The whole intervention is furniture.
In his family-therapy practice, Dr. Milton H. Erickson found that if the family dynamic was unsettled and dysfunctional there was inevitably one person who was the stirrer, the troublemaker. In his sessions he would symbolically isolate this rotten apple by seating him or her apart from the others, if only by a few feet. Slowly the other family members would see the physically separate person as the source of their difficulty.1
⚠ No confrontation, no accusation, no interpretation offered aloud. A chair is moved and, over some weeks, everyone in the room arrives at a conclusion.
⚠ It is the smallest technique in Law 42 and probably the most powerful, and the corpus should be precise about what it actually does — because the passage says something different from what Greene concludes from it.
⚠ Read the sentence again.
Slowly the other family members would see the physically separate person as the source of their difficulty.
⚠ The seeing follows the seating. Not: the separation made an existing pattern legible. The separation caused the attribution.
⚠ And that means the technique has no truth condition. Seat the actual stirrer apart and the family identifies the stirrer. Seat the quietest person in the family apart and — on the mechanism as described — the family identifies the quietest person.
⚠ The corpus should record this alongside Greene's third isolation, which is defined as slander and insinuation. Both work identically whether the diagnosis is right or wrong. ⚠ Law 42 has now supplied two techniques for making a group blame someone and none for finding out whether they should.
⚠ What makes the Erickson version defensible is not built into the method. It is that a clinician made the diagnosis first, in a professional setting, with the family's own account in front of him — and Greene's reader has none of that and is told to run the same manoeuvre.
⚠ The passage's closing line is the best sentence in Law 42's Keys and it does not belong to Law 42.
Stirrers thrive by hiding in the group, disguising their actions among the reactions of others. Render their actions visible and they lose their power to upset.1
⚠ The mechanism here is illumination, not removal. The stirrer's power comes from attribution failure — their moves are read as everyone's reactions — and the countermeasure is to break the camouflage.
⚠ Nothing in that requires banishment. The person can stay in the room, in the job, in the family, and simply stop being invisible.
⚠ Which is a materially better technique than the chapter's title for the most common real case. You usually cannot banish anyone. You can almost always make a pattern legible — and the passage says legibility alone removes the capacity to upset.
⚠ The corpus should record the divergence plainly: Law 42 is called strike the shepherd, and its most transferable case says turn the light on.
🚩 ⚠ There was inevitably one person who was the stirrer.
Another universal, and this one is a clinical claim rather than a rhetorical one. [CONTESTED].
⚠ The dominant framework in family systems runs the other way. The person carrying the visible trouble is generally treated as the identified patient — the member in whom a systemic dysfunction surfaces — rather than as its cause, and the whole point of the framing is that removing or blaming that person leaves the system intact to produce another one.
⚠ The corpus is not in a position to adjudicate what Erickson himself held, because Greene names no work. 🚩 A named clinician, an extensive published output, and no book, paper, case or date. 🚩 SECONDARY WITHOUT PRIMARY.
⚠ But the two readings make different predictions and one of them is testable. If the stirrer is the cause, removal ends the trouble. If the stirrer is the symptom, removal produces a replacement within a predictable period — and that is exactly the observation the chapter would need in order to distinguish them, and never makes.
🚩 ⚠ The phrase rotten apple is Greene's.
A clinician moving a chair to change how a family attributes its distress is being described in the vocabulary of spoilage — ⚠ and it is the same substitution the corpus recorded on the Dante page, where a city's defender is called a cancer.
⚠ The vocabulary is not decorative. It is doing the argument's work. A rotten apple cannot be improved, does not have a perspective, and is not owed a hearing. Once the metaphor is accepted, the chapter's "they are irredeemable" follows without needing to be argued.
🚩 ⚠ And the purpose is swapped. Erickson's separation is an intervention on the family, aimed at changing a dynamic in the room. Greene's extraction is: "Once you recognize who the stirrer is, pointing it out to other people will accomplish a great deal." ⚠ The therapeutic aim is repair. The extracted aim is coalition-building against a person. Same manoeuvre, opposite objective, and no acknowledgement of the swap.
⚠ The constraints on the Erickson version are what make it safe, and every one of them is dropped in the translation.
If only by a few feet. ⚠ The intervention is minimal, and a minimal intervention can be wrong without being catastrophic. The chair moves back.
It happens in the room, in front of everyone, including the person being seated apart. ⚠ The target can see the manoeuvre and respond to it — which is not true of slander and insinuation, the chapter's other route to the same effect.
And it is performed by a third party with no stake in the outcome. ⚠ The therapist does not benefit from the family blaming any particular member. Greene's reader is the person the stirrer is stirring against — the accuser, the beneficiary and the judge, in one body.
⚠ So the corpus should record the honest transfer. What generalises is the observation: attribution in a group is enormously sensitive to physical and procedural arrangement, and a small change in who is standing where can move a whole room's account of what is wrong.
⚠ What does not generalise is the legitimacy. The technique's safety in Erickson's hands came entirely from the diagnosis preceding it, the reversibility, and the operator's disinterest — and none of the three survives into Law 42's version.
Use the visibility half, not the seating half. ⚠ Render their actions visible and they lose their power to upset. The transferable move is to break the camouflage — name the pattern, put the sequence of events somewhere everyone can see it, describe what happened without characterising who did it. That works on a real stirrer and does nothing to an innocent person, which is exactly the property the seating lacks.
