It's 1807 and Talleyrand has decided Napoleon is going to ruin France. He wants to move against him, and for that he needs one partner — one person willing to conspire against an emperor.
Think about who you'd pick. Someone you've known for years. Someone who owes you.
Talleyrand picked Joseph Fouché: head of the secret police, his most hated enemy, and a man who had previously tried to have him assassinated.1
That's not recklessness. Greene lays out three separate reasons, and each one is a reason the obvious choice would have been worse.
Greene's account of Talleyrand's thinking is unusually specific, so it's worth taking as a list:
He knew that their former hatred would create an opportunity for an emotional reconciliation. He knew that Fouché would expect nothing from him, and in fact would work to prove that he was worthy of Talleyrand's choice; a person who has something to prove will move mountains for you. Finally, he knew that his relationship with Fouché would be based on mutual self-interest, and would not be contaminated by personal feeling.1
Reconciliation is an event. Hatred converted has a moment attached to it — a scene, a handshake, a decision. Friendship has no such moment; it's a background condition, and background conditions don't generate commitment.
Something to prove is fuel. The former enemy starts at a deficit and knows it. Greene: "a person who has something to prove will move mountains for you." The friend starts at a surplus, and a surplus generates nothing.
The third reason is the one people misread as coldness, and it's the most useful of the three.
"Not contaminated by personal feeling" means the relationship runs on interests — and interests are legible. You can check them, price them, and notice when they change.
Affection is not checkable. There's no reading you can take, no quarterly number, nothing that tells you the balance has shifted. Which is precisely what killed Michael III: he thought he'd secured a person, and had no instrument for noticing that the person's position had inverted.
An alliance you can audit is worth more than one you have to trust.
Greene's instruction: "Whenever you can, bury the hatchet with an enemy, and make a point of putting him in your service."1
Here's the mechanism underneath all three, stated as one thing.
Law 2's other half — the burden of the favor — says the friend you help accumulates an undischargeable debt that ferments into resentment. The friend is carrying a liability they can't put down.
The former enemy is carrying the opposite balance. They arrive owing you nothing and expecting nothing, with a documented reason for you to distrust them, and therefore with everything to gain from performance. Every good day is a data point in their favour.
So the two halves of the law are one mechanism read at two signs. Favour creates a debt that corrodes. Enmity converted creates a credit position the other person is motivated to build.
That's why Greene's advice isn't sentimental and isn't really about forgiveness. He's telling you which ledger produces effort.
The 1971 case is smaller and more precise.
Kissinger had been the target of an unsuccessful kidnapping conspiracy — the Berrigan brothers, four more Catholic priests, four nuns.2
What he did next: privately, without informing the Secret Service or the Justice Department, he arranged a Saturday-morning meeting with three of the alleged kidnappers.2
He explained he'd have most American soldiers out of Vietnam by mid-1972, and charmed them completely. They gave him some "Kidnap Kissinger" buttons. One of them stayed a friend for years and visited him on several occasions.2
Two details carry the weight.
Saturday morning, no agencies. The setting removes the institutional frame entirely. Through Justice, they're defendants and he's a complainant, and nothing can happen except procedure. In a room on a weekend with nobody logging it, they're people who disagree about a war.
He gave them something real. Not a plea, not a warning — a timeline. Information that mattered to them, which is the only currency that would have worked on people who had been willing to commit a felony over the issue.
Greene notes it wasn't a one-off: Kissinger "made a policy of working with those who disagreed with him," and colleagues said "he seemed to get along better with his enemies than with his friends."2
The Talleyrand–Fouché case has an inconvenient feature that Greene reports and then walks past, and it's worth stopping on because it's the strongest test of the claim.
The conspiracy failed. They did not topple Napoleon.1
By outcome, then, picking Fouché didn't work. So why is this the illustration?
Greene's answer is in the sentence right after:
although the conspirators did not succeed in toppling Napoleon, the union of such powerful but unlikely partners generated much interest in the cause; opposition to the emperor slowly began to spread. And from then on, Talleyrand and Fouché had a fruitful working relationship.1
Two returns from a failed operation.
First, the improbability was itself the message. Talleyrand and Fouché appearing on the same side was information: it told everyone watching that opposition to Napoleon had crossed factional lines and was no longer one faction's grievance. A friend as partner would have carried no such signal, because two allies agreeing is not news.
Second, the alliance survived its own failure. That's the real evidence. A partnership built on affection that fails a major operation usually ends in blame — affection makes betrayal feel like betrayal. A partnership built on mutual interest survives a loss intact, because nobody was owed anything in the first place and the interests are still there in the morning.
So the case proves something narrower and more useful than "recruit your enemies and you'll win." It proves the relationship is more durable than the friendship version under exactly the conditions that break friendships. Greene chose a case where the plan failed, which is either careless or the most honest selection in the chapter.
There's someone at work who moved against you. Blocked your project, briefed against you, took a role you wanted and wasn't gracious about it.
You've been routing around them for eight months. Stop and price it.
Routing around costs you a permanent tax — every plan gets shaped by avoiding one person, and everyone can see it. And they're paying the same tax, which means there are two of you spending real effort on a cold war neither of you is winning.
So: find the thing you can give them that costs you least and helps them most. Not an apology — apologies are about the past and they'll hear it as positioning. Something they need. Information, access, cover in a meeting, a piece of scope they'd want.
Give it without a request attached. Then say nothing about it.
