The fourth search method changes the object of the search:
Sometimes in your search for weaknesses it is not what but who that matters. In today's versions of the court, there is often someone behind the scenes who has a great deal of power, a tremendous influence over the person superficially on top. These behind-the-scenes powerbrokers are the group's weak link: Win their favor and you indirectly influence the king.1
And a second application:
even in a group of people acting with the appearance of one will—as when a group under attack closes ranks to resist an outsider—there is always a weak link in the chain. Find the one person who will bend under pressure.1
Two different claims under one heading, and only the first is about influence.
Greene gives the reason later, in the Richelieu interpretation, and it is the sharpest sentence in the method:
This person may have more weaknesses than the king himself, because his power depends on all kinds of capricious factors outside his control.2
That is a genuine structural observation and it holds up.
A king's position is institutional. It survives his being disliked, absent, ill or wrong. There is a procedure by which he holds it and the procedure does not consult anybody's affection.
A favourite's position is relational. It exists entirely because one person continues to want them near, and that person's mood, health, marriage and attention are all outside the favourite's control.
So the favourite lives with a permanent insecurity the principal does not have — and insecurity is the raw material every other method in this chapter is looking for. The structure manufactures the weakness, which means you can locate it from an org chart before you have met anybody.
That is the most operationally useful thing in Law 33: it is the only method that identifies a target by position rather than by observation.
The closing-ranks version — "there is always a weak link in the chain. Find the one person who will bend" — is not about influence. It is about breaking a coalition, and its mechanics are unrelated.
The favourite is approached because they can carry your influence upward. The bending member of a closed group is approached because they can stop resisting, and what you get is not a channel but a defection.
Different targets, different asks, different outcomes if you fail. Greene runs them together because both answer who? rather than what?, which is a similarity of question and not of technique.
And the second claim is the weaker one. There is always a weak link is the same universal, unfalsifiable move as the law's founding premise: any failure to find the bender is attributable to insufficient searching, and any group that holds simply had its weak link inadequately probed.
The chapter's case is unusual because it shows the method applied, exhausted, and reapplied.
First application. In 1615 the young bishop ends a speech to the three estates by addressing only the Queen Mother — "he looked directly at and only at the queen mother", praise "so glowing that it actually offended some in the Church."3
The reading behind it: *"Not that Marie was obviously weak—she governed both France and her son; but Richelieu saw that she was really an insecure woman who needed constant masculine attention."*2
Then the sub-application, which is the method inside the method: Marie's own favourite was Concino Concini, and Richelieu "played him perfectly—attending to him as if he were the king."4 The weak link had a weak link.
Then the reapplication. Once Richelieu has the premiership, Marie is spent: "Now Richelieu no longer needed Marie de Médicis. He stopped visiting and courting her… Instead he concentrated on the king."5
Greene's own summary: *"Once Richelieu had the position he desired—prime minister—he discarded the queen mother, moving on to the next weak link in the chain: the king's own character."*2
The weak link is a rung, not a destination. The method is iterative and each application makes the previous one disposable, which the chapter states plainly and does not examine.
The part Greene does not price: Richelieu spent roughly eight years — 1615 to 1623 — cultivating a woman he then discarded, and the discarding was visible.
That has a cost the chapter cannot see because it never follows a case past its success. A person who has watched you use and drop the favourite is a person who now knows what you do to favourites, and the next rung is standing right there.
Louis XIII "hated the cardinal with a passion, only tolerating him out of love for Marie."6 Richelieu's eighteen years were spent managing a king who had watched him work.
The method's iteration is visible to everyone above the current rung, which is presumably why Greene's advice at this point shifts from win their favour to make yourself indispensable — you cannot go on being liked once the pattern is legible, so you switch to being necessary. He does not say this; the case shows it.
You need something from someone whose door you cannot get through.
Stop working on the door. Draw the actual map: who does this person listen to, and whose position depends entirely on them?
Those are two different lists and the intersection is the target. Somebody whose influence is real and whose standing is relational rather than institutional — a chief of staff, a long-serving assistant, a favoured deputy, the person who has been there for years without a title that would survive a reorganisation.
Their insecurity is structural and you do not need to detect it. It comes with the position, and it consists of one thing: nothing they have is theirs.
What that person needs is almost never money or favours. It is recognition of the influence they actually have and are never publicly credited with — treating their judgement as consequential, in private, without asking for anything in the same conversation.
