning are the same operation performed by the two parties.
History — Mao and the Long March
That page reads the Long March from Mao's side: a fall that looked like cowardice, house arrest on a farm in Hunan, stripped of power by rivals who thought his peasant strategy backward — and an apparent acceptance of it that troubled his friends more than the fall did.
Law 15 uses the same episode from Chiang's side, as proof that stopping halfway is fatal.
Set the two against each other and something falls out that neither has: they describe the same ten years and cannot both be right about what was decisive. Greene makes Chiang's decision to stop the operative cause. The Mao page makes it Mao's own conduct in the interval — what he did with a position of near-total weakness.
They yield opposite advice, which is why it matters. Greene's reading addresses the stronger party and has nothing for the weaker one; the Mao page addresses the weaker party and has nothing for Chiang's constraint. The surviving 10,000 were not passive between 1937 and 1947, and a law organised entirely around what the victor does has no slot for the defeated party's agency at all — which is how the chapter can present a decade as a single omission.
Sharpest implication. The chapter's real content is not be ruthless. It is that a half-measure is a distinct thing from both action and inaction, and worse than either — because it converts a grievance into a classification. That claim survives the removal of everything violent in the law, and it is the part worth keeping: if you are not going to resolve something, the question is not how hard to hit but whether to touch it at all.
Generative questions.