Which, in this matchup, favours Foreman — he has a devastating punch, and sooner or later Ali has to come to him.
So Ali announces, in press conferences before the fight, that he is going to change his style and punch it out with Foreman.
"No one, least of all Foreman, believed this for a second. That plan would be suicide on Ali's part; he was playing the comedian, as usual."2
Then, before the fight, Ali's trainer loosens the ropes around the ring — "something a trainer would do if his boxer were intending to slug it out." Nobody believed that either. "It had to be a setup."3
And then Ali did exactly what he had said he would do. He went right up to Foreman and slugged it out, upset his opponent's entire strategy, and Foreman wore himself out throwing wild punches and taking counterpunches until Ali landed a right cross that knocked him out.4
Greene's closing line is the finding: "Foreman walked into a trap—the trap he had been told to expect."5
What makes this case unusual is not that a deception worked. It is that the truth was delivered twice, through two independent channels, and neither got through.
The verbal announcement could be dismissed as showmanship — Ali talked constantly, and most of it was theatre. Fine.
But the loosened ropes are a different kind of signal entirely. They are a physical preparation, made by a trainer, at cost, with a specific technical meaning that everyone in boxing understands. That is not a press conference. It is the sort of evidence a professional is supposed to read.
And it was discarded on a stronger ground than the announcement: it had to be a setup. The second warning was not merely disbelieved — it was actively re-interpreted as confirmation of the opposite.
That is the mechanism, and it is much worse than ordinary deception.
Greene names it flatly: "The habit of assuming that a person's behavior will fit its previous patterns is so strong that not even Ali's announcement of a strategy change was enough to upset it."6
Work out what that requires.
Foreman had a model: Ali dances. Ten years of evidence behind it. When contradicting information arrived, the model did not merely resist — it consumed the contradiction and grew stronger, because a strong prior converts disconfirming evidence into evidence of deception.
Ali says he'll slug it out → he is lying, therefore he will dance. The ropes are loosened → an elaborate feint, therefore he will dance.
Each new piece of contrary evidence increased Foreman's confidence in the original model, because each one was interpreted as an attempt to move him off it.
A prior that can absorb its own disconfirmation is not a belief. It is a trap that the holder maintains. And this is the same structure logged at Law 12: Gould saw all the clues about Gordon-Gordon and processed each as an anomaly inside an established classification. "There had been many clues before then… it took the loss of millions." Two laws apart, same mechanism, uncross-referenced.
The strategic content here is genuinely counterintuitive and worth stating precisely.
Ali had two ways to conceal his plan.
Conceal it. Say nothing, train in secret, hope nothing leaks. This is expensive — camps leak, sparring partners talk, and a change that large is hard to hide. And if it does leak, it leaks as credible intelligence, because it was hidden.
Announce it. Free, requires no secrecy, and cannot be leaked because it is already public. And the announcement is inoculating — once Foreman has considered and rejected the true plan, he is more resistant to it than if he had never heard it, because he has now actively reasoned his way past it.
The second is not merely cheaper. It is stronger, and the strength comes from the target's own reasoning rather than from the operator's concealment.
There is a condition attached: it only works against someone with a strong prior. A fighter with no fixed model of Ali — a debutant, or someone who had never seen him — would simply have taken the announcement at face value and prepared for a brawl. Ali's ten-year pattern was not the vulnerability he had to overcome. It was the instrument. The very consistency that made him readable is what made the announcement unbelievable.
That inverts the chapter it sits in. Law 17 says predictability is a liability to be scrambled. Ali's predictability was the asset that made the deception free.
Greene files this as Law 17's Reversal, and the Reversal's stated content is that predictability has uses: "By creating a pattern for people to be familiar and comfortable with, you can lull them to sleep… it allows you on rare occasions to do something completely against the pattern, unsettling your opponent so deeply he will fall to the ground without being pushed."7
That is correct and it undoes the body text more thoroughly than Greene acknowledges.
The chapter's main argument classifies predictability as animality — the condition of creatures that can be hunted. The Reversal says predictability is a stored asset: it accumulates while you behave consistently and is spent, once, when you break it.
Those are not compatible. Under the body text you should scramble continuously; under the Reversal, continuous scrambling means you never accumulate anything to spend. Ali is a ten-year deposit and a single withdrawal, and Greene has no way to say which strategy the chapter recommends.
You are about to do something the other side would guard against if they believed you.
The instinct is secrecy. Keep it quiet, prepare quietly, spring it.
Consider the opposite, and consider it seriously rather than as a paradox. Say what you are going to do, plainly, in the room, on the record.
Two things have to be true for this to be the stronger move.
They need a strong prior about you. A track record, a reputation, ten years of behaving one way. Without that, the announcement is just information and they will act on it. The asset being spent is your own consistency, and if you have not accumulated any, there is nothing to spend.
And the announcement has to be costless if believed. Ali would have been fine fighting a Foreman who took him seriously; he simply preferred the alternative. If being believed would be catastrophic, do not test it.
What you get in exchange is worth more than concealment. A secret can leak, and when it leaks it arrives as credible intelligence. A public statement cannot leak, and — this is the part that does the work — once they have reasoned their way past it, they are more committed to the wrong model than before you spoke, because now they have invested their own judgement in dismissing it.