Then the discipline the case makes necessary. Do not treat them as a rung. Richelieu's iteration worked in a court where he could survive being seen to discard people; almost nobody else can. If you get what you needed and the relationship immediately cools, you have advertised exactly what you are, to an audience that includes the person you were trying to reach.
The version that survives repetition is the one where the relationship is not conditional on the ask — which costs you the eight years Richelieu spent and does not cost you the eighteen he spent afterwards managing a man who had seen him do it.
Strongest evidence. A structural argument that explains rather than asserts — relational power is inherently insecure — plus a case that shows the method applied three times in one career, with a nested application in the middle.
Tension — two unrelated methods under one heading. The favourite carries influence upward; the bending coalition member defects. Different targets, different mechanics.
Tension — there is always a weak link is unfalsifiable. Same shape as the law's founding premise.
Tension — the iteration's cost is never counted. Richelieu's discarding of Marie was visible to the man he spent the next eighteen years managing.
🚩 [POPULAR SOURCE] · 🚩 SINGLE SOURCE · 🚩 SECONDARY WITHOUT PRIMARY — no source for the 1615 speech, the Concini relationship, or the 1622 cardinalate. The dates are conventional; the readings of everyone's inner states are Greene's.
Open questions. Does the structural insecurity claim hold where the relational position is long-established — a favourite of thirty years may have accumulated institutional protection and be less vulnerable than the theory predicts. And: is there any case in the corpus of an operator identifying the wrong weak link, or does the method's org-chart shortcut also inherit the chapter's missing failure mode?
Richelieu is on his second page in this vault, and the first one is worth reading against this.
Richelieu and Marillac's Brother is the sequel, and it is the cost of the iteration arriving. In 1631 — the Day of the Dupes — a plot to remove Richelieu reaches the top of the French government, and the queen mother is at the centre of it. The woman he cultivated for eight years and then discarded spends the following decade trying to destroy him, and on that day nearly does.
The corpus contains the bill for this method and the chapter does not. Greene's account ends with Richelieu governing France for eighteen years; the vault holds the year the discarded rung came back.
Within Law 33, this method is the odd one out and the strongest for it. The other five all read a person — gestures, childhood, contrasts, voids, uncontrollable emotion. This one reads a structure, and structures are legible without proximity, do not require a baseline, and do not depend on the operator's interpretive skill.
Which raises a question about the plan as a whole: it is the only method that would work on somebody you have never met, and it is buried fourth in a list of six. See The Strategic Plan for Finding Weakness.
Organisational analysis — the influence map that is not the org chart. BANT Qualification formalises the problem of identifying who can actually authorise something, on the working assumption that titles and decision rights diverge and that selling to the wrong person is the commonest failure.
Greene's method is that problem's older sibling, and the sales frame supplies a distinction he lacks.
The insight neither produces alone: BANT separates authority from influence — the person who can say yes and the person who can say no are frequently different, and a third person shapes what gets considered at all. Greene's weak link collapses these into one figure. But the three roles have completely different vulnerabilities: the authoriser is institutionally protected, the blocker's power is entirely negative and therefore precarious, and the shaper's influence is invisible and therefore chronically unacknowledged. The third is the one this method actually describes, and it is the one whose need is most reliably unmet, because their contribution is by construction never on the record. That is a sharper target definition than "someone behind the scenes" and it comes from a sales methodology rather than from a manual about courts.
Network theory — the tie whose removal disconnects. Technology Lock-In and Path Dependence examines how a system's behaviour is determined less by its components than by the structure of dependencies between them, and how a single load-bearing link can constrain everything downstream of it.
Applied here it reframes what a weak link is, and the reframing is not semantic.
What the pairing produces: Greene means weak as fragile — the person most likely to bend. The structural reading is weak as thin — the connection through which everything must pass and which has no redundancy. Those identify different people. A favourite who is emotionally fragile but easily replaced is a poor target; an unglamorous person through whom all the information flows and for whom there is no substitute is the actual chokepoint, and they may be entirely secure. That is Rockefeller's railways applied to an organisation of people, and it is a better version of this method than the one the chapter gives — find the input everyone treats as a utility works on staff exactly as it works on freight.
Sharpest implication. This is the only method in the chapter that identifies a target by position rather than by observation — you can locate it from an org chart, before meeting anyone, because relational power manufactures its own insecurity: nothing the favourite has is theirs. But Greene's weak means fragile, and the more useful reading is thin — the connection with no redundancy, whose holder may be perfectly secure. And the method is iterative, which the chapter states and never prices: Richelieu's discarding of Marie was watched by the man he then spent eighteen years managing.
Generative questions.