The Alibi of Unanimity

A summer afternoon in a small town in Texas. Population 5,607. The thermometer outside the drugstore reads 104°F. That is 40°C.

One person speaks up: let's drive to Abilene and have dinner. Fifty-three miles each way, 106 miles there and back. In kilometers that is over 170. The car has no air conditioning.

And nobody objects. One says it sounds good, the next says it sounds good, the one after that says it sounds good. All four get in the car.

After they return, one of them says they had not wanted to go at all. Then the next says the same thing. So does the person who proposed it. All four had wanted to stay home.

Everyone assenting to a decision no one wanted. This is called the Abilene paradox. In 1974 an organizational behavior scholar named Harvey gave it that name, presenting it as something that had really happened in his own family.Original citation — J. B. Harvey, "The Abilene Paradox: The Management of Agreement," Organizational Dynamics 3(1), 1974. The widely circulated attribution to the Harvard Business Review is wrong. Every passage this piece carries was read directly in the 1988 reprint, and reprint page numbers vary by source, so we give none.

Move it to a setting we know. The company dinner nobody ever asked for, already on the calendar. The organizer booked it believing everyone wanted it, and everyone else went along believing the organizer wanted it.

One image is left standing here. Nobody set the direction, and the car is already moving.

The question usually put to this case is: all four hated the idea, so why did none of them say so. We ask a different one. Why did they say let's go? The wrong culprit has been booked for this case.

The First Suspect Has an Alibi

The suspect called first is groupthink — the phenomenon in which a highly cohesive group folds away its own dissent to preserve unanimity. Think of being the wet blanket in a group chat that is already excited. Once the mood has already tipped to one side, a word against it is read as an attitude rather than as content.

The explanation is a strong one. The tighter the group, the more dissent is pressed down, and pressed-down dissent looks like unanimity. Widely read explainers and encyclopedia entries write Abilene the same way: each person is against it but, mistakenly believing the others agree, holds back from expressing it.

We counted for ourselves how far that explanation actually circulates. The subject is web documents surfacing high in search as of 10 August 2026, and what we read through the body text and coded comes to 13 English explainers. Three Korean-language items were read at headline level only.The window we swept is wider than that — six searches and five fetches surfaced roughly 55 addresses. The 55 is the window we scanned, not the denominator of the coding. The result fell heavily to one side. Vocabulary in the silence, avoidance, and reading-the-room family was overwhelming, and the frame "they tell each other the opposite" appeared in not a single item. As prescription, "honest and open communication" dominated.

What that sample tells us goes exactly that far and no further. Documents that surface high in search are themselves skewed, and we could not reach the primary text of a university textbook. So we will not widen this to "this is what people know." Widely read explainers write it this way. That much.

The specialist literature is a different matter. Follow-up comparative work has already separated the two constructs along eight axes.The eight axes = cohesion · leadership style · an external enemy · individual commitment · felt coercion · responsibility · satisfaction with the outcome · the target of blame. The man who supplied the name separated them too. Of events another researcher had offered as cases of groupthink, he wrote that he "would conceptualize as mismanaged agreement." So we will not be the ones to say that the academy blurs the two. They were separated long ago, and the explanation in circulation still speaks in terms of silence. Why that gap matters is this: the moment you explain it by silence, the two illnesses reduce to the same sentence. Both become "something that happened because nobody spoke up." One item of vocabulary makes two illnesses reducible to the same sentence, and that reduction travels alongside the prescription "let's get people talking more." What we observed extends to the fact that the two appear together; which of them produced the other is something this sample cannot settle.

But this suspect has two alibis. They are not counter-evidence imported from outside — they come from asking whether the original text is consistent with itself in the passages where it describes its own case.

First, there was no dissent in the room to press down. The first two items of the original's symptom list record that on both the reading of the situation and the response required, every member privately agreed. You cannot charge someone with suppressing dissent when there was no suppressed opposition.

Second, everyone voiced their dissatisfaction immediately after the decision. The fifth symptom records frustration and anger, the formation of subgroups, and mutual blame. The comparative literature on groupthink offers the opposite as characteristic of that illness: members express satisfaction with the decision. That contrast, though, was not read off the primary text; it comes through follow-up literature.

