France, 1848. The presidential election has come down to Louis-Adolphe Thiers, "the man of order," against General Cavaignac, "the rabble-rouser of the right."1
Thiers works out he is hopelessly behind and starts looking for a way through.
His eye lands on a lowly deputy in the parliament — grand-nephew of the great general, and in Greene's phrase "a bit of an imbecile."2 But the name will win, in a country that wants a strong ruler. So: put him up, get him elected, run him, and push him offstage in due course.
The first part of the plan worked to perfection, and Napoleon was elected by a large margin.2
Three years later Louis Bonaparte dissolved parliament, declared himself emperor, and ruled France for eighteen years.2
The context is the part people skip, and it changes what kind of decision this was.
Thiers was not choosing between a safe plan and a clever one. He was losing. Greene's word is "desperately" — he "searched desperately for a solution."1 There is no version of 1848 where Thiers wins as Thiers.
That reframes the whole thing. The puppet plan was not overreach; it was the only move on the board that did not end in Cavaignac. Judged against doing nothing, it was correct, and it delivered exactly what it promised in the timeframe it promised it.
Which is precisely why it belongs in this chapter and not in a chapter about greed. The failure is not that he chose badly among his options. It is that his plan ended at the election.
Look at what Thiers actually selected for.
He needed one thing — votes, at scale, immediately, in a country "yearning for a strong ruler."2 The name Bonaparte supplied that and nothing else could have. It is not an endorsement or a policy or a coalition; it is a piece of stored public feeling attached to a living man, available for use, requiring no persuasion at all.
And he needed a second thing: the man attached to the name had to be removable afterward. His evidence for that was the impression of stupidity.
Two requirements, and he sourced them from two different properties of the same person. The first from the surname, the second from the personality. He treated them as independent.
Here is the structural problem, and it is not a foresight failure in the way Greene says.
The name is not neutral cargo. The reason the name won the election is the reason the name outranked Thiers. A country that will elect a man for being a Bonaparte is a country that will keep obeying him for being a Bonaparte — the popular feeling Thiers borrowed was not lent to him, it was lent to the name, and it stayed with the name. On the day after the election, the deputy had a national mandate and Thiers had a plan.
So the two requirements were never independent. Every unit of electability the name supplied was a unit of removability it destroyed, and the exchange rate was one to one. A less famous Bonaparte would have been easier to discard and would not have won. There is no dial setting that delivers both.
Greene's diagnosis is different and, on the text, weaker:
The problem was that Thiers had not foreseen one simple fact: This "imbecile" was in fact a man of enormous ambition.2
That makes the error a misjudgement of character — Thiers read the man wrong, and with better reading he would have chosen someone else. But there was no one else. The whole plan is the surname, and the surname is what beat him. Ambition made Louis Bonaparte the one who took the opening; the opening was there for whoever held that name.
Greene locates the failure in an unpredictable fact about a person. It was a predictable fact about the instrument.
The chapter gives no account of the interval and it is the part a planner would want.
Three years is a long time. It is long enough for Thiers to have noticed the drift, and long enough for a counter-move if one had been designed. Nothing is recorded because — on the evidence of the plan as described — nothing was designed. The plan had two steps, elect him and eventually push him offstage, and the second step is stated in the passive with no mechanism, no timetable, and no named lever.
Eventually would be pushed offstage. That is not a step. It is a wish located after the part that had detail.
This is the chapter's actual lesson and it is more useful than the one printed. A plan is not long because it has a distant end; it is long because every stage has the same resolution as the first one. Thiers's plan had one stage at high resolution and one at zero, and he experienced it as a two-stage plan.
Read the paragraph again and ask who, on the chapter's own facts, planned all the way to the end.
Thiers's plan: win the election. Resolution drops to zero immediately afterward.
Louis Bonaparte's: accept a nomination from a man who intends to discard him, win on a name, hold the presidency for three years, dissolve the parliament, take the crown, keep it for eighteen. Every stage happened. Nothing about it was improvised into existence at the last moment — a coup against a sitting legislature requires an army's cooperation arranged in advance.
The exemplar of Law 29 in this anecdote is the puppet, and Greene files the episode under Thiers's failure without remarking on it.
That is not a small oversight, because it inverts the reader's position. The chapter is written to a reader who identifies with the planner — you are Thiers, you are Bismarck, you are the one with the design. Here the reader's stand-in loses to a subordinate who was running the longer plan the entire time, and the chapter's framing keeps that subordinate in the role of unforeseen circumstance. He is not circumstance. He is the other strategist, and the only reason he reads as weather is that Greene narrates from Thiers's chair.
The framing is Greene's and is carried here rather than smoothed.
"This Bonaparte seemed a bit of an imbecile."2 The quotation marks around "imbecile" later in the same passage signal that Greene knows it is Thiers's assessment rather than a verdict. But the sentence that introduces the man is in Greene's own narrating voice, and no correction is offered.
What is worth noting is that the underestimation is doing structural work for the chapter. If Louis Bonaparte was in fact a capable operator whom the political class misread — which is the ordinary historical reading — then Thiers's error is not a foresight lapse at all but a straightforward misappraisal, and the case stops illustrating Law 29 and starts illustrating Law 19 or Law 43. The chapter needs him to be stupid, because a stupid man who wins is evidence about the plan, and a clever man who wins is evidence about the man.
You need something and there is exactly one thing that will get it. A person, a partner, a channel, a name — whatever it is, it is the only unit on the board with enough force to move the thing you are stuck on.
