Someone reorganised your team. They genuinely thought it would help. They said so, they meant it, and you believe them.
Eight months on: two people have left, the thing you were good at is split between three people who each do a third of it badly, and your Wednesdays now contain a meeting that exists to manage the confusion.
Was that a good decision?
There's a strong pull to say well, their heart was in the right place. Greene's point is that this sentence is a measurement error — not an act of generosity, an error, of the same kind as misreading a thermometer.1
Greene's operative sentence is "you measure their strategy and their power by what you can see and feel."1
The reason is availability. You have direct access to consequences and none at all to someone's interior. Intentions reach you only as reports about interiors, and reports are authored.
He says so outright — "how often are someone's intentions made the issue only to cloud and deceive!"1 Intentions aren't merely unobservable; they're the part of the account the other person controls, and therefore the part most worth manipulating. Anyone can supply a motive after the fact. Nobody can supply a different outcome.
The rule reduces to: prefer the variable your counterparty can't edit.
Then the instruction acquires a temperature. You must learn to "inwardly laugh"1 at anyone's account of their own good intentions.
Not discount. Not note and set aside. Laugh — privately, behind a face showing nothing.
Discounting stated motives is a sound epistemic policy. Laughing at them is a stance toward everyone you deal with, with a prediction baked in: whatever they say about why, it's cover.
Greene does notice something true under the sneer. People "always assuming that they have acted out of goodness"1 mostly believe their own account — the cover story is generally sincere. So the laugh lands not on liars but on humans narrating themselves favourably, and he has no separate category for the person doing that in good faith. On his account, everyone is.
The sentence doesn't stop where you'd expect. Those moral judgements are "really an excuse for the accumulation of power."1
Two claims stacked, one argued.
One: moral judgements are unreliable instruments for assessing actions. Defensible.
Two: moral judgements are themselves power moves. Asserted in a subordinate clause.
The difference matters. One tells you to set moral language aside as noisy data. Two tells you anyone using it is making a bid — so moral objection can no longer be received at all, only diagnosed. "What you did was wrong" loses any reading in which it's simply true.
Balzac, in the nearby margin, gives the doctrine neat: "There is no good and bad, there are only circumstances."2
The passage's own worked image is the most carefully built thing in the Preface, and four phrases carry it.3
"Your opponent sits opposite you." Across a board, not beside you. Every relationship covered is adversarial by default — and Greene applied the rule to "your friend or rival"1 a paragraph earlier, so the friend is at the board too.
"Taking nothing personally." The book returns to this phrase at Law 48 as the first psychological requirement of formlessness. Here it's sportsmanship. There it's armour.
"The politeness of those you are playing with more than their good and sweet intentions." The doctrine at its best. Politeness is observable — a behaviour, delivered reliably, at cost. Good intentions are a claim. Not style over substance: verifiable substance over unverifiable.
"Do not be distracted by anything else." Where the cost lands. Anything else includes the constraints someone was under and the difference between malice and incompetence — both of which change what you should do next, neither readable from effects.
A chess game has no version where two players build something together, so a model built on it can't represent that case.
You're writing up the restructure, and you'll write the paragraph twice.
First, the one that comes naturally: the reorg was well-intentioned but poorly executed. Read it back. It's already done its work — "well-intentioned" absorbs the impact, "poorly executed" names a problem with no owner. Nobody will act on it. You wrote it to be sayable.
Now delete every word about motives and write what you can see: two departures, one capability split three ways, one recurring meeting, eight months.
Side by side, the second is harder to read, impossible to argue with, and accuses nobody. You didn't get colder. You removed the variable the other person controls.
Before sending, ask what the rule can't: does this person keep producing this outcome? One bad result with good intent is someone who got it wrong. The fourth is a pattern, and intent has stopped mattering — not because it's irrelevant but because it's been tested. That's the line between assessment and contempt.
Skip the laugh. It isn't part of the measurement. It's the part that costs you the relationships you'd still be learning from.
Effects-only measurement is excellent at assessment and incapable of investing ahead of evidence.
Every collaboration and every hire begins where there are no effects yet — only what someone says and your read on whether they mean it. The rule says wait for results: sound for an adversary, and a description of how not to start anything.
The second gap bites harder. Effects don't distinguish malice from incompetence, and one calls for training while the other calls for distance. "Do not be distracted by anything else"3 forbids the enquiry that would separate them.
The overreach is scope. Offered for "your friend or rival"1 unadjusted, and applied to a friend it produces someone who can be assessed but never trusted — because trust is a bet placed before the effects are in.
The standard defence of this book — in this vault's own source stub, and from most of its readers — is that it's descriptive rather than prescriptive.
That is a defence by intention, and by the rule on this page it's inadmissible. The book must be assessed on what you can see and feel: what it does to readers, what it licenses, what it gets quoted for.
