At the end of Law 15, after everything else, Greene writes this:
"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."1
Read it cold and it is reasonable. Read it in position — as the last paragraph of the chapter it closes — and something is wrong with it, and the wrongness is worth naming precisely rather than being smoothed over.
The qualification does not fit the chapter it closes. This build flags it in place rather than resolving it, per the standing rule on the source's flagged material: state the situation factually, completely, without extra hedging and without editorialising.
Four pages of annihilation language, and the cases are not metaphorical.
Hsiang Yu cuts his own throat.2 Empress Wu smothers a newborn, poisons a niece, poisons an heir, and executes an empress.3 Moses, called the law's first practitioner, has "every last offender slaughtered" after the Golden Calf, and instructs his followers to "utterly destroy" the tribes of Canaan and "show no mercy to them."4 Cesare Borgia's captains are throttled indoors while their troops are destroyed outside.5 In the margin, Ramón María Narváez is asked on his deathbed whether he forgives his enemies and answers: "I do not have to forgive my enemies, I have had them all shot."6
The chapter's own vocabulary: exterminated, crushed, annihilate, no mercy, "never go halfway."7
Not one of the law's cases is a banishment.
The two words that carry the most weight in the qualification are of course.
They present the limitation as something the reader already understood — as though the preceding four pages had obviously been figurative, and only a careless reader could have thought otherwise. Nothing in those pages supports that. The cases are killings, described approvingly, with the killing identified as the reason they worked.
Of course is a retroactive frame. It does not add a limit to the argument; it asserts that the limit was always present, and it does so at the one point in the chapter where a reader might otherwise stop and ask what exactly is being recommended.
This is worth logging as a technique rather than a slip, because it is efficient. A stated qualification invites scrutiny of the gap between it and the text. A qualification presented as obvious invites the reader to feel they had understood all along.
Set the framing aside, though, because the recommendation is not empty. Exile is a real strategic instrument and Greene's description of why it works is accurate: "Sufficiently weakened and then exiled from your court forever, your enemies are rendered harmless. They have no hope of recovering, insinuating themselves and hurting you."1
Two conditions are doing the work there, and both are strong.
Sufficiently weakened. Exile alone is nothing — a powerful enemy sent away is a powerful enemy elsewhere, which is the Long March. The weakening has to come first.
From your court forever. The permanence is the second condition, and courts are the specific setting where it can be enforced. A court has a boundary and a gatekeeper. Most environments do not.
Where both hold, banishment achieves the law's stated goal — removal of the capacity to act on you — at a fraction of the cost, and without the reputational consequences of the alternative. It is the better instrument. Greene is right about that and wrong to present it as what he had been describing.
The paragraph does not stop there, and the next sentence is the real problem:
"And 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. If you cannot banish them immediately, then plot for the best time to act."8
Take that seriously. When total victory is unavailable and banishment is unavailable, the law's recommended posture is wariness.
Now recall what the chapter has spent four pages calling fatal. Hsiang Yu's error was leaving an enemy alive and watching him. Chiang Kai-shek's error was leaving an enemy alive and turning away. The half-crushed viper's whole point is that vigilance does not contain it — "time makes the venom grow stronger."9
So the law's fallback is the behaviour the law identifies as the failure mode. And it is not a small case: for most readers, most of the time, neither annihilation nor permanent exile is available, which means the fallback is not an edge condition. It is the ordinary condition, and the chapter's answer to it is the thing the chapter is against.
One reading fits everything on the page: the qualification is not addressed to the strategic argument at all. It is addressed to the reader's discomfort.
It arrives exactly where a reader would be most likely to close the book — after the newborn, after Moses, after the deathbed line about having them all shot. It does no strategic work; the preceding argument neither requires nor is improved by it. It changes none of the cases. What it does is retroactively reclassify what was just read as something less alarming than it was, using two words, and then move on.
That reading is consistent with the Preface's documented technique of pre-labelling the reader's objections as incomplete understanding — the same move, at a smaller scale, applied at the moment of maximum resistance.
You have decided somebody has to go.
Not destroyed — you are not in that business and neither, on the page, is anyone reading this. Just gone: out of the decision, off the project, no longer someone whose agreement you need.
The chapter's two useful conditions are the ones to actually work with, and they come in order.
Weakened first. Removing someone who still has standing produces a person with a grievance and an audience, which is worse than where you started. So the sequence is not remove-then-justify; it is establish the case, let the position become obviously untenable to people other than you, and only then move. If you cannot name three people who would independently agree the situation had become unworkable, it is too early.
Then permanent, and permanence is structural. Exile fails when the boundary is porous — when they still have the standing calls, the old alliances, a route back through someone who owes them. Removing a title while leaving the relationships intact is not banishment, it is a demotion, and demotions are the half-measure this whole law is about.