You will notice, when you consider it, a small note of unease. Everyone does. And the way it usually gets handled is with a sentence like we'll sort that out later.
Do not write that sentence. Write the next one instead.
Take the property that makes the instrument powerful — the reach, the reputation, the surname — and ask what it does to you after it has done what you need. Not whether the person is trustworthy. What the property does, mechanically, when the goal has been achieved and you no longer need it pointed outward.
If the answer is the same thing, at me, you have not found a risk to manage. You have found that the plan has one stage.
Then the test that costs something: name the lever. Not the intention to remove them — the actual lever, the thing you would pull, the person who would back you, the clause. If you cannot write it in a sentence today, it does not exist, and it will not have been invented by the time you need it. Thiers had three years and never found one, because there was never one to find.
Strongest evidence. The anecdote is unusually clean as a structure: a plan with a stated first step that worked exactly as designed, and a stated second step that never had a mechanism. Whatever is true about the personalities, the shape of the failure is legible from the text alone.
Tension — Greene's diagnosis and his own narrative disagree. He says the unforeseen fact was Bonaparte's ambition.2 The narrative says the instrument was a surname, and a surname's power does not depend on its bearer's ambition. Documented above; not resolved.
Tension — the case may not belong to this law. If the underestimation was a misreading of a capable man, this is a Law 19 case (knowing who you're dealing with), not a Law 29 case. Greene's framing forecloses the question by asserting the imbecility.
🚩 [POPULAR SOURCE] · 🚩 SINGLE SOURCE · 🚩 SECONDARY WITHOUT PRIMARY — the entire episode is given in one paragraph with no citation, no source, and no dates beyond 1848 and "three years later." The 1851 coup and the Second Empire are conventional history; Thiers's motives and his private assessment of Bonaparte are asserted.
Open questions. Was there any lever available to Thiers in 1849–51, or is this a case where the correct play was to lose the election? And how much of the "imbecile" characterisation is Greene's, how much Thiers's, and how much a durable nineteenth-century libel the chapter has simply inherited?
The Greene corpus holds the other Bonaparte at exactly the same moment, and the pair is instructive.
Napoleon, Le Petit Caporal opens with a twenty-six-year-old whose command "was a joke" to his own officers — too short, too young, too badly groomed. Same family, same underestimation by the professionals around him, same subsequent forty years of being wrong about it.
Greene uses the two episodes for opposite purposes. Napoleon I's underestimation is a story about the man overcoming it and becoming the greatest manager of men in history. Louis Bonaparte's is a story about somebody else's planning failure, and the underestimation is left standing as though it were accurate. The corpus contains the observation that Bonapartes get systematically underrated by their political class, twice, and never notices it is the same observation.
Against the vault's Law 28 material the contrast is sharper still. Ivan IV was thought an idiot by the boyars and cultivated it deliberately; Ivan IV and the Five-Year Silence establishes that underestimation can be manufactured and is then never retested, because it protects the people holding it. That is a live alternative reading of 1848 that the chapter never entertains: nothing in Greene's paragraph distinguishes a genuinely stupid man who got lucky from a man who let Thiers keep a convenient belief for three years.
Cognitive psychology — the question he actually answered. The Question You Actually Answered describes the swap: faced with a question requiring expertise you do not have — is this a good car — the mind quietly answers an easier one, do I trust the seller, and hands back the answer with full confidence and no notice that a substitution occurred.
Thiers's hard question was can I remove this man in three years. That requires modelling a constitutional crisis, an army's loyalties, and a public's attachment — none of which he could do standing in 1848. The easy question his mind answered instead was does this man seem impressive. He answered it correctly. Louis Bonaparte did not seem impressive.
The insight neither frame produces alone: attribute substitution is normally illustrated with cases where the easy question is a decent proxy — trusting the seller does correlate with the car being sound. Here the easy question is anti-correlated with the hard one. The men who seem least impressive are, in a hereditary-name system, exactly the men whose power does not depend on seeming impressive, and therefore exactly the men least removable by the means Thiers had. He did not merely answer a different question; he answered one whose correct answer pointed the wrong way. That is a failure mode the psychology literature's examples do not surface, because they are drawn from domains where proxies are noisy rather than inverted.
Story craft — the backpack, and why this anecdote lands. Backpacks — Share Your Plan, Then Derail It names the technique of telling the audience your plan in specific detail so they hold it in their own body, and then breaking it — Dicks's storyteller describing exactly how he will retrieve the spoon, reaching for it, and finding it gone, at which point the room gasps because they had placed the spoon there themselves.
Greene's paragraph is built this way, whether by instinct or design. The plan is given in full — the name, the election, the eventual push offstage — then "the first part of the plan worked to perfection," which loads the backpack, and then three years and eighteen years arrive in a single sentence.
What the craft frame adds is a warning about the evidence. The anecdote's persuasive force is a property of its construction, not of its support, and the construction is doing the work a citation would normally do. A shared-plan-then-derailed structure produces conviction in the listener regardless of whether the plan was ever articulated that cleanly by anyone in 1848 — and Greene's paragraph, which cites nothing, is entirely dependent on it. This is worth carrying beyond this page: the cases in this book that feel most conclusive are frequently the ones built in this shape.
Sharpest implication. When one instrument is the only thing powerful enough to solve your problem, check whether the property that makes it powerful is the same property that makes it uncontrollable — because if it is, there is no version of the plan that works, and the feeling of having found the answer is the sensation of the trap closing. Thiers's plan did not fail. It executed perfectly and the execution was the defeat.
Generative questions.