What you're testing for is the deficit reflex. A converted enemy overcorrects — they'll want to square it, fast, and they'll do it with effort rather than warmth. If that comes back, you have a working relationship, and it will run on interests you can both name.
If nothing comes back, you've spent one favour and learned they're not convertible, which is worth knowing precisely.
One rule throughout: do not try to make them like you. The value is not affection. Kissinger's colleagues noticed he did better with enemies, and the reason is that nobody was pretending. Keep the transaction visible and it stays checkable.
Greene then makes a turn that most readers of this law miss, and it's the more interesting half.
Without enemies around us, we grow lazy. An enemy at our heels sharpens our wits, keeping us focused and alert. It is sometimes better, then, to use enemies as enemies rather than transforming them into friends or allies.3
So conversion isn't always the play. Sometimes the enemy is more valuable left in place, doing the job of keeping you sharp.
That's a real tension inside one law, four lines apart, and Greene doesn't mark it as one. Recruit them, or keep them — the chapter gives you both and no rule for choosing.
The nearest thing to a rule comes at the end: "you are far better off with a declared opponent or two than not knowing where your real enemies lie."4 Which suggests the variable is visibility, not hostility. A declared enemy is a known quantity. Converting them may buy you a partner and cost you an instrument.
Strongest support. Two cases from different centuries and domains, both with named actors and specified mechanisms, plus a stated three-part rationale rather than a moral. The Kissinger case is unusually well-attested by the standards of this book.
Tension — recruit them or keep them? l.631 says bury the hatchet and put the enemy in your service. l.637 says it is "sometimes better… to use enemies as enemies." Both are stated as general guidance. No criterion is offered. Preserved.
Tension — the flagship case failed. The conspiracy didn't work. Greene keeps the illustration and repositions the payoff to the alliance's durability. Defensible, but it means the case does not demonstrate the thing the law's title promises.
Tension — against Law 15. Crush Your Enemy Totally will argue that a surviving enemy always returns. This law says a surviving enemy is your best hire. Documented pair for the tension log; not resolved here.
🚩 SINGLE SOURCE · 🚩 SECONDARY WITHOUT PRIMARY — no citation for Talleyrand's reasoning, which is presented as his interior thought process across three numbered motives. That is a strong claim about a private mental state with nothing behind it. [POPULAR SOURCE] throughout.
Open questions.
Greene's Never Invoke Gratitude, from The Laws of Human Nature, warns against reminding people of favours because the reminder generates resentment.
Read against this page, the two produce a joint recommendation neither states.
If invoking a debt breeds resentment, and if the former enemy arrives carrying no debt, then the converted enemy is the one relationship where you never face the temptation in the first place. There's nothing to invoke. The motivation is self-generated and self-renewing, and you don't have to manage it.
Which reframes Law 2's advice as something more specific than "distrust friends." It's a claim about which relationships require ongoing maintenance. The friend needs constant topping-up that makes things worse; the recruited enemy runs on their own fuel until the deficit closes.
That last clause is where the two books stop helping, and it's the open question above.
→ psychology · Reciprocal Altruism
The evolutionary account: cooperation between non-kin is stable when there's repeated interaction and a memory of who defected, because the shadow of future encounters makes defection expensive.
Talleyrand and Fouché have that structure and one strange inversion.
Standard reciprocal altruism predicts that a known defector is the worst possible partner — Fouché had already tried to have Talleyrand killed, which is a maximum-severity defection on record. The model says exclude him.
Talleyrand read the same fact as a qualification. And the reason both can be right is that the model prices reputation as a deterrent, while Talleyrand priced it as leverage. Fouché's documented treachery is exactly what makes his cooperation costly to fake and valuable to demonstrate — he cannot coast, because the record is public.
The insight the pairing produces: in a repeated game with observers, a partner with a bad public record is more constrained than one with a clean one, because they have no reputational slack to spend. The clean partner can defect once and survive it. The known defector cannot. Prior betrayal, once converted, functions as a hostage — and that's a use of reputation neither the evolutionary model nor Greene's version states, though Talleyrand appears to have been operating on it.
→ history · Emperor Sung and the Banquet of Resignations
Sung's three demonstrations — the estates, the cup of wine, the returned conspiracy papers — are this law's mechanism run by someone holding total power.
Placing them together exposes a variable Greene keeps invisible.
Talleyrand recruited Fouché from parity. Neither could destroy the other; the reconciliation was a negotiation between two people who each had something the other needed.
Sung converted Liu and Ch'ien Shu from total dominance. He wasn't negotiating. He was declining to execute.
Both produce loyal former enemies, and the loyalty is not the same substance. Talleyrand got a partner whose interests happened to align and would realign if conditions changed. Sung got men who had been to the edge and been pulled back, which is a bond that doesn't track interest at all.
So the technique has two modes with different failure profiles, and the chapter blends them. The parity version is durable but conditional — it survives a failed conspiracy, as Talleyrand's did, and would not survive a genuine divergence of interest. The dominance version is unconditional but requires you to have already won, and cannot be run by anyone who actually needs allies.
The people who most need this law can only access its weaker half.
Sharpest implication. The thing that makes a former enemy valuable is not that they've changed their mind. It's that they hold a deficit and know you know — and the whole mechanism runs on that asymmetry, which means it decays as they close it. Greene sells conversion as a permanent upgrade over friendship. The structure he describes is a temporary state with an expiry date nobody prices, and the person on the other end of a closed deficit is just a friend again, with all of Law 2's problems restored.
Generative questions.