And when neither is available, which is most of the time, do not take the chapter's advice. Wariness is not a strategy; it is a permanent tax on your attention that the chapter itself says does not work. The thing that does work is the one Greene's own qualification gestures at without naming: reduce the contact surface. Not their power — your exposure. No shared deliverables, no forum where they have standing over you, no dependency that requires their sign-off. You are not making them harmless. You are making yourself unreachable, which is achievable without anybody being removed at all.
The evidence for the mismatch is simply the chapter's own contents: five cases, all killings, and a closing sentence saying killing is not what was meant.
Tension: the fallback contradicts the law. Wariness is the recommended posture when banishment is unavailable, and wariness is what the law calls fatal. Documented, not resolved.
Tension: the qualification and the Reversal pull opposite ways. The Reversal, immediately following, says it is "almost always wiser to crush your enemy" and that if they plot revenge years later you should "simply crush them again."10 The qualification softens; the Reversal re-hardens, twenty lines later, and the chapter closes on the harder position.
Tension: none of the cases are re-read in light of it. If banishment was always what was meant, Empress Wu — the law's flagship observance — is not an example of the law. Greene does not revisit her.
🚩 SINGLE SOURCE · 🚩 SECONDARY WITHOUT PRIMARY throughout the chapter's case material. The Moses passages are scripture paraphrased and deployed as strategic evidence with no citation. [POPULAR SOURCE]
Open questions. Is there any case anywhere in the book where banishment alone produced the outcome the law claims, or is the instrument recommended and never demonstrated? And the structural one: given that most readers can neither annihilate nor exile, what fraction of this book's advice is written for a position almost none of its readers occupy?
Against Clausewitz and the Negotiation Trap, built from the same chapter, the qualification fails on its own terms. Clausewitz's mechanism does not care whether an enemy is dead or exiled — it cares whether they retain standing to be negotiated with. An exiled enemy with residual claims is precisely the counterparty that argument warns about. So banishment satisfies the emotional half of Law 15 and not the structural half, and Greene's closing sentence quietly abandons one of his own two arguments without saying so.
Against Empress Wu Chao, the conflict is flatter. Wu is the law's only long-horizon success and her method was murder; the one banishment she used (an illegitimate son, exiled on trumped-up charges) was an instrument among several, not the strategy. The law's best evidence is incompatible with the law's stated limit, and the chapter runs them four pages apart without contact.
Psychology — Moral Agency and Categorical Responsibility
That page opens on the explanatory abyss under blame. At what point in a causal chain does someone become responsible? They committed murder — but did they choose the rage disorder, the testosterone, the impulse-control wiring, the developmental environment, the genes? Each step back dissolves the categorical boundary that "responsible" requires.
Greene's of course is the same operation run in the opposite direction, and much faster.
Where the psychology page dissolves a moral category by tracing it backward until it will not hold, Greene installs one instantly, with two words, and never traces anything. Murder and banishment are presented as obviously distinct kinds of act — and on the page they are separated by nothing except the assertion.
What the pair produces: the sharpness of a moral boundary and its defensibility are unrelated. The psychology page shows a boundary everyone treats as absolute becoming unfindable under examination. Greene shows a boundary nobody had drawn becoming absolute without any examination at all. Both cases point at the same property — moral categories are asserted rather than derived, and the assertion is what does the work. Which is why of course is so efficient: it does not argue for the boundary, it reports that the boundary was already there, and a reported boundary is far harder to contest than an argued one. You cannot rebut a premise nobody defended.
Business — Us vs Them, Not Us vs Us
That page's argument is about direction. Internal competition — leaderboards, top closer of the week — produces short-term gains and slow structural damage: people stop helping, hoard information, protect leads, sabotage. Aim the same competitive energy at an external rival and cohesion rises instead.
The connection to banishment is not obvious until you ask what removing someone does to everyone else.
Law 15 treats elimination as a two-party transaction: you and the enemy, and when they are gone the problem is over. The team page says that is never what happens, because the removal is observed. Colleagues watch someone get exiled and update their model of what the environment is — and specifically, they learn that internal conflict is resolved by removal.
The insight the pair produces: banishment converts a two-party problem into an n-party signal, and the signal is precisely the one the business page identifies as corrosive. Every observer now knows the cost of being on the wrong side of an internal dispute, which is exactly the condition under which people stop sharing information and start protecting themselves. So the instrument Greene recommends as the cheaper option is only cheaper on the ledger he is looking at. Wu could ignore this because her court had no productive collaboration to damage. Any environment where the remaining people have to work together afterwards is paying a cost the chapter has no line item for — and it scales with how visible the removal was, which is the opposite of what "make an example" recommends.
Sharpest implication. The qualification's real function is not to limit the argument but to make it readable. It changes nothing about what the chapter recommends or how its cases worked, and it arrives precisely where a reader would otherwise stop — which means the most important sentence in Law 15 is doing rhetorical work rather than strategic work, and it is the sentence most likely to be remembered as the chapter's actual position.
Generative questions.