You gave someone a job when they needed one.
Eighteen months on, something has gone slightly wrong and you can't name it. They're a bit shorter with you than they used to be. They correct you in front of other people now, lightly, as a joke. When you mention — carefully, not as a reproach — that things were difficult back when they started, something crosses their face.
So you do the obvious thing. You give them more. A better title, a bigger number, more room.
And it gets worse.
That's the part nobody warns you about, and it isn't ingratitude in the ordinary moral sense. It's not that they're a bad person who forgot. It's that the help itself is doing something to them, continuously, and every top-up makes it heavier.
Greene states the mechanism once, plainly, and it's the most psychologically precise passage in Law 2:
People want to feel they deserve their good fortune. The receipt of a favor can become oppressive: It means you have been chosen because you are a friend, not necessarily because you are deserving. There is almost a touch of condescension in the act of hiring friends that secretly afflicts them.1
Take the pieces apart, because there are three separate claims stacked there.
One: people need to believe they earned what they have. Not prefer — need.
Two: a favour is evidence against that belief. It doesn't merely fail to support it; it actively supplies a competing explanation for the same outcome.
Three: the competing explanation is humiliating, and it arrives attached to the person who's still standing there.
That third one is the engine. If a stranger hands you an unearned advantage, the reminder leaves when they do. If your benefactor is your boss, the evidence is in the building every day.
The intuition most people have is that gratitude and resentment are opposites, and that a big enough favour produces a lot of the first and none of the second.
Greene's model says they're the same reading of the same fact, sorted by what it costs you to hold.
Here's the fork. The favour proves you were chosen. Now: chosen for what?
If you can construe it as chosen-for-merit, the favour flatters you and gratitude is cheap. If you can only construe it as chosen-for-closeness, the favour says something about you that you'd rather not know — that the thing you're now doing, and possibly doing well, came to you for a reason unrelated to whether you could do it.
And the person who gave it is the one person who can never confirm the flattering version, because they were there. Your benefactor is a permanent witness to the wrong story.
That's the condescension Greene says "secretly afflicts them."1 Not condescension in the giver's manner — the giver is usually being warm. Condescension in the structure of the transaction, which persists whatever anyone's manner is.
The failure doesn't announce itself, and Greene's description of the timeline is one of the better things in the chapter:
The injury will come out slowly: A little more honesty, flashes of resentment and envy here and there, and before you know it your friendship fades.1
A little more honesty. That's the tell, and it's an odd one, because more honesty ought to be good.
It reads as candour. It's actually the beginning of the correction — the person adjusting the ledger downward so that what they have starts to look like what they earned. Small deflations of you, delivered as frankness. A joke about a decision you made. A slightly sharper disagreement than the situation needs.
Then the closing line, which is the one worth memorising:
The more favors and gifts you supply to revive the friendship, the less gratitude you receive.1
The relationship has an inverse gradient. Your standard repair tool makes the problem worse in direct proportion to how hard you use it. Which is why the whole thing is so hard to see from inside: the diagnostic that would normally work — I helped and it got worse, so I should help more — is the mechanism accelerating.
There's a second failure running underneath, and it's about information rather than resentment.
Friends often agree on things in order to avoid an argument. They cover up their unpleasant qualities so as to not offend each other. They laugh extra hard at each other's jokes. Since honesty rarely strengthens friendship, you may never know how a friend truly feels.2
So at exactly the moment the resentment starts accumulating, the channel that would tell you about it is the one channel guaranteed to be compromised.
Greene's conclusion: "When you decide to hire a friend, you gradually discover the qualities he or she has kept hidden."1 Not qualities they developed. Qualities the friendship was suppressing all along, and that a working relationship stops paying them to suppress.
You didn't change them. You removed the reason for the performance.
Law 2's transgression case is this mechanism running to completion, and the escalation is legible if you count it.
Michael III made a stable groom his chief councillor. Then, as Basilius proved "avaricious for the perks of power": doubled his salary, tripled it, ennobled him, and married him to his own mistress.3
Four escalating concessions. And Greene's framing of the logic — "Keeping such a trusted friend and adviser satisfied was worth any price"3 — is the inverse gradient stated as an emperor's policy.
Now the psychological reading rather than the political one.
Basilius's appetite looks like greed and isn't, or isn't only. Every gift Michael gave arrived as a gift — which is to say, as further evidence that Basilius held his position by favour rather than desert. The debt never closed, because gifts don't close debts. They document them.
What Basilius needed was a version of his position he could tell himself he'd earned. Michael never once supplied that, and couldn't have, because Michael was the witness.
So Greene's interpretation lands exactly:
They forget the favors they have received and imagine they have earned their success by their own merits.4
That's not moral failure. That's the only available repair. To live with the position, Basilius had to delete the horse — and the man who remembered the horse was the emperor.
The murder resolves an accounting problem.
You're about to help someone who will keep working near you afterward. Do three things before you do it.
First, find the merit and say it, before the help lands. Not afterward, where it reads as consolation. If you're hiring a friend, name the specific thing that made them the right call, to them and in front of other people, and make it true — if you can't find one, you've learned something and should stop.
Second, convert the gift into a transaction as fast as you can. A gift stays open forever; a wage closes monthly. Give them something hard with a real standard attached and let them clear it. People discharge debts through work they can point at, and you want that arriving early, while they still want to.
Third, watch for the little more honesty. Not hostility — you'll catch hostility. The soft correction delivered as candour, the joke with a small edge, the disagreement one degree sharper than the topic warrants. Log it. Three in a month is the leak.
And when you see it, do the thing that feels wrong: give them less, not more. Less help, more responsibility. The instinct will be to warm the relationship with another favour, and that is the one move guaranteed to deepen it.
