With gratitude to The Canary Collective for being a second experimenter.
A child is unravelling in a classroom somewhere in this province right now, and the adults in the room have already decided who leaves. The child stays; the other twenty-five file into the hallway with their relief, and a teacher pulls the door shut on a single disabled body left alone with the carpet squares and smell of old lunches.
We call this a room clear.
The teacher who executes it rarely chose it and though she grieves it, she becomes habituated to the mechanics, because she is following the plan—the safety plan, the behaviour plan, the directive that arrived from a boardroom. She has been taught to read her own discomfort as a feeling to manage rather than a signal to obey.
Stanley Milgram had a name for the place she is standing. He called it the agentic state: the condition in which a person hands their moral authorship upward to whoever issued the instruction, becomes the hand and not the will, and continues to do harm they would refuse to author themselves. The famous number from his laboratory—the one that gets taught, repeated, weaponised into a shrug about human nature—is sixty-five percent, the proportion of ordinary people who delivered what they believed were lethal shocks because a man in a grey coat told them the experiment required it. That number does the heavy ideological lifting. It lets us believe obedience is fixed, ambient, simply what people are.
The number nobody teaches is zero.
RadioLab has an excellent episode that summarises Milgram’s work:
The variation they leave out
Milgram ran his study in more than twenty configurations, and a variation pertinent to this conversation is the fifteenth. Two experimenters stand in the room. At the decisive moment, one insists the participant continue and the other says stop.
Faced with two equally legitimate authorities pulling against each other, every participant set the switch down; compliance fell to zero, and people stopped early, exactly at the seam where authority split. The lesson buried in that result rearranges everything the baseline appears to prove.
Obedience lives in the structure, not the soul. A single coherent voice of authority manufactures executioners out of decent people; a single dissenting voice, equally credentialed and standing in the same room, empties the role entirely.
One.
The willingness to harm collapses the instant the front shows a crack, which means the front is the thing the institution must protect above all else—above the child, above the evidence, above its own stated values—because the front is doing all the work.
The economics of closing ranks
This is where Dillbary and Miceli enter, and where the trouble begins. In Of Sinners and Scapegoats, they argue something most of us prefer to leave unexamined: punishing the innocent is usually filed as an error the system labours to minimise, when in truth it is, in their word, “a choice”.
When identifying the actual source of a harm is expensive, an enforcer will rationally elect to sanction the whole group and absorb the innocent as collateral, because the maths favours the sweep over the search. Mass arrests, checkpoints, profiling—the practice persists because it pays.
On End Collective Punishment, I have already read this paper as a description of the staff room: collective punishment endures because it preserves internal group cohesion, spares the institution the costly labour of looking honestly at its own failure to support a child, and converts a problem of inadequate funding and untrained adults into a problem located inside one small body. The room clear is efficient in precisely their sense. It resolves the immediate disturbance, protects the comfort of the majority, and asks no one to name the real injurer, which would be the design itself.
Their model and Milgram’s variation describe the same machine from opposite ends, and they make contrary predictions about the one thing that matters. The economic account assumes the group holds—that cohesion is durable, that the sweep continues, that the lone refusenik is marginal and absorbable and, when troublesome, simply becomes the next thing to be swept up.
Condition 15 says the opposite with brutal clarity: cohesion is brittle, and one credible insider who refuses to perform consensus drops the whole apparatus to nothing. The institution behaves as though Dillbary and Miceli have the final word. It invests staggering energy in the appearance of unanimity precisely because, somewhere beneath the policy language, it knows Milgram does.
The room that performs agreement
Watch how a school defends its front and you are watching an institution that has read Condition 15 in its own body and resolved to deny you the second experimenter.
You walk into the meeting and seven professionals are already seated, already familiar with one another, already agreed; the chairs are arranged, the language is pre-aligned, the conclusion was reached before you arrived and your role is to witness consensus rather than disturb it. Sara Ahmed gave us the grammar for this. The institution that issues a commitment to inclusion and then does nothing has performed non-performativity—the saying that substitutes for the doing—and the family who keeps asking why the saying produced no doing becomes the killjoy, the one who spoils the table’s good feeling, the body marked as the problem for naming the problem.
Elaine Scarry showed us how pain is unmade by bureaucratic language, and the institution has learned the lesson she diagnosed with a fluency that should frighten us. It will never say the student could not be accommodated, because that sentence admits a duty and confesses a failure to meet it.
Instead, it says that the student presented with an escalation staff could not reasonably have anticipated and that the team responded appropriately and followed the plan as written.
The grammar performs two operations at once:
- It removes the adults as authors—the room became clear, a situation arose, supports were in place
- It installs a culprit in the only body small enough to carry the weight, so the disturbance acquires a source and the source is the child
She is no longer the object of the harm; she is recast as its origin, the unforeseeable weather the diligent staff merely weathered. This is the institution’s anaesthetic, and it works on the staff first, because a teacher who can say the team followed the plan has been relieved of the unbearable knowledge that an option stood in front of her wearing her own shoes.
And here is what the performance is for. The seamless front, the pre-agreed conclusion, the euphemism that dissolves agency—all of it exists to guarantee that the divided room of Condition 15 never assembles. Keep the authorities aligned and the harm proceeds at sixty-five percent and climbing. The cohesion the economic model treats as a sturdy given is, in fact, a thing the institution sweats to maintain hour by hour, because it has felt how little it would take to lose.
The teacher is the second experimenter
This is why the ethical teacher is dangerous out of all proportion to her place in the hierarchy, and why the institution fears her in a register it reserves for no parent.
