Someone inside your organisation needs to go, and you cannot do it alone. So you find others who feel the same way.
That instinct is where most plots begin, and Greene is unusually blunt about the arithmetic:
If there is someone on the inside whom you need to get rid of or thwart, the natural tendency is to consider conspiring with others in your group who feel the same way.1
In most conspiracies the goal is some large-scale action to topple the leader and seize power. The stakes are high, which is why conspiracies are so often difficult and dangerous.1
And the failure mode, which is not clever counter-intelligence:
The main weakness in any conspiracy is usually human nature: the higher the number of people who are in on the plot, the higher the odds that someone will reveal it, whether deliberately or accidentally.1
Franklin, quoted:
"Three may keep a secret if two of them are dead."1
Then the part that makes it genuinely unmanageable:
No matter how confident you may be of your fellow conspirators, you cannot know for certain what is going on in their minds—the doubts they may be having, the people they may be talking to.1
This is not a solvable problem, only a reducible one. Every additional member adds capability and adds an independent chance of exposure, and there is no monitoring arrangement that closes the gap, because what leaks is not information but doubt.
Greene offers what he calls a few precautions, and each is aimed at a specific leak path.
Keep the number minimal.
Keep the number of conspirators as small as possible.1
Compartmentalise.
Involve them in the details of the plot only as necessary; the less they know, the less they have to blab.1
Reveal the timing last.
Revealing the schedule of your plan as late as possible before you all act will give them no time to back out.1
That is a colder instruction than it first reads. It is not about secrecy — it is about removing the window in which a participant could reconsider. The precaution assumes your own people are the risk, which they are.
And then do not change anything.
Then, once the plan is described, stick to it. Nothing sows more doubts in conspirators' minds than last-minute changes.1
The last is the least obvious and probably the most useful. A revision looks like improvisation, improvisation looks like a plan that is not working, and a participant who suspects the plan is not working starts calculating their own position.
Then the warning that reaches past the plot itself:
Even given all this insurance, keep in mind that most conspiracies fail and, in their failure, create all kinds of unintended consequences.1
And the example, which is well chosen:
Even the successful plot to assassinate Julius Caesar led not to the restoration of the Roman Republic, as the conspirators intended, but eventually to the undemocratic regime of the emperor Augustus.1
Note the word successful. The operation worked exactly as designed. Caesar died. And the outcome was the opposite of the objective — the Republic did not return, and what replaced Caesar was a more durable version of the thing he had been killed for being.
Greene's closing formulation of the bind:
Too few conspirators and you lack the strength to control the consequences; too many and the conspiracy will be exposed before it bears fruit.1
That is a genuine scissors and the chapter offers no way out of it. Secrecy and capability trade against each other directly: the size that keeps you undetected is the size that cannot govern what happens next.
Which is why the chapter, immediately after, recommends something else entirely:
In destroying anything from within, you must be patient and resist the lure of large-scale, dramatic action.1
As Canaris showed, the placement of little wrenches in the machinery is just as destructive in the long run, and safer because harder to trace.1
And a redefinition of what counts as a win:
Consider the ability to dissuade your opponents from acting aggressively or to make their plans misfire as a kind of battlefield victory, even if your triumph is surreptitious. A few such victories and your enemy will fall apart from within.1
The contrast with the conspiracy is exact on every axis. The plot requires many people, one moment, and total commitment; the wrench requires one person, many moments, and no commitment at all. The plot has a spectacular success condition and an unmanageable aftermath. The wrench has no visible successes and no aftermath to manage.
The case that proves it is in the same chapter: Canaris's Schwarze Kapelle mounted several assassination plots and every one of them misfired. What actually worked was the quiet, continuous, individually deniable sabotage of Sealion, Gibraltar and the Italian assessment — none of which required a second person to keep a secret. See Canaris and the Invisible Enemy.
You want something inside your organisation stopped, and you are assembling allies.
Start by asking whether the dramatic version is necessary or merely satisfying. Most people reach for the coordinated move because it feels proportionate to the grievance, and proportion is not a strategic criterion. The question is what actually has to change, and the honest answer is frequently smaller than the plot you were designing.
Then count the people, and treat each one as an independent probability rather than as an ally. Every addition brings capability and an unmonitorable inner life. Greene's precautions all follow from this and are worth taking literally: minimum numbers, need-to-know details, timing disclosed last, and no changes once the plan is stated, because a revision reads as a plan that is failing.
Then look hard for the version that requires nobody else. This is the actual recommendation of the chapter and it is buried under the conspiracy material: a wrench in the machinery, placed by one person, repeatedly, with no coordination to expose. Every objection to that approach is about satisfaction rather than effectiveness — it is slow, invisible, and nobody will know you did it.
