Filippo Maria was the last of the Visconti dukes of Milan, and Greene describes not an incident but a habit: he "consciously did the opposite of what everyone expected of him."1
Follow one courtier through it, because the sequence is a machine.
The duke suddenly showers the man with attention. Naturally, the courtier comes to expect a promotion. So the duke starts treating him with the utmost disdain.
Confused, the man decides to leave court. The duke recalls him and starts treating him well again.
Doubly confused, the courtier reasons it out: perhaps his expectation of promotion had become obvious, and offensive. So he begins behaving as though he expects no honour at all. The duke rebukes him for his lack of ambition and sends him away.2
Expect promotion — punished. Stop expecting promotion — punished. Leave — recalled. Stay — dismissed.
There is no posture available that is not an error.
Greene states the solution the court arrived at, and it is worth reading as a job description rather than as advice:
"The secret of dealing with Filippo was simple: Do not presume to know what he wants. Do not try to guess what will please him. Never inject your will; just surrender to his will. Then wait to see what happens."3
Read the four instructions as a list of things a person is being asked to stop doing.
Do not model your superior. Do not anticipate. Do not form intentions. Wait.
That is not obedience — obedience still requires knowing what is wanted. It is the removal of independent judgement as such, and it is a much stranger thing to have installed in a court than fear would be.
And the result: "Amidst the confusion and uncertainty he created, the duke ruled supreme, unchallenged and at peace."4
An arbitrary tyrant who punished at random would produce fear and flight. Visconti produces neither — the courtier keeps trying, keeps reasoning, keeps adjusting. He is not fleeing; he is working.
The difference is that the reversals are triggered by the courtier himself. Each reversal comes precisely when the courtier has formed a stable expectation, which means the man's own conclusions are the trigger. He is not being punished for what he does. He is being punished for having settled on a model.
That is what makes it inescapable by effort. A person punished for actions can change actions. A person punished for having reached a conclusion has only one remedy — stop concluding — and stopping concluding is the thing that makes them useless to themselves.
The duke is not attacking the courtier's behaviour. He is attacking the courtier's capacity to predict, and once that is gone the man has no basis for any independent move.
Greene's assessment is that Visconti "ruled supreme, unchallenged and at peace," and files it as a success. Take the success at face value and then ask what it cost, because the chapter has no line for it.
A court operating under the four instructions above is a court of people who do not anticipate, do not form intentions, and wait. Every quality that makes a subordinate useful is on that list. An adviser who does not presume to know what the ruler wants cannot prepare anything. A minister who never injects his will cannot administer.
So the duke bought security by disabling the faculty he would otherwise have been able to delegate to — and what he has at the end is peace, and a court that can only wait.
That may be an entirely rational trade for a fifteenth-century Milanese duke, whose principal occupational risk was assassination by the people nearest him. Worth saying plainly rather than moralising: the technique works, and what it purchases is the absence of threat, and what it spends is the capacity of everyone around you to act.
Greene's success criterion here is unchallenged, and against that criterion the trade is free. Against any criterion involving what the court actually produced, it is not, and the chapter never proposes one.
The uncomfortable part is that the machine does not require malice, or even intent.
Every element Greene describes — attention followed by coldness, a recall after a withdrawal, a rebuke for the posture that was rewarded last month — is producible by an ordinary person who is simply inconsistent: whose mood, priorities or attention vary, and who does not notice that the people below them are attempting to model those variations.
The courtier's experience is identical either way. He cannot distinguish a duke who is oscillating deliberately from one who is merely erratic, because from below both present as the rule keeps changing. And the four instructions are equally the correct survival strategy in both cases.
Which means the mechanism's output — a room full of people who have stopped forming independent judgements — is available to anyone inconsistent enough, at no cost and with no plan. The Visconti case reads as a portrait of calculated cruelty. It is at least as useful as a description of what unmanaged inconsistency does to the people beneath it.
You keep getting it wrong with someone above you, and you cannot work out the rule.
Notice the specific texture of it, because it is diagnostic. It is not that you fail — it is that the thing rewarded last time is the thing criticised this time. You showed initiative and were told to check first. You checked first and were told to use your judgement. Each individual instance is arguable; the pattern is not, and naming the pattern sounds like an excuse.
Your instinct is to model harder. Build a better theory of what they want. That is the courtier's move and it is the one the mechanism feeds on — every model you build is more invested effort waiting to be invalidated, and there is no correct model to reach.
So do the thing that removes the ambiguity rather than resolving it. Get the standard written down once, in neutral terms, for a specific piece of work. What does good look like here, and who decides. Not as a challenge — as a request for a spec. If a standard emerges, you now have something stable to work against and the oscillation loses its purchase, because your performance is measured against a document rather than against a mood.
If no standard emerges after a genuine attempt, you have learned the important thing, and it is not about you: the unreadability is load-bearing. It is how the position works. No amount of accuracy on your side will resolve it, and the correct response is a decision about exposure rather than about effort.
