In 1878 Jay Gould set up a company to compete with Western Union's telegraph monopoly.
The directors of Western Union did the sensible thing: they bought him out. It cost a hefty sum, but they'd removed an irritating competitor.1
A few months later Gould was back, complaining he'd been treated unfairly, starting a second company to compete with Western Union and its new acquisition.
They bought him out again.
Then the pattern began a third time — and this time Gould "went for the jugular." He staged a bloody takeover struggle and took complete control of Western Union.1
Two purchases had taught the directors what Jay Gould was: a nuisance who could be paid to go away. That lesson was the weapon.
Greene's formulation:
Another effective smoke screen is the pattern, the establishment of a series of actions that seduce the victim into believing you will continue in the same way. The pattern plays on the psychology of anticipation: Our behavior conforms to patterns, or so we like to think.2
And the closing line, which is the sharpest sentence in the passage:
The pattern is powerful in that it deceives the other person into expecting the opposite of what you are really doing.1
The opposite — not merely something different. That's the distinguishing feature of this technique, and it's what separates it from a plain surprise.
Every other technique in Law 3 asks the target to believe something. This one asks them to conclude something.
A decoy is offered. A cover story is told. Both are supplied by the deceiver, which means both can in principle be doubted — the target knows, at some level, that they received information from an interested party.
A pattern is different. Gould never told Western Union's directors that he was a nuisance who could be bought off. They worked it out themselves, from evidence, twice, and they were right both times.
That's the elegance. The belief isn't installed; it's derived. And people defend conclusions they reached themselves far more stubbornly than claims they were handed, because the conclusion feels like their own competence rather than someone else's testimony.
By the third approach, the directors weren't being credulous. They were being experienced, which is much harder to argue anyone out of.
The costly part of this technique is easy to skip past, so it's worth pricing.
Gould's first two companies were real. He had to found them, fund them, and make them genuinely threatening enough that buying him out was the rational response. Western Union paid "a hefty sum" each time — Gould was collecting, not spending.
But that's the wrong ledger. The correct one is: what did the first two rounds cost him in time and risk, and what did they buy?
They bought a prior. After two identical episodes, the directors held a well-supported model — Gould competes, Gould gets bought, Gould goes away — and that model determined their response to the third approach before they'd examined it.
Which is why the third move worked without needing to be hidden. Gould didn't conceal the takeover attempt. He didn't have to. The directors saw a third competing company and filed it under the existing pattern, and by the time the filing turned out to be wrong the company was gone.
Note also what the pattern did to their urgency. A first-time threat gets full attention and a fast, careful response. A third-time threat gets the standard treatment: route it to the people who handled the last two, resolve it the way it was resolved before. The pattern doesn't just produce a wrong expectation — it produces a procedure, and procedures run without anyone senior looking.
Gould's real product wasn't a false belief. It was a routine.
This technique has an unusual property: it is expensive and single-use against a given target.
Two funded companies, months of work, and a reputation for being bought off — spent to take one company once. Western Union's directors will not fall for it again, and neither will anyone who watched.
Which explains why the pattern deception is rare in a way blandness isn't. A bland exterior can run for a lifetime against everyone simultaneously. A pattern must be built against one target, over time, and it's destroyed by its own success.
Greene doesn't mark this, and it matters for whether the technique is available to anyone reading. It requires: a target that persists, resources to fund the false iterations, and enough time for the prior to set. Most people have none of the three.
Someone has approached you three times with roughly the same thing and you've handled it the same way three times.
That is the condition, and the tell is not in what they're doing — it's in what you are doing.
Check for the routine. Has this been delegated? Is there a standard response? Did you personally read the terms this time, or did you recognise the situation and apply the previous answer?
If a recurring approach has stopped reaching your desk, the pattern has done its work regardless of whether anyone intended it. Every established procedure is a place where judgement has been replaced by recognition.
So the practical rule: on the third instance of anything, look at it as though it were the first. Not more suspiciously — that's not the failure. The failure is that you stopped looking at all, because you already knew. Read the actual terms. Ask what's different this time. It costs twenty minutes and the pattern technique cannot survive it.
And a second check, for the other direction. When you catch yourself thinking this is just X again, notice that the sentence contains a prediction. Say what you're predicting out loud: I expect this ends with us buying them out for a few million. Stated as a prediction, it becomes something you can test. Left as recognition, it's invisible.
Worth separating, because the technique's premise is a claim about people generally.
"Our behavior conforms to patterns, or so we like to think"2 — Greene's hedge in that clause is doing quiet work. The technique doesn't require behaviour to be patterned. It requires the observer to believe it is.