Then run it defensively, which is the half that will actually save you something. When someone tells you plainly what they intend and your reaction is they can't mean that, stop. That reaction is the mechanism. Ask what specifically makes it implausible, and check whether the answer is evidence about this situation or a model built from their past. If it is the second, you are Foreman, and the fact that you can explain why they must be bluffing is not reassurance — it is the symptom.
The strongest feature is the double signal. One warning discarded is ordinary; two independent warnings — one verbal, one physical and costly — discarded, with the second actively converted into confirmation of the opposite, is strong evidence that a prior was doing the work rather than a shortage of information.
Tension: the Reversal contradicts the body text on what predictability is, and the chapter never chooses. Animality in the main argument; a stored asset here. Continuous scrambling and one-shot spending are incompatible strategies.
Tension: this is the chapter's cleanest case and it argues against the law. Ali's ten-year consistency is the reason the manoeuvre worked. Law 17's instruction would have destroyed the asset it depends on.
Tension: it is a one-shot instrument presented without a use limit. Having done this once, Ali's word is now possibly true, and the entire mechanism depended on it being reliably false. Greene gives no indication that the technique consumes itself.
🚩 SINGLE SOURCE · 🚩 SECONDARY WITHOUT PRIMARY — no source is named for the press conferences, the trainer's decision, or Foreman's state of mind. The claim about what everyone believed ("No one, least of all Foreman, believed this for a second") is an assertion about the interior of a man Greene did not interview. The fight itself is among the most heavily documented events in sport; none of this has been checked against that record in this build, and the "rope-a-dope" reading of the bout is contested in the boxing literature. [POPULAR SOURCE] · [CONTESTED] on the tactical account.
Open questions. Did Ali intend the announcement as a deception, or did he simply say what he meant and get lucky in being disbelieved? The two are indistinguishable from outside and imply completely different things about whether this is a technique. And: is the effect symmetric — can a person with a strong prior be walked into a true belief by being told a lie they will dismiss?
The convergence with Law 12 is exact and Greene never draws it. Gould and Lord Gordon-Gordon runs the identical mechanism: an early classification thereafter processes every clue as an anomaly, and "there had been many clues before then… it took the loss of millions for him to see through the scheme." Gould was professionally paranoid and saw everything.
Two laws apart, the book demonstrates twice that a prior survives its own disconfirming evidence, and never states it as a principle. The Law 12 page concluded that vigilance is the wrong defence because it operates downstream of classification. Foreman is that conclusion in a boxing ring — he was not inattentive, he was pre-committed, and his attentiveness was spent converting warnings into confirmations.
Within Law 17, this case sits against Reykjavik as a matched pair on the durability of an opponent's model. Spassky was destroyed by a pattern with no content — he could not read Fischer and could not stop trying. Foreman was destroyed by a pattern with too much content — he read Ali perfectly and could not stop believing it. Between them: an opponent's model of you is neither easily broken by evidence nor easily maintained in its absence, and both failure modes are exploitable by someone who knows which one they are facing.
Psychology — Stereotype and A Priori Judgment
That page's territory is the prior formed in advance of the encounter — a categorisation that arrives before the evidence and then organises how the evidence is read, so that what is perceived is shaped by what was expected.
Foreman is that operating on an individual rather than a group, and the pair produces the sharper version of both.
The usual worry about a priori judgment is that it makes you miss disconfirming evidence — you do not notice what does not fit. Foreman did not miss anything. He received the disconfirming evidence twice, examined it, and assigned it a meaning that strengthened the prior.
That is a second and more robust failure mode: not blindness but absorption. A prior strong enough to interpret its own counter-evidence is unfalsifiable from inside, and the more sophisticated the holder, the better they are at generating the interpretation that saves it — Foreman's it had to be a setup is a genuinely intelligent inference and it is what beat him.
Which means the standard remedy for stereotyping — pay attention to the individual case — is insufficient against the strong form. Attention is what Foreman spent. The only thing that discriminates is noticing that your explanation for the anomaly was generated by you rather than supplied by the situation, and that is a much harder thing to catch.
Business — The Self-Persuasion Principle
That page's mechanism: a conclusion the prospect reaches themselves is held far more firmly than one supplied by the operator, so you engineer the path and let them arrive.
Ali is the adversarial application, and it is the same principle pointed at a false conclusion instead of a true one.
Foreman was not told Ali would dance. He worked it out — from the announcement, from the ropes, from his own knowledge of Ali's career — and he held the conclusion with the conviction that belongs to something you reasoned your way to yourself.
What the pair produces: self-persuasion has no truth-preference, and the operator's honesty is not what determines the outcome. The sales version is usually defended on the grounds that the prospect is reaching a true conclusion under their own steam. Ali's case shows the mechanism runs identically toward a false one, and runs better when the operator tells the truth — because a truth the target dismisses is more inoculating than a lie they detect.
That is genuinely uncomfortable for the sales framing. It means the technique's ethics cannot rest on the operator being truthful, since Ali was truthful and it was the deception, and the target's own reasoning did all the deceiving.
Sharpest implication. Ali told the truth and it functioned as the deception, because a strong prior converts contrary evidence into confirmation. Which means honesty is available as a manoeuvre against anyone who is certain about you — and that the reliability which made him readable was not the vulnerability Law 17 says it is, but the asset that made the whole thing free.
Generative questions.