The Mongols did something before their campaigns that looks, on a casualty list, like a wasted battle.
They would fight an opening engagement whose only purpose was to test the opponent's strength and resolve. They would never deal with an enemy until they had gauged his morale.1
Men died in it. Nothing was taken. And it was one of the most efficient things they did, because it also revealed something of the enemy's strategy and thought — a set-up battle that returned a reading no scout could produce.1
Nietzsche states the principle in one line, and it is the cleanest sentence in the chapter:
One attacks someone not only so as to harm him or to overpower him but perhaps only so as to learn how strong he is.1
Greene's argument for why this is necessary rests on a specific gap in what observation gives you.
You can watch someone for a long time and still not know how much fight they have in them. And without that number you will get them wrong in a predictable direction — over- or underestimating them according to your own levels of fear and confidence.1
The stakes of getting it wrong are asymmetric:
Someone hiding his cowardice and lack of resolve can be made to surrender with a single violent push; someone desperate who has little to lose will fight to the bitter end.1
Those two people can look identical across a table. They diverge only under pressure, which means the only instrument that reads them is pressure.
The interpersonal form appears a few lines earlier in the same chapter and is the same instrument at lower cost.
People are easiest to read in moments of crisis, when they either reveal a weakness or work so hard to disguise it that the effort shows.1 The problem is that crises arrive on their own schedule.
So you make one.
You can actively probe them by doing things that seem harmless but get a response — maybe say something bold or provocative, then see how they react.1
Making people emotional, pushing their buttons, touches some deep part of their nature. And then one of two useful things happens: they let slip something true, or they put on a mask that you — in the laboratory situation you have created — can see the edges of.1
That phrase is the whole idea. The probe is not an attack. It is an experiment with a controlled input.
The chapter's cases are all about converting one measurement into a whole strategy.
Hannibal. Told by his intelligence that the Roman general opposite him was arrogant and hot-tempered, he would deliberately play weak — luring the man into a rash attack.1 One temperament reading; an entire battle plan.
Churchill. Once he saw that Hitler had a paranoid streak and became irrational at the merest hint of vulnerability, he knew how to unhinge him: feign an attack on some marginal area like the Balkans, make him see threats on all sides, and get him to spread his defences thin — a critical military mistake.1
Lee Atwater, 1988. Working for the elder George Bush against Senator Robert Dole, Atwater discovered that Dole had a terrible temper his aides struggled to control. He devised endless stratagems to push the buttons.1
Greene names the double payoff there, and it's worth separating:
Not only did an upset Dole look unpresidential to the American public, but an emotional and angry man rarely thinks straight.1
One effect is on the audience. The other is on the opponent's own cognition. The same provocation buys both.
A disturbed mind is one you can control and unbalance at will.1
Two costs, actually, and the chapter mentions neither.
The Mongol set-up battle killed people to acquire information. Greene presents it as economical — and by the logic of the campaign it was, because the reading prevented larger errors later. But the economy is only visible from the top of the organisation. Nobody in that opening engagement was told what it was for.
The interpersonal version has a subtler price. A probe is indistinguishable from a provocation to the person receiving it, and it is remembered as one. Atwater's stratagems worked and also made Dole a lasting enemy. You are buying a measurement with a relationship, and the chapter never puts the two on the same ledger.
The move that reduces the price is in the phrasing Greene almost gives: things that seem harmless. A probe whose deniability is intact costs nothing. A probe that reads as an attack has bought its information at full price and started a war for the same money.
You are about to commit to something against someone whose resolve you can't read — a negotiation where you don't know if they'll walk, a competitor you don't know will retaliate, a stakeholder you don't know will actually block you.
Do not build the plan on your estimate. Buy a reading first, and make it small.
The shape is: a request slightly beyond what's reasonable, made casually, early, in a low-stakes setting. A deadline floated that's tighter than it needs to be. A position stated a notch stronger than you'll hold. An ask attached to something else so that refusing it costs them nothing.
Then watch three things, in this order.
Speed. How fast the resistance comes. Instant firmness is usually a rehearsed position; a pause before firmness is usually a real one.
Where it goes. Do they push back on the substance, or on you for asking? The second tells you what they're actually protecting.
Recovery. How long until they're normal again. This is the reading Atwater bought — a man whose composure takes a while to come back has a lever with a long handle.
Then keep the deniability. The probe must be cheap to abandon: "fair enough, that was probably ambitious." If you cannot back out of it in one sentence, you have not run an experiment, you have opened a front.