A parent who objects is structurally easy to absorb. We arrive already coded as biased, emotional, over-involved, the difficult mother whose love disqualifies her testimony; the room can file us under “interested party” and proceed, cohesion intact. I have sat in those chairs and felt myself converted into evidence of my own unreliability in real time. The economic model handles me with ease. I am the innocent the sweep is willing to absorb.
A teacher who breaks ranks in public is a different order of event entirely.
Just a Parent
She is the contradiction the model cannot price. She holds the same credential as the adults running the plan, stands inside the same institutional frame, speaks the same professional language—and she says, where it can be heard, this is wrong, and I will not perform agreement with it.
That is Condition 15. She is the second experimenter, and her dissent does what a thousand parental objections structurally cannot: it splits the authority, and the split is the mechanism. This is what the Canary Collective does that End Collective Punishment, by design, cannot.
A teacher writing under her own moral authority about what she has watched her colleagues do is not adding information to a debate. She is fracturing a front.
Kelly Oliver would call her a witness in the strong sense—someone who testifies to what exceeds the official record and, in testifying, restores the possibility of response.
The institution understands this perfectly, which is why it turns on her with a fury that confirms the whole analysis. The teacher who speaks is investigated, isolated, reframed as a conduct problem, reported to the college, quietly counselled about professionalism and fit.
Watch closely: the system is now running Dillbary and Miceli on its own dissenter, scapegoating the one who broke ranks in order to repair the front she cracked. The collective punishment turns inward to defend the conditions of collective punishment. They would rather sweep a good teacher than survive a divided room, and the desperation of that choice is the loudest confession available that Condition 15 is true.
The ones who stay
I have been describing the dissenter as a single figure, and she is two. One has been pushed out and now speaks from the pavement with a clarity that exit purchased—disillusioned, unbought, legible, holding the megaphone and the moral high ground that comes free to anyone who has stopped having something to lose.
The other keeps a foot in each world: present in the building, credentialed, complicit by location, calibrating hour by hour exactly how much truth the room will tolerate before it removes her. The exile owns the language.
The conversation between them—the one outside calling to the one inside who feels the guilt move in her chest—is the transmission line, and it is the precise thing Milgram’s lab could only gesture at, because the lab assumed the second authority had to be standing in the room rather than living in the dissenter’s head, reached the night before and the night before that.
Albert Hirschman named this grammar fifty years ago in Exit, Voice, and Loyalty. A member who watches an institution rot reaches for one of two responses: leave, or stay and make trouble. His genuinely unsettling finding is that the people most capable of effective voice tend to leave first, because they are the most principled and the most pained, and the instant they go the institution sheds the very pressure that might have reformed it.
So the teacher who stays despite her disillusionment, who chooses the friction of voice over the clean relief of the door, is carrying the heaviest load in the building, and carrying it at the price of her own peace. What reads from outside as loyalty—that soft, suspect word—is the attachment that holds a principled person in place long enough for her voice to land.
The binary that sorts every adult into pure-and-outside or compromised-and-inside is the institution’s favourite story, because it coaxes its most dangerous members toward the one exit that renders them harmless.
A teacher who believes that staying makes her a coward is a teacher being talked, gently, into removing her own voice from the room. The guilt is a disciplinary technology. Dancing the line is courage. Here is the edge, and I will hold it rather than smooth it. The gradualism stays moral only while it genuinely tests the threshold. The identical posture that lets a principled woman stay and press can curdle into the alibi that lets a comfortable one stay and rest, where “I’m changing it from within” becomes the line she recites for fifteen years while the room clears continue on schedule. The test is whether she is actually risking ejection or merely invoking the risk to excuse standing still. The honest line-dancer feels the edge because she keeps stepping on it and getting warned; the self-deceiver describes the edge with great eloquence from the safety of the centre.
Sara Ahmed walked the whole arc—she stayed as the killjoy and stayed, pressing, until she judged that staying had tipped into lending her name to the institution’s performance of having listened, and then she left and said precisely why. The line is real. It moves. And exit, taken at the right moment, is the honest terminus of voice rather than its betrayal.
The second experimenter, then, is plural, and the work is a relay. The exile names the harm where it can be read; the insider carries that named contradiction back through the staff-room door in a body that will be present when the directive comes; and the line each of them walks—when to speak, when to stay, when to finally go—is the entire substance of the courage.
The arithmetic of refusal
So the two frameworks sit on my desk making their opposite promises, and I have decided which one to organise around.
The economics of collective punishment describe the world the institution wants you to accept as permanent—efficient, cohesive, self-justifying, a sweep that always continues because the maths always favours it and the group always holds. It is a counsel of exhaustion dressed as analysis, and it asks the disabled child and her family to understand their suffering as the rational by-product of a system behaving sensibly.
Condition 15 describes the exit. One body in a room. One adult, credible and unhidden, refusing to perform the consensus; one teacher who writes it down where it can be read; one voice from inside that splits the authority.
The work of End Collective Punishment in Schools and the work of the Canary Collective are the same work approached from the two sides of the same door—parent and teacher, the doubled contradiction—and the point of building them is not to win an argument the institution has already priced into its budget. The point is to assemble the divided room on purpose, again and again, in public, until the facade cracks.
A child is alone on the carpet squares right now.
Invoke Condition 15. Become the authority that disagrees, or reach the one who can before the door swings; say it out loud, with your face showing, in the single moment the room insists on consensus. The front holds only while we pretend to agree. Uncage your voice! Be a second experimenter.