Do the diagnosis before the arrangement, and separately. ⚠ If you find yourself designing a room, an agenda or a seating plan that will lead others to a conclusion, you have already made the finding privately. Write it down first, with what it rests on. The arrangement will produce agreement either way, so the agreement will not tell you whether you were right.
Run the replacement test. ⚠ If the trouble is caused by a person, removing them ends it. If the trouble is produced by the arrangement, a new stirrer appears within a few months and everyone is surprised. The second outcome is common enough that it is worth predicting out loud in advance — it is the cheapest available check on your own diagnosis.
⚠ And notice who is holding the chair. Erickson had no stake in which family member was blamed. If you do, the technique will work and its result is not evidence. The minimum correction is a third party who can see the same facts and reach a different conclusion — and if there is nobody in that position, that itself is the finding.
Strongest evidence. The mechanism is specific, minimal and observable: physical arrangement shifts group attribution, slowly and without argument. ⚠ And the closing line — stirrers thrive by hiding in the group, disguising their actions among the reactions of others — is a real account of how the behaviour is sustained, and it implies its own countermeasure.
Tension — the passage says the separation caused the attribution and the chapter treats it as revealing one. ⚠ Preserved: on the mechanism as described, the technique has no truth condition.
Tension — Greene's extracted purpose is coalition-building; Erickson's stated setting is family repair.
🚩 [POPULAR SOURCE]. 🚩 SECONDARY WITHOUT PRIMARY — a named clinician with a large published output, no work, paper, case or date. 🚩 [CONTESTED] — "inevitably one person" runs against the mainstream family-systems reading, in which the member carrying the trouble is the identified patient rather than its cause. 🚩 Rotten apple is Greene's vocabulary and it does the argument's work.
Open questions. ⚠ Did the families get better? The passage reports that they came to see the separated person as the source, and stops. Whether the dynamic improved is not stated, and it is the only outcome that would distinguish the two readings. ⚠ And how long does an induced attribution last once the chair is moved back?
⚠ This is Milton Erickson's second page in the build. The first is the ironwood tree at Law 37, where he redirects a patient's attention through a symbol rather than an argument.
⚠ Both times the shape is the same and it is not the shape of the chapter he is in. At Law 37 he is used to illustrate spectacle and is really illustrating indirection. Here he is used to illustrate striking the shepherd and is really illustrating attribution.
⚠ And both times the manoeuvre is minimal, reversible and non-verbal — a tree, a chair — while the chapters around them recommend banishment, spectacle and slander. The corpus should log the pattern: Erickson is the book's most consistently small operator, imported into arguments for large moves.
Behavioural mechanics — manufacturing the perception rather than reporting it. Attacking Your Own Church to Bind It Together records Jim Jones engineering gunshots, threats and a faked shooting against himself, so that the congregation would feel besieged and cling closer — manufacturing an enemy when real ones were insufficient, because a group united against a threat is more controllable than one merely offered benefits.
Slowly the other family members would see the physically separate person as the source of their difficulty.
The insight neither produces alone: ⚠ the two techniques sit at opposite ends of a scale and share one engine — a group's account of what is wrong with it is installable from outside, and the installation does not require any claim to be true. ⚠ Jones installed an external enemy; the chair installs an internal one. Both produce cohesion, and in both cases the cohesion is real while the account is arbitrary. ⚠ Which supplies the diagnostic the chapter needs and does not have. The question is not is this person the problem — a group will converge on whoever is arranged into the position — but who arranged the room, and what do they get if the group agrees? ⚠ Erickson gets nothing, which is why his version is a treatment. Jones gets everything, which is why his is a mechanism of control. And Greene's reader is somewhere in between and is never asked to notice which.
Business — the three layers blame peels in. Onion of Blame describes resistance coming apart in a predictable order: circumstances first (I don't have time), then others (I need to talk to my partner), and sometimes — not always — self (I'm not sure I'd actually follow through).
Stirrers thrive by hiding in the group, disguising their actions among the reactions of others.
What the pairing produces: ⚠ the onion gives the sequence a distressed family is already moving through, and it explains why the chair works so easily. A family in trouble has usually exhausted circumstances — money, work, the house — and is somewhere in others. ⚠ Seating one member apart does not introduce a new idea; it resolves an argument the group is already having and stops the sequence at the second layer. Which is where everybody would prefer it to stop, because the third layer is self. ⚠ That is the technique's real leverage and also its danger: it is offering the group the answer it most wants, one step before the answer that would cost them something. ⚠ And it gives the corpus a sharper reading of Law 42 as a whole. The chapter is, structurally, an others-layer instrument — trouble traced to a single external author, no self-layer anywhere in it — which is exactly why it is so satisfying and exactly why it should be run last rather than first.
Sharpest implication. ⚠ The seeing follows the seating — the separation caused the attribution rather than revealing it, so the technique has no truth condition, and Law 42 now supplies two ways to make a group blame someone and none for finding out whether they should. ⚠ Render their actions visible and they lose their power to upset is a different law from the chapter's title: illumination, not removal, and available when banishment is not. ⚠ The safety was in the constraints — a few feet, reversible, in the room, by a party with no stake — and all four are dropped in translation. ⚠ And the chapter is an others-layer instrument, which is why it satisfies.
Generative questions.