Law 47's Reversal contains one of the more visible self-contradictions in the book, and it happens across a paragraph break:
"Your caution after victory should never make you hesitate, or lose momentum, but rather act as a safeguard against rash action. On the other hand, momentum as a phenomenon is greatly overrated."1
Momentum is a thing you must not lose. Momentum is a phenomenon that is greatly overrated. Adjacent sentences, same paragraph, no acknowledgement.
The contradiction is worth preserving rather than smoothing, because it is a real seam in the argument and because a reader taking either sentence alone will get the wrong instruction. But it is also resolvable, and the resolution turns out to be more useful than either sentence.
There are two things called momentum and Greene uses both without marking the switch.
Momentum as tempo. The rate at which you act. This is real, it is yours, and slowing it after a win is exactly the over-caution the Reversal's first half attacks. Nothing metaphysical about it — you either keep doing things or you stop.
Momentum as force. The belief that a run of successes has acquired a property of its own, which will carry the next one, and which is depleted if you interrupt it. This is what he calls overrated, and the denial is: "You create your own successes, and if they follow one upon the other, it is your own doing."2
Under that reading the paragraph is coherent. Maintain your rate; do not believe in the streak. And the distinction has teeth, because the two prescribe opposite things at exactly the moment they most feel identical — the force reading says keep going with this, and the tempo reading is entirely satisfied by keeping going with something else.
The mechanism Greene gives against it is the strongest thing in the passage: "Belief in momentum will only make you emotional, less prone to act strategically, and more apt to repeat the same methods."2
That last clause is the one that connects. Elsewhere in the chapter he argues that the pattern which produced your success is the likeliest thing to destroy you. But a pattern needs a justification to survive scrutiny — someone eventually asks why you are doing the same thing again, and because it worked is a thin answer that invites an argument about whether conditions have changed.
Momentum is the answer that closes that conversation. It converts repetition from a choice into a necessity: you are not repeating the method, you are preserving the run, and interrupting it would cost you something. See The Pattern That Worked Is the Trap.
So momentum-belief is not primarily a mistaken theory about probability. It is the licence that lets pattern repetition go unexamined, and that is why attacking it matters more than the metaphysics would suggest.
Greene's dismissal is contemptuous and revealing: "Leave momentum for those who have nothing better to rely upon."2 Read as a diagnostic rather than an insult, it says momentum is what you cite when you cannot name the cause — the explanation that arrives when the real one is unavailable or unflattering.
The momentum seam can be repaired. This one cannot, and it is larger.
Four paragraphs earlier, in the Keys to Power, Greene instructs the victor to be "recognizing the role of luck and circumstance in your success,"3 and builds the Borgia case entirely on the claim that unrecognised good fortune destroys people.
Then, in the Reversal: "You create your own successes, and if they follow one upon the other, it is your own doing."2
Those are not two senses of a word. They are incompatible descriptions of where success comes from, in the same chapter, a page apart.
What each is for explains why both are there. The luck passage is deployed against complacency — you need the reader to feel that the ground under a win is unstable. The self-authorship passage is deployed against streak-thinking — you need the reader to feel that nothing is carrying them and they are producing each result individually. Both do useful work on their respective errors and they cannot both be an account of causation.
Preserved, not resolved. The honest synthesis available in the material: your successes are your doing, and their timing is not — which permits the humility the first passage wants without granting the streak the second passage denies. Greene does not write that sentence, and the chapter would be better with it.
The Reversal's first half describes a failure the rest of the chapter never mentions: "There are some who become more cautious than ever after a victory, which they see as just giving them more possessions to worry about and protect."1
This is the exact opposite of the law and it is a real condition. Victory converts you from an attacker into a holder, and holders play differently — the same move now risks something that previously did not exist, so its expected cost rises even though its expected return has not changed.
Greene's correction is a distinction rather than a prohibition: caution "should never make you hesitate, or lose momentum, but rather act as a safeguard against rash action."1
Safeguard against rash action, not brake on action. The difference is whether the caution is attached to how you move or to whether you move — a filter versus a governor. A filter examines each move and rejects the reckless ones; a governor reduces the number of moves. They feel the same from inside, and only the first is what Law 47 is asking for.
🚩 REPLICATION RISK, and unusually the flag points in an interesting direction.
Greene's denial of momentum-as-force reads as settled common sense, and it aligns with what was, when The 48 Laws was published in 1998, one of the most-cited results in behavioural science: that the "hot hand" in basketball is an illusion produced by misreading randomness.
That result has since been substantially revised. Work in the mid-2010s identified a selection bias in the original analytical method which, once corrected, removes much of the original evidence for the fallacy and leaves some support for streak effects being real in certain settings. The debunking was itself partly debunked.
[CONTESTED] — this does not vindicate momentum-belief in any strong form, and it says nothing about business or war, where the sequences are neither independent nor short. But it means the confident "greatly overrated" is a stronger claim than the evidence supported then or supports now, and the vault should not pass it on as established.
The argument that survives regardless is the functional one above: whatever the truth about streaks, believing in one supplies a licence to stop examining your method, and that cost is paid whether or not the streak is real.
If momentum is not a force, what is a run of successes evidence of?
Greene's answer is: your own doing. But a run has genuine informational content that his framing discards. It is evidence about your method's fit to current conditions, about the conditions themselves persisting, about your reputation shifting in ways that make the next thing easier, and — since a rising reputation genuinely does reduce resistance — about a mechanism by which past success causally assists future success without anything mystical being involved.