Which leaves a question. If there was no dissent in the room to press down, why did individuals say the opposite of what they wanted? We will not settle the answer in advance as a problem of courage. First we have to look at the body again.

They Did Not Fall Silent. They Said the Opposite.

The original paper lists six symptoms, and the third is the soft spot. It records that organization members fail to communicate their desires and beliefs accurately to one another, and then the sentence steps up a level in force. "In fact, they do just the opposite." They do precisely the reverse — and so they lead one another into misperceiving the collective reality.

What the Abilene group actually put into the room was not silence. It was an utterance: "Yeah, it's a great idea. Let's go to Abilene." From four people, every one of whom wanted to stay right where they were.

Actively supplying assent that runs against your own preference. In this piece we will call that forged assent. The name is not the original's vocabulary but one we have attached here. The original writes it as members having "communicated inaccurate data," and as doing "just the opposite." Picture the thumb typing "sounds great!" into a group chat while the person behind it wants none of it. They did not fail to speak. They spoke the reverse.

So what was supplied to that room was not agreement but contaminated data. The passage where the original paper restates the paradox later runs the same way: organizations take action in contradiction to the data they already hold for dealing with their problems, and so compound their problems instead of solving them. The Abilene decision is not wrong because it was pressed down. It is wrong because it ran normally on a contaminated input.

There is a fact inside the same paper that cuts against us. That paper also contains the word "collusion," and elements of consensus-preserving silence sit alongside it. A widely read encyclopedia entry defines Abilene as the suppression of expression. So one could object that silence-versus-forgery is an overstated dichotomy.

Granted. But our claim is not that there was no silence at all. It is that the original's primary emphasis falls not on silence but on reverse transmission, and that the explanation in circulation lost exactly that half. The original speaks of inaccurate transmission and then, immediately after, raises the force and nails it down. And what was actually placed in the room was not suppressed expression but expression uttered in the opposite direction. In the circulating sample we read, that frame appeared zero times. What we are correcting is not the meaning of the original but the half that vanished on the way from the original to the versions in circulation.

For forgery to happen, something has to be in place first: a state in which everyone's own view is the same and only the estimates of everyone else's views are wrong. This is called pluralistic ignorance. It is the moment in a lecture hall when no hand goes up. The whole room is thinking "I must be the only one who didn't follow that," and so nobody asks. The same misreading, though, takes different forms in different rooms. In the lecture hall it comes out as silence. In the room in Abilene it comes out as an utterance of assent.

The misreading shows up in experiments too. In one campus study, students believed they were more uncomfortable with a given practice than the average student was. Their own attitude and their estimate of everyone else's attitude were systematically out of line. Sample sizes and scales by study, however, sat behind the primary text we could not reach, so we give no figures.

Here the strongest objection arrives. Isn't Abilene a management parable with no empirical basis? It is a fair point. The original paper is a theoretical essay built on a single family anecdote from a hot summer day. The author presented it as a real experience of his own family, and attached no label marking it as illustration.

So we put no empirical weight on that anecdote. We use Abilene as a name tag — as grounds for who formalized what, and nothing more. What we build is the mechanism underneath it.

The lineage of that mechanism runs in two branches of unequal strength. The pluralistic-ignorance branch is carried by an independent experiment. A different research team ran a coordination game on the question of why bad norms persist, and wrote in its conclusion that the results give "clear support for the pluralistic ignorance perspective." Two layers separate here. That the experiment's conclusion supports pluralistic ignorance is a citation. Laying forgery in the meeting room on top of it is our transplant — what that experiment measured was norm persistence in a coordination game, not a meeting room. The other branch is thinner, and that story belongs to the next section.

Now run one full turn. The original's symptom list is itself the chain.

Each person privately wants the same thing → each transmits the opposite of their own preference to the others → a decision is built on invalid information → an implementation nobody wanted arrives, along with mutual blame → and they leave the room with the underlying problem untouched.