And the discipline that makes this honest rather than merely effective: decide in advance what result would change your plan. A probe you would ignore either way is not intelligence-gathering. It's just poking someone.
The Mongol reconnaissance-in-force is well attested as a real practice, and the Atwater material is close to consensus in accounts of the 1988 primary. The Churchill–Hitler Balkans deception is real, though Greene's causal claim — that it worked because Churchill had diagnosed paranoia — is his own reading. [POPULAR SOURCE]
The technique has a circularity the chapter shares with the masked opposite. You probe to test a hypothesis about someone. But the probe also creates the state you then measure: a person made angry by a deliberate provocation is displaying a response to provocation, not a stable trait. Atwater's Dole looked like a man with an uncontrollable temper. He was also a man being systematically targeted by an opposing campaign, which would test anyone's composure. The chapter reads the result as a discovery about Dole. It is at least partly a discovery about what Atwater did.
And the interpersonal version has no stopping rule. Greene's advice is to push buttons until something gives. He does not say how to distinguish a probe that has taken a measurement from one that has done damage, and the difference is only visible afterwards. The deniability constraint above is this page's addition.
This sits in direct tension with Strategy 9: The Counterattack Strategy, and the two chapters are four apart.
Strategy 9's core instruction is to let them move first — patience as the master skill, Ted Williams waiting for his pitch, Napoleon feigning weakness at Austerlitz and letting the czar come. The information arrives because you did not act.
This chapter says make them move so you can read the movement.
Both are ways of purchasing information with tempo, and they are billed to different accounts. Waiting costs initiative and pays nothing else. Probing costs deniability and, sometimes, the relationship. The book presents both as mastery and never says which conditions select for which.
The reconciliation available: wait when the other side has to act eventually, probe when they don't. Napoleon could wait at Austerlitz because the allied army had come to fight and could not simply sit. Churchill could not wait for Hitler to reveal his threshold for paranoia, because nothing was forcing the disclosure. The variable is whether time is applying pressure to them or only to you.
There is also a clean convergence with Strategy 10: Deterrence Strategies — the deterrence chapter's whole apparatus works by making an opponent's estimate of your resolve too expensive to test. Deterrence is the countermeasure to this page. Robert the Bruce's tit-for-tat is a policy designed so that every probe returns the same costly answer, and stops being worth running.
Psychology → Stress as a Measurement Instrument, and What It Distorts. The chapter's premise — that character is revealed under pressure and not otherwise — matches a long-standing assumption in selection and assessment: interviews, stress tests, and crucible experiences are all built on the idea that the real person surfaces when load is applied.
The insight the pairing produces: pressure does not reveal a hidden stable trait so much as reveal what the person's resources are currently able to cover, and resources vary. The same man is resolute rested and brittle exhausted; the same negotiator holds firm with an alternative and folds without one. Which means every probe returns a reading valid for a specific moment and set of circumstances, and treating it as a fixed measurement of character is the error. Hannibal's arrogant Roman was arrogant in front of his own army in a political culture that punished caution. Atwater's Dole was tired, mid-primary, and being worked on. The reading is real; the noun it gets attached to — "a man with a temper" — is the fabrication, and it is the fabrication that gets carried into the next campaign.
Business → Sales Qualification as a Probe Programme. Commercial practice has independently arrived at this: the deliberately mistimed ask, the price floated high to observe the reaction, the artificial deadline, the request for a commitment nobody needs yet. All of it is reconnaissance-in-force with a friendlier vocabulary, and it is measuring the same thing — resolve, and where the real authority sits.
The insight neither domain gives alone: the probe reveals the organisation, not the individual, and that is its highest-value output. When you float something slightly unreasonable, what comes back is not one person's resistance but the shape of who they had to consult, how long it took, and which objection they were willing to own versus attribute upward. Greene reads probes for individual temperament because he is writing about commanders. In any structure larger than two people, the more useful signal is the routing — the delay before the answer is a map of the decision chain, and it is nearly impossible to acquire any other way.
Some facts about people cannot be observed, only elicited — resolve is one of them, and there is no amount of watching that substitutes. Which means every serious plan has a piece of information in it that you can only buy by acting first and slightly wrongly, on purpose, in a way you can walk back. The people who look uncannily well-informed about their opponents are usually just the ones willing to pay that small price early rather than a large one later.