If you cannot stop being their benefactor, get them a role where someone else is.
Worth noticing as a fact about the book rather than about people.
Greene's marginal column around Law 2 carries an unusual stack of quotations, and they all say the same thing:
Six sources, four traditions, roughly two thousand years, one claim.
Two readings and Greene offers neither. Either this really is a human constant and the convergence is evidence — which would partly answer the 🚩 SINGLE SOURCE flag hanging over the rest of the chapter. Or Greene assembled six quotations that agreed with him, which is what an anthology permits and what its structure conceals.
The Mahābhārata line is the one that adds something the others don't: it locates the problem in asymmetry rather than in friendship. Friendship between equals is fine. It's the rich man and the pauper that can't hold — which is a narrower and more testable claim than the one the chapter title makes.
Strongest support. The mechanism is specified well enough to be falsifiable: it predicts a direction (more help → less gratitude), a timeline (slow, via small honesties), and a trigger condition (the benefactor remains present and senior). Most of Greene's psychology is not this precise.
Tension — the law is stated wider than the mechanism supports. Everything here depends on asymmetry and proximity, not on friendship. The identical dynamic should appear between any benefactor and beneficiary who keep working together, friends or not — and should be absent between friends who never exchange favours. Greene titles it a friendship problem anyway.
Tension — with the vault's own reading of the same transaction. Sacrifice as Social Currency reads the favour from the receiver's side as a binding debt that produces loyalty. This page reads it from the giver's side as a destabilising liability that produces resentment. Same act, opposite predicted outcomes. Neither page names what decides it. Documented, not resolved.
Tension — no exit is offered. Greene's practical advice reduces to "If you never expect gratitude from a friend, you will be pleasantly surprised."11 That's an adjustment to the giver's expectations, not a fix. He identifies a mechanism and then declines to intervene in it.
🚩 SINGLE SOURCE · 🚩 SECONDARY WITHOUT PRIMARY — the psychology is asserted from anecdote and quotation with no empirical citation. 🚩 CONSENSUS MISREPRESENTATION does not apply: Greene claims no research, which is more honest than inventing it. [POPULAR SOURCE] throughout.
Open questions.
Greene's Never Invoke Gratitude, from The Laws of Human Nature, tells you not to remind people what you did for them, because the reminder produces resentment rather than compliance.
Set the two side by side and the later book turns out to be the weaker claim.
Never Invoke Gratitude locates the injury in the reminding. Fix your behaviour — don't mention the debt — and the relationship survives. That's actionable and comforting.
This page says the injury is in the receiving, and runs whether or not anyone mentions anything. Michael never once reminded Basilius about the horse. The corrosion took years and needed no help.
So the twenty-years-later Greene has softened his own finding into advice. The earlier position is bleaker and better specified: if the debt works silently, then tact is not a solution, and the only real interventions are structural — supply a merit story, convert the gift to a wage, or put the person somewhere you are not their benefactor.
Same author, and the book with less psychology in it has the sharper psychology.
→ behavioral-mechanics · Sacrifice as Social Currency (mandatory psychology → behavioral-mechanics handshake: this page describes a mechanism with a direct tactical counterpart)
The tactical page treats the favour as an instrument: give, and you create an obligation you can later draw on. That's the operator's version, and it's how the favour is taught everywhere from sales training to statecraft.
This page is the same transaction seen from inside the person carrying it — and what it reports is that the obligation doesn't sit there inertly waiting to be drawn on. It ferments.
Put them together and you get the timing rule neither page states.
If the favour creates a debt and the debt converts to resentment on a schedule, then a favour is not a stored asset. It's a depreciating one, and the depreciation accelerates the longer the beneficiary has to live next to the evidence. Call it in early and you're drawing on gratitude. Call it in late and you're drawing on a grievance, and you'll get compliance with interest charged — the little more honesty, the sharpened disagreement, eventually the refusal delivered with an impudent look.
Which means the operator's version has an expiry date on it that operators are never told about. Michael III held his favour for years, kept adding to it, and cashed it exactly once — asking for money back — and that request is what triggered the murder. He drew on an asset that had turned into a liability without changing its appearance.
→ business · Reciprocity and Obligation as Manipulation Technique
The business page handles reciprocity as a designed funnel mechanic: give value first, unrequested, and the recipient becomes disproportionately likely to buy.
It works, and this page explains the boundary condition that makes it work — which the marketing literature generally doesn't state.
Commercial reciprocity is transactional and terminal. The free tool, the sample, the useful newsletter: the recipient buys, the ledger closes, everyone leaves. The benefactor is a company, not a person in the room, and there is no ongoing relationship in which the recipient's self-concept has to accommodate having been helped.
Change either variable and the mechanism inverts. Make the benefactor a named individual the recipient must keep seeing, and remove the settling transaction, and you get Basilius.
The insight neither field produces alone: reciprocity is safe to deploy exactly to the degree that it can be discharged and exited. The techniques that scale commercially are the ones with a checkout button. Applied to a colleague, a friend or a subordinate — where there's no checkout and no exit — the same technique doesn't create obligation, it creates a permanent asymmetry the other person has to metabolise. Marketers get gratitude because their customers get to leave.
Sharpest implication. The standard repair for a cooling relationship — do something generous — is, in any relationship with a standing power asymmetry, the thing that broke it. Greene's line is the whole finding in one clause: the more favors and gifts you supply to revive the friendship, the less gratitude you receive.1 The tool and the wound are the same object, which is why intelligent, well-meaning people run this failure to completion while congratulating themselves on their generosity.
Generative questions.
sacrifice-as-social-currency resolved by asking whether the receiver can ever reach parity?