The last link matters most in this piece. The paper writes that if the generic issue — the inability to manage agreement — is not dealt with, "the cycle repeats itself with greater intensity." The feedback is not our reading. It is the original's own wording.

Why the car at the top of this piece pulled out is explained here for the first time. Nobody set the direction. Four people transmitted one. And this does not end after one turn.

The Room in Abilene Is Not a Quiet Room. It Is a Talkative One.

Change the culprit and the whole picture of the room changes with it. Forgery is utterance. Utterance makes noise. So the room in Abilene is not quiet. Assent goes back and forth. The picture of a good mood and a meeting that ends ahead of schedule is not an observation recorded in the original but a prediction we drew out of this mechanism. If that prediction holds, the danger sign stands on exactly the opposite side from where we were taught to look for it. This picture stands on the pluralistic ignorance established earlier. Everyone is misreading everyone else, so the noisier the room, the more the misreadings harden one another.

What a room like that needs cannot possibly be "more talking."

One more scene attaches here, on thin evidence. A single experiment and one computational model built after it are the whole of this story. It is the scene in which a person who privately objects takes the lead in suppressing objections from others. Think of the person who hated a practice most as a new hire and, once senior, enforces it on juniors more strictly than anyone. Why? Because the cheapest way to prove sincerity is to police someone else. That experiment's sample size and effect size were released only as far as the abstract, and we could not reach them. Reports of independent replication we did not find within our search range, and neither did we find reports of failed replication. So this paragraph is not something our argument rests on. It is something our argument predicts. If it holds, the picture sharpens; if it does not, the conclusion of the previous section stands unchanged on the original's own wording.

A different objection arrives here. Isn't this already explained as an information cascade? The structure in which people back out information from the behavior of those ahead of them and discard their own signal was formalized in finance long ago, and everything from fashions to customs is explained on that frame.

The rebuttal only goes so far. Two axes separate the two, and to those we add one item confirmed on the Abilene side only.

First, what gets discarded differs. What a cascade discards is private information. The original paper writes that at some stage a decision maker will "ignore his private information" and act only on the information obtained from previous decisions, and calls that judgment "entirely rational." What Abilene discards is not information but one's own preference. In the room in Abilene, what gets updated on seeing someone else's utterance is not one's preference but one's estimate of what everyone else wants. While that estimate updates in the wrong direction, the preference is not updated — it is simply thrown away. Where the preference itself genuinely rests on other people, as with going out for dinner, adjusting to what others want would be a normal update. But the first two symptoms in the original record that everyone in this room was already privately on the side of staying.

Second, the head of the chain differs. In that model, the first action in the chain still stands on the actor's own signal, and information stops later in the chain. That is why the initial signal governs the whole chain. Abilene is forged from the first utterance. On the first two symptoms, everyone in the room was privately on the same side, so the person who first said "let's go" was also speaking against their own preference. At no point in the chain did a genuine signal ever enter.

Third, the aftermath differs. For Abilene the original names frustration and mutual blame as symptoms. The cascade paper we read does not define this phase in those terms. So this axis is not the result of putting the two illnesses face to face — it is an item confirmed on the Abilene side only.

There is also a discriminator we will not use. We will not use the axis "only Abilene flips suddenly." The cascade paper itself writes that cascades are fragile — that "small shocks can frequently lead to large shifts in behavior."

Put the two illnesses side by side. The dividing axis is the first column; the remaining columns are consequences of that axis.

The same split turned up in rooms of non-human actors as well: in one experiment running 100 scenarios across ten domains on eight language models, agents that privately disagreed with a norm conformed publicly 64–94% of the time, and 52–92% even with the false-consensus framing stripped out (this is a preprint, not yet peer reviewed, with a single author). There are reports pointing the other way too — another study reproducing a 12-member jury ended with 17 of 18 runs failing to reach agreement at all (Abilene-type convergence itself failed there). The conformity experiment does not refer to the Abilene paradox itself, so joining the two phenomena under one structure is our reading. That the same split appears in actors for whom human stakes do not exist is, given that those actors learned from human speech, closer to a mirror reflecting the same structure than to independent evidence — but that forgery is engraved on that mirror is a signal that forgery attaches to the rules of discourse rather than to individual timidity.