That last one matters. There is a real pathway from success to success, and it runs through other people's beliefs rather than through physics. Doors open, calls get returned, the benefit of the doubt is extended. Someone on a visible run is operating in a materially easier environment, and treating that as your own doing is precisely the attribution error the same chapter warns about four paragraphs earlier.
So the correct position is narrower than either of Greene's. The run is not carrying you, and it is not nothing: it is a temporary improvement in your operating conditions, granted by observers, which will be withdrawn when the run visibly ends. Which is an argument for using it quickly rather than for believing in it.
Three or four things have gone your way and the sequence has started to feel like a property of the situation.
Test which momentum you are actually holding. Ask what would be lost if you did the next thing differently — different approach, different target, same energy. If the honest answer is nothing, I'd just be doing something else, you have tempo, and tempo is fine. If the answer has a sense of interruption in it, of breaking something, of the run being spent — that is the force belief, and it is doing no work for you that a clear reason could not do better.
Then make yourself name the cause. Not it's going well — the specific reason each of the last three worked. If you can produce three distinct reasons, you have a method and you can decide about it. If they collapse into one word, or into we're on a roll, you are relying on the thing Greene says to leave for people with nothing better.
Separately, check which direction your caution is pointing. Is it rejecting specific reckless moves, or is it reducing how many moves you make? The second one is the over-cautious failure and it does not feel like fear; it feels like stewardship.
And use the easy conditions while you have them. The doors are open because people have noticed, and that is real, temporary, and not yours. Spend it on the thing that would have been hard to arrange last year.
Evidence quality: [POPULAR SOURCE] throughout, no citations of any kind in the Reversal. The momentum denial is [CONTESTED] — see the replication note above; the empirical picture on streak effects was revised after publication and in the direction Greene did not anticipate.
Tension (unresolved, preserved): "recognizing the role of luck and circumstance in your success"3 versus "you create your own successes, and if they follow one upon the other, it is your own doing."2 Same chapter, one page apart, incompatible as accounts of causation. Each is deployed against a different reader error. Greene does not acknowledge the collision.
Tension (resolvable, unmarked): never lose momentum versus momentum is greatly overrated, resolved above as tempo versus force. Greene switches senses mid-paragraph without signalling it.
Open questions: if a visible run genuinely improves your operating conditions through other people's beliefs, then something real is carrying you — just not the thing momentum-believers think. How should that be scored? Neither Greene's position accommodates it.
The Reversal is where Greene habitually undercuts his own chapter, and this is an aggressive instance: the law says stop after victory, and the Reversal's first half warns against stopping too much. That is a useful corrective and it also means the chapter's actual instruction is a band rather than a rule, with neither edge specified.
Across the corpus the momentum question is unstable. Mastery is built on accumulation — years of deliberate practice producing capabilities that compound, which is a momentum claim in everything but name. The 33 Strategies of War treats tempo and the disruption of an enemy's rhythm as central, which requires momentum to be a real property of an opponent's position. Here it is dismissed as a crutch. The three books are not describing the same phenomenon and none of them defines the term.
The convergence is with Greene's consistent hostility to any explanation that removes agency from the operator. Luck, momentum, fate, the market, the times — the corpus reliably treats these as excuses, and reliably makes an exception when it needs the reader to feel unsafe. Law 47 needs the reader to feel unsafe about luck and self-reliant about streaks, and takes both positions within a page.
Record When Hot: Studying Hot Streaks — the working practice takes the opposite position and is more useful for it. Practitioners who track their own output overwhelmingly report clustering, and their response is operational rather than metaphysical: when it is going well, extend the session, defer the administrative work, capture more than you need. That does not require momentum to be a force. It requires only that whatever is producing the run — fit between method and conditions, a configuration of attention, an unfinished thought still warm — is temporary and not reproducible on demand. The insight the pairing produces, which neither reaches alone: Greene's dismissal and the practitioner's behaviour are compatible, because they are answering different questions. He is asking will this continue, and the answer is you have no idea. They are asking is now different from later, and the answer is demonstrably yes. Which yields a rule the chapter cannot state: treat a run as information about the present rather than as a prediction about the future. Harvest it, do not extrapolate from it. That is exactly the discipline the streak-believer lacks, and exactly the opportunity the streak-denier forfeits.
Flow State Intensity — flow is the closest psychological correlate of felt momentum and it settles the ontology in a way that neither of Greene's sentences manages. The state is real, measurable, and consequential for performance; and it is a property of the person's engagement with a task, not of the sequence of outcomes. That distinction is the whole disagreement. The streak-believer feels something genuine and misattributes it to the run; Greene, correctly denying the run has force, denies the felt thing along with it and therefore has nothing to say about a state that materially affects output. What the pairing produces: momentum-as-experienced is a real internal condition with a false external theory attached, which explains why the belief is so resistant to argument — the evidence for it is proprioceptive, and telling someone their streak is statistical noise does not touch what they are actually reporting. It also explains the specific danger Greene identifies. Flow narrows attention onto the task and away from the frame, which is precisely why "less prone to act strategically, and more apt to repeat the same methods" follows: the state that makes execution excellent is the state that makes the question should I be doing this at all unavailable.
The transferable claim is the harvest rule from the first handshake: treat a run as information about present conditions rather than as a forecast, which licenses intensified use without licensing extrapolation. That is checkable in any domain with recorded output and it predicts a specific error in both directions — believers over-extend, deniers under-harvest.
The open edge is the reputational pathway raised above. If a visible run materially eases your operating conditions through observers' beliefs, then past success does causally assist future success, and neither Greene's position accounts for it. How to score that without sliding back into streak-thinking is unresolved.