That cuts both ways, which is why it's worth running rather than assuming the answer. Much of what this book produces is recognition — seeing a move run on you that you'd previously experienced as your own inadequacy. That's an effect too, and a good one.
The point is only that "he was just describing it" is the one answer the book rules out of order.
Strongest evidence. The availability argument stands without the cynicism attached to it: intentions arrive as authored reports, effects don't. The politeness line applies the same insight well — verified behaviour beats claimed disposition.
Tension 1 — two claims, one defended. That moral judgements are poor instruments is argued; that they are "really an excuse for the accumulation of power"1 is asserted in a subordinate clause, and it's the claim that makes moral objection unhearable.
Tension 2 — the Preface contradicts this page four lines later. l.384 says be distracted by nothing but results.3 l.388 calls hidden motives "the single greatest piece of knowledge you can have."4 Opposite instructions on where to look, unacknowledged. Developed at Study Everyone, Trust No One Completely.
Tension 3 — the rule can't be applied to the book without contradicting the book's own defence.
🚩 SINGLE SOURCE · 🚩 SECONDARY WITHOUT PRIMARY. The passage contains no case, no citation and no example — pure assertion, unusually for this book. All claims [POPULAR SOURCE].
Open questions. Is there a formulation that keeps the availability argument while preserving the malice/incompetence distinction — or does admitting that distinction readmit motive-reading and its manipulability? And does this doctrine model trust at all?
Greene abandons this rule four lines after stating it, then spends two later books doing so systematically.
The local version is the sharpest evidence: at l.388 hidden motives become "the single greatest piece of knowledge you can have."4 Two paragraphs after telling you to attend to nothing but results. He doesn't qualify the first rule or note that the second overrides it.
Then the same reversal at book scale. Seven Courtier Types, from The Laws of Human Nature, is entirely motive-reading — identifying the flatterer, the schemer, the gatekeeper, each a claim about what someone is trying to do, made before the effects.
The later book is right, and the reason shows what's wrong here. Effects-only is a retrospective instrument; it can only score what has happened. Anything predictive requires the motive-reading this passage forbids. Greene needs it to write a book about anticipating people, and opens by ruling it out.
The reconciliation both books support: judge conduct already taken by effects, and what someone is about to do by reading them. Two questions, two instruments. The Preface collapses them and the later corpus quietly abandons the collapse.
Handshake 1 — history: Karma — Consequence Mechanics
Greene's rule has a very old competitor, and they disagree about something specific and checkable.
The karma material is consequence machinery: acts propagate, lawfully rather than judicially. So far this looks like Greene — outcomes, not opinions about outcomes. But the traditions divide on the input. Where intention (cetanā) is what makes an act karmically weighty, the same visible outcome carries different consequence depending on the interior state that produced it. Where the mechanics run on the act itself, it doesn't.
So this corpus has spent two millennia on precisely the question Greene settles in one clause, and hasn't settled it — which is itself evidence the question is harder than the clause admits.
The insight neither produces alone: the two are answering different questions that look like one. Greene asks how do I assess this person's power and strategy? — where intention is worthless, being unobservable and manipulable. The karmic account asks what did this act do to the person who committed it? — where intention is the entire variable, because the interior state is what gets laid down as disposition. Both are consequence-based; they differ on whose consequences are measured. Greene has no account of what running his doctrine does to the person running it, and this is the domain that does.
Handshake 2 — psychology: Fantasy Bond
Aimed where Greene aims it, this rule is cynical. Aimed where he'd never aim it, it's the standard clinical corrective.
A fantasy bond is a connection maintained on the idea of the relationship rather than what it delivers. It survives on stated intent and remembered warmth, and it survives specifically because the effects are never audited. They didn't mean it. They're under pressure. They love me really. Each of those is an intention offered in place of an outcome.
Now apply "measure their strategy by what you can see and feel"1 there. You get the intervention: stop scoring the relationship on declared feeling and score it on what happens. Who calls. What happens when you need something. How you feel on the drive home. Not cruelty — the thing that lets someone finally see a situation they've been narrating for years.
What the two produce together: the effects rule isn't intrinsically adversarial. It's a corrective for whichever variable is being used as cover, and in intimate life the cover is almost always intention. Which reframes the page. Greene presents a competitive instrument for reading opponents across a board; its highest-value use is on the relationships he seats at that board as opponents,1 where it functions as clarity rather than contempt — and where the "inwardly laugh" instruction is precisely the part that would wreck it.
Sharpest implication. The rule is sound as epistemic policy — prefer the variable the other party can't author. Including the subordinate clause turns it into something else, because "moral judgements are really an excuse for the accumulation of power" makes moral objection undeliverable. Keep the measurement, drop the clause and the laugh, and you have one of the book's most useful instruments. Keep all three and you have someone who can assess everybody and be reached by nobody.
Generative questions.