So what should be done.

What the Cockpit Fixed Was Not the Chance to Speak but the Price of Speaking

Start with what the received view actually prescribes. In the circulating sample we read, the dominant prescription was "honest and open communication." One common form that prescription takes in a meeting room is going once around the table so everyone gets a turn to speak.

Now set that prescription down in the room we have just seen. The room in Abilene is already talkative. What a device that adds speaking turns aims at is silence in the room, and the main illness in this room is not silence. The main target is off, so the prescription is not so much wrong as beside the point. That much follows from swapping the culprit. Had the culprit been silence, the same prescription would have been the right answer. That the added turns also raise the volume of forged utterance is the next step, and that step is our derivation rather than a measurement.

There is one fact pointing the other way. What broke Abilene open was, in the end, talking. One person opened his mouth after they got back, and the rest followed in a line. But the place that confession came out was not before the decision — it was after the decision and the implementation were both finished. The phase of frustration and mutual blame that the original records as the fifth symptom is exactly that place. When there is nothing left to overturn, the price of speaking has already come down. This reading is not something the original states; it is what we drew from the ordering of the symptom list.

One more thing is sold in the same bundle as these devices for adding speaking turns: role-played dissent, in the form of a designated devil's advocate or a red team. The verdict that this does not work does not follow from the step just taken. Its grounds are separate. But we should say up front that what those grounds actually measured reaches a neighboring axis, not the axis on which our verdict sits.

Let us separate dissent that arrives because it was assigned as a role — assigned dissent — from dissent that arrives because the person actually thinks that way — genuine dissent. Think of a fire drill. When everyone knows it is a drill, nobody really runs. Same motion, no signal riding on it.

Here is the measurement. One experiment reports that spontaneous dissent outperformed every devil's advocate condition, and that assigned dissent instead made people defend their initial position more firmly. In a field study of 201 employees and managers, genuine disagreement likewise proved more effective than disagreement manufactured by technique at preventing confirmatory information search — the habit of looking only for information favorable to one's own position.

That is the extent of the range. What these experiments measured is the number of solutions, original thinking, and the information-search habit just mentioned. Not "decision quality" in general. One meta-analysis, whose dependent variable may be something else entirely, reports the opposite direction, and whether that is a genuine conflict or a different question cannot be settled before the primary texts are compared.For all three studies, effect sizes and test statistics sat in primary texts we could not reach. The dependent variable these measurements captured also differs from the dependent variable of the judgment we place later. What we mean to speak about is whether mutual misreading of private preferences falls; what the experiments measured is the number of ideas generated and an information-search habit. The measurement came as far as the neighboring axis, and the last step across to the Abilene axis is our derivation.

The Abilene-specific reason is our derivation too, not a measurement. Dissent that arrives as a role carries not one fragment of that person's private preference, so by definition it cannot supply what Abilene lacks: what each person actually wants.

That is not to say the hesitation just before speaking up does not exist. It does. What we deny is not the hesitation itself but a prescription that leaves it filed under individual courage and stops there.

The devices that die and the devices that live do not stand in the same column. What dies is the device that adds chances to speak. The turn around the table, the declaration that "let's be candid here," the allocation of speaking time. And the role-played-dissent device (on different grounds, but on the same side in that it does not bring the price down). What lives is the device that shifts the burden of speaking. A sentence fixed in advance, the same whoever says it. In one line: the turn changes who speaks, and the fixed sentence changes what speaking costs. Where Abilene is blocked is not the turn. It is the price.

That is as far as the original's wording and our own measurements carry. The mechanism of forgery, the recurring cycle, the coding of what the received view prescribes. From here on it is our derivation. If speaking is a problem of price rather than of courage, then an industry already exists that brought that price down by regulation rather than by character. That is why we go outside now. There too a vehicle was heading toward an ending nobody wanted, and the people holding the observations were all aboard it.