And they mostly do, because it's true mostly. Extrapolating from repeated behaviour is a good heuristic that works nearly all the time, which is exactly why it's exploitable. You cannot fix this by abandoning the heuristic; you'd be wrong far more often.
Greene offers a technique and no defence. The honest position is that the defence is expensive and partial: you can only afford to re-examine a few recurring situations properly, so pick the ones where being wrong would be worst.
Strongest support. A dated, named commercial case with three iterations and a clear outcome — and unusually, the mechanism is demonstrated rather than asserted, since the first two rounds are visible as setup rather than inferred.
Tension — the case may be simpler than the technique. Gould could plausibly have intended to be bought out the first two times and only reached for the takeover when the opportunity appeared. Greene presents three moves as one designed sequence. Nothing in the account establishes premeditation, and a man who repeatedly extracted payments and then took the company is fully explained without a plan.
Tension — against the smoke-screen rule. Greene files this under Part II, whose stated principle is the bland and unremarkable front. Gould's method is the opposite: highly visible, aggressive, expensive, and memorable. It works through repetition rather than through being unnoticed.
🚩 SINGLE SOURCE · 🚩 SECONDARY WITHOUT PRIMARY — no citation; the 1878 date is the only specific, and the directors' reasoning is narrated from outside. [POPULAR SOURCE].
Open questions.
Greene's Expectation Assimilation, from the vault's behavioural corpus, holds that people absorb new input into whatever frame they already carry — the incoming fact gets bent to fit the model rather than the model bending to fit the fact.
That is the pattern technique's engine, described from the target's side, and the pairing sharpens both.
Greene's account treats the directors as predicting — expecting Gould to behave as before. Expectation assimilation says something worse: they weren't predicting and then being surprised. They were perceiving the third approach as an instance of the first two, which means the differences were being filed away as noise in real time.
Which explains the otherwise puzzling part of the case. The third move was a bloody takeover struggle — not a subtle thing. It should have been visible. Under a prediction model, the directors made a bad forecast and got caught out. Under assimilation, they were looking directly at a takeover and processing it as another buyout negotiation until it was finished.
The pattern doesn't produce a wrong expectation about the future. It produces a wrong perception of the present, which is why the defence has to be procedural rather than attentive.
→ psychology · Habituation and Deprivation
The psychology page holds that a repeated, unchanging stimulus stops being processed — attention is spent on change, and the steady state gets written off.
Gould's first two companies were a habituation protocol, and reading them that way changes what the technique is.
The obvious reading is that Gould built a belief. The habituation reading is that he built a tolerance: each identical approach reduced the alerting value of the next one, until a competing telegraph company was no longer a stimulus at Western Union at all. It was weather.
And that predicts something the belief model doesn't — that the effect should scale with repetition rather than with plausibility. Gould didn't need his third company to be more convincing than the first two. He needed it to be the same, which is much easier and much cheaper.
The insight neither field yields alone: the pattern technique's active ingredient is sameness, not credibility. Which means the counter is not better scepticism — you cannot be sceptical of something you have stopped registering — but deliberate re-alerting: a rule that fires on the count rather than on the content. Third occurrence, escalate regardless. A trigger that doesn't depend on anyone noticing anything is the only kind that survives habituation, and it's exactly the kind of dull procedural control that organisations remove when nothing has gone wrong for a while.
→ history · Exploiting Enemy Expectations
The Chinese strategic material treats the enemy's expectations as terrain — something you survey, then shape, then move through. The value of a manoeuvre is measured against what the opponent has been led to anticipate.
Gould fits that frame and adds a variable it doesn't carry: cost of construction.
The strategic tradition mostly assumes expectations can be shaped cheaply — a feint, a false camp, a rumour. Gould's expectation cost him two real companies, funded and operated, over multiple years. That is an enormous capital investment in a single opponent's model of him.
Which produces a distinction worth having, and it applies well beyond either source: expectations can be shaped by assertion (cheap, fast, deniable, weak) or by track record (expensive, slow, undeniable, and nearly impossible to dislodge). Military feints are the first kind. Gould's was the second.
And the second is the only one that survives a competent opponent. Western Union's directors were not fools and would have discounted anything Gould merely claimed. What they could not discount was two completed transactions in which he had behaved exactly as the model said. You cannot be sceptical of a pattern you personally generated the evidence for — and that's the real reason this technique beats every other deception in Law 3 against a sophisticated target.
Sharpest implication. The directors were never deceived. Twice they observed Gould accurately, reasoned correctly, and acted sensibly — and the accumulated correctness is the whole weapon. This technique doesn't defeat judgement, it converts judgement into procedure, and a procedure runs without anyone senior in the room. The most dangerous position to be in is not ignorance about someone; it's being reliably right about them until the moment they change.
Generative questions.