And run it the other way, because this is the half that matters more. If people around you hesitate, check rather than decide, and bring you things half-finished — ask what they would have to believe about you to behave that way. You will probably not find a deliberate strategy. You may find that you have been inconsistent without noticing, and have accidentally bought yourself an unchallenged and unproductive court.
The strongest feature is the completeness of the sequence. Greene gives all four branches — expect, stop expecting, leave, stay — and the punishment for each. That structure is what makes this a mechanism rather than a story about a difficult man.
Tension: the chapter counts an unchallenged court as an unqualified success. The four survival instructions describe a court that cannot anticipate, cannot form intentions, and can only wait, which is also a court that cannot do anything. Greene's success criterion is unchallenged and no other criterion is proposed.
Tension: the mechanism does not require intent, and the chapter's framing assumes it. Ordinary inconsistency produces an identical experience from below and identical adaptations. Nothing in the passage distinguishes calculated oscillation from erraticism, and the courtier could not tell either.
🚩 SINGLE SOURCE · 🚩 SECONDARY WITHOUT PRIMARY — no source is named. The behaviour is reported as habitual practice ("he might suddenly shower a courtier with attention") rather than as a documented episode, and the courtier's internal reasoning — his confusion, his inference about having been obvious, his adjustment — is narrated as fact about an unnamed person. This is a reconstruction, not a record. [POPULAR SOURCE] · [LOW CONFIDENCE] on the specific sequence.
Open questions. Filippo Maria was, by other accounts, an intensely fearful man who is said to have been highly secretive about his own movements. If the oscillation was anxiety rather than strategy, does the mechanism care — and does the chapter's reading survive the possibility that its purest example of calculated technique was a symptom? And: what happened to Milan's administration during a reign conducted this way?
Within Law 17, Visconti is the chapter's purest mechanism and its least famous case, which is itself worth noting — Reykjavik gets the pages and turns on a different variable entirely (a stakes asymmetry, named by Spassky), while the Visconti passage actually demonstrates what the law claims.
Against the Reversal, the pairing exposes the chapter's unresolved position on what predictability is. The body text calls it animality — the condition of creatures that can be hunted. The Reversal calls it an asset: a comfortable front you build deliberately in order to spend it later. Visconti has no pattern to spend, which means his version of the technique forgoes the Reversal's entire benefit and the chapter does not notice the trade.
The sharpest structural convergence is with Law 11. Seek the Weak Master and the whole dependency chapter are written for the subordinate — how to become indispensable to a ruler. Visconti is the countermeasure, and it is filed six laws away with no cross-reference: a master who oscillates makes indispensability unattainable, because the courtier cannot determine what he would need to be indispensable at. This build logged at Law 11 that the Machiavelli epigraph was the law's undeclared countermeasure run from above. Visconti is the second instance of the same gap — a book written for the ambitious subordinate keeps printing, without comment, the moves a competent master uses to defeat them.
Behavioral mechanics — Intermittent Reinforcement and Control Dynamics
Intermittent reinforcement's finding is that unpredictable reward produces stronger, more persistent attachment than consistent reward — the behaviour becomes far more resistant to extinction precisely because the subject cannot tell a pause from an ending.
Visconti is that schedule administered by a person, and the pair explains something the chapter reports without accounting for: why the courtier does not simply leave.
Greene's framing is terror, and terror predicts flight. It does not predict a man who is recalled, comes back, reasons carefully, and adjusts again. Intermittent reinforcement does predict exactly that. The warm periods are real and unpredictable, which makes them impossible to write off, and the cold periods cannot be distinguished from an interval before the next warm one.
What the pair produces: the courtier is not being frightened into staying, he is being conditioned into it — and the two feel completely different from inside. Fear is aversive and the person knows they want to go. This is not aversive; the courtier experiences hope, repeatedly, and each cycle deepens the investment. Which is why the mechanism can hold someone in place for years without their ever concluding they are trapped, and why "just surrender to his will" reads to the person doing it as patience rather than as capture.
Business — Consistent, Persistent, Predictable Actions — The Trust Builder
That page's rule is that trust is built by being consistent, persistent and predictable, and it names the failure mode precisely: operators who alternate between two stances communicate that they are performing one of them, and "unpredictable operators destroy trust faster than they build it."
Visconti is that failure mode run deliberately, at scale, and successfully.
Together they identify what the sales page's rule is actually optimising for, and it is not trust in the abstract. Predictability lets the other party act on your behalf without checking — that is the whole commercial value of it. Unpredictability produces the opposite: a counterparty who checks everything, initiates nothing, and waits.
Which means the two pages are not in conflict. They are answering the question what do you want from this person in opposite directions, and the answer determines the technique. A duke who needs no one to act on his behalf can afford Visconti's method; an operator whose revenue depends on other people moving without supervision cannot. The chapter presents unpredictability as a general instrument of power, and its own case shows it is specifically the instrument of someone whose objective is that nothing happens.
Sharpest implication. Visconti is not punishing behaviour — he is punishing the act of having reached a conclusion, which leaves only one remedy, and the remedy disables the person applying it. The technique's product is not fear but the surrender of independent judgement, and a ruler who has bought that has bought a court that can only wait.
Generative questions.