Then, if you must coordinate, plan the aftermath before the act. Caesar's assassins had a flawless operation and no answer to the following morning, and they got Augustus. The scissors is real: the group small enough to stay secret is too small to control what happens next, so if you cannot describe who governs the day after, you are not planning a strategy, you are planning an event.
And be honest that the odds are poor. Greene says most conspiracies fail, and the one he names as successful produced the opposite of its goal.
The general claim — that conspiracies scale badly because exposure risk rises with membership — is sound and well supported historically. The Caesar example is accurate and apt. [POPULAR SOURCE]
The precautions are offered without a single case of them working. Greene supplies a plot that succeeded and backfired, a general claim that most fail, and four pieces of advice attached to neither. A chapter that recommends against conspiracies and then supplies technique for running them anyway is doing both jobs badly — the analysis is the valuable part and the tips read as an obligation to be useful.
And "reveal the schedule as late as possible so they have no time to back out" deserves to be named for what it is. That is a technique for removing your own collaborators' ability to reconsider a serious and possibly criminal act. It is presented as operational hygiene. Greene's framing treats your co-conspirators' second thoughts as a security vulnerability rather than as information — and in a plot with a genuine moral dimension, the participant having doubts may simply be right.
The Caesar point is stronger than Greene uses it. He deploys it as a caution about unintended consequences, but the lesson is more specific: the conspiracy's constraint on size is what guaranteed the failure. The plot needed to be small to stay secret, so it was too small to hold Rome afterwards, and the vacuum was filled by someone who had not been in the room. That is not bad luck — it is the structure of the scissors, and it means the aftermath problem is not solvable by better planning within the constraint.
Against Strategy 31: The Inner-Front Strategy this section functions as the chapter's internal warning against its own most obvious application. The chapter teaches you to get inside and work from within; this passage says that the dramatic version of "working from within" is the one that reliably fails.
Against Canaris and the Invisible Enemy the case demonstrates both halves. The Schwarze Kapelle's assassination plots all misfired; the solitary, deniable, unspectacular sabotage worked for a decade. Greene names Canaris in this very passage as the model for the wrench, and does not note that the same man ran the conspiracy version alongside it and got nothing from it.
Against Strategy 26: The Strategy of the Void the cell doctrine there is the structural answer to the problem stated here — small, dispersed, compartmentalised units precisely because a large coordinated body cannot keep a secret or survive detection. Greene poses the problem in Strategy 31 and solved it in Strategy 26, five chapters earlier, without cross-referencing either way.
Against Strategy 5: The Command-and-Control Strategy there is a genuine tension. Strategy 5 holds that divided leadership causes history's greatest defeats and that unity of command is decisive. A conspiracy is by construction a body with no unified command — a group of equals who cannot appoint a leader without creating exactly the figure they are conspiring against. That is a real structural reason conspiracies fail, and neither chapter names it.
Psychology → What leaks is doubt, not information. Greene's diagnosis is that you cannot know the doubts they may be having, the people they may be talking to — the risk is a state of mind rather than a security failure. The vault's Cognitive Dissonance holds the mechanism: a person holding a commitment they are no longer sure of is under pressure to resolve it, and talking is a resolution.
The insight the pairing produces: the security problem in any coordinated secret is not that people are careless but that unresolved commitment is psychologically expensive to hold, so participants will act to relieve it — by confessing, by hedging, by telling one person. Which is why Greene's timing precaution is technically correct and morally ugly: shortening the window is the only intervention that works, because it reduces the duration of the discomfort rather than trying to manage it. And it identifies the honest version of the same insight for anyone building anything difficult with other people: a collaborator's doubts are not a leak risk to be engineered around, they are the earliest available signal that the plan has a problem, and a structure that suppresses them has traded information for security.
History → The size that keeps a secret cannot govern the aftermath. Caesar's assassins succeeded and produced Augustus. Set that beside The Culminating Point of Victory, where the arc of success carries past the point at which it can be held.
Together they produce a claim about a whole class of actions: any operation whose secrecy requirement caps its membership has, by that same cap, guaranteed it cannot control what follows — which means the plan's constraint and the plan's failure mode are the same variable. This is why coups so reliably produce someone other than the plotters, why the person who removes an obstacle rarely inherits the space, and why the aftermath cannot be fixed by better planning inside the constraint. The only real answers are to make the act small enough not to create a vacuum, or to accept that you are clearing a path for whoever is standing nearby — and the second is a decision most conspirators never notice they are making.
The group small enough to keep a secret is too small to govern what happens afterwards — which is not bad luck but the structure of the thing, and it is why Caesar's assassins ran a flawless operation and got Augustus. The alternative the chapter buries is the one that works: a wrench placed by one person, repeatedly, with no coordination to expose and no aftermath to control.