One winter night, an airliner was approaching landing. While it circled at altitude over a landing-gear problem, it ran out of fuel. In its probable cause, the accident investigation body's final report wrote that contributing to the accident was the failure of the other two flight crew members either to fully comprehend the criticality of the fuel state or "to successfully communicate their concern to the captain."US National Transportation Safety Board accident report AAR-79-7 (adopted 7 June 1979) · accident of 28 December 1978.

The voice recording shows it was not that no expression occurred. The first officer did make "several subtle comments." They were simply "not given in a positive or direct tone." Direct expression came only after the engines had flamed out.

And here comes the strongest move in the case. It is not that a rule was created. The rule already existed. The airline's flight operations manual had said, from before the accident, that all crew members noting a departure from prescribed procedures and safe practices should "immediately advise the captain."

The duty to speak was already written down, and it failed anyway. So what the investigating body's recommendation aimed at was not procedure. Its wording urges emphasis on "participative management" for captains and "assertiveness training" for the other cockpit crew members. This is, in effect, the origin of what is now called crew resource management. That pair is the physical instance of the line we have just drawn. The chance and the duty to speak were already there; what was missing was the price of speaking.

A device that does not stop at writing down a duty to speak but fixes what to say and how through training and regulation, moving the burden of utterance off individual character and onto the organization. We will call this a speech protocol. It works the way a cabin announcement is always the same sentence. When there is a set thing to say, the person saying it does not carry the weight of it alone. Note, though, that what the recommendation actually wrote reaches as far as training and participative management; "the same fixed sentence whoever says it" is what we drew out as an extension of that. And in the cockpit, what enforced that extension was not an individual but the airline and the regulator. To do the same thing in a meeting room, there has to be a separate party that makes the sentence, adopts it, and enforces it.

The analogy stops in two places. One is the object that failed to be transmitted. What failed to reach the room in the cockpit was the fuel state — an observation, that is, information — while what Abilene discards is preference. So the two rooms overlap on one axis, the price of expression, not on the content of the deficit. The other is hierarchy. The deficit in the cockpit is largely an axis of rank. That the steeper the authority gradient, the more the first officer's speech is suppressed, is a standard concept in that field. This passage, however, rests on a secondary summary whose primary text we could not reach, so it goes as far as circumstantial. Whether the same failure appears in a cockpit of equals is something we did not confirm. Abilene's deficit is a mutual misreading that survives with no hierarchy at all. So that industry's prescription covers only part of Abilene. Devolving authority fixes the rank-type failure, and what remains is Abilene's share.

Three objections are left.

First: wouldn't one anonymous vote settle this? The soft spot is not the tally but what comes after it. A phase returns in which what you learned anonymously has to be said publicly, and an anonymous survey does not lower the price of speaking in that phase. In one line: an anonymous survey is a diagnostic instrument, not a therapeutic one.The effects and side effects of the intervention itself are also still contested — one study reports no evidence of a boomerang effect (the backfire in which telling people the norm hardens their resistance instead) among lighter drinkers in any of its four samples, and a review that set out to test boomerang concerns systematically is still at protocol stage. What can be said now is not "this is solved" but "this moves conditionally." Non-anonymous self-report given in front of an interviewer also inflates on its own. In the 1976 American National Election Studies, 78% of respondents said they had voted, while official turnout was 61%. Self-report had inflated 17 percentage points in the socially desirable direction. The direction survey methodology took to strip that overstatement out was precisely a set of techniques that raise the anonymity of the response.

That limit does not kill after-the-fact surveys across the board. There are two places we use this instrument, and they demand different precision. Used as a diagnostic, one bit of direction is enough. Was the majority against it or not. Used as a therapy, an actual decision has to be reversed on top of that tally, so the size and stability of the distribution are needed too. Since the overstatement was observed in interview responses rather than anonymous ones, what that limit shakes is the second use.

Second: isn't the real culprit hierarchy and power? The objection is that a subordinate speaking against his own preference is not a misreading but an accurate reading — he forges because he knows what the boss wants. That case is conformity, not Abilene. Abilene's discriminating condition is that even the person with the most authority did not want it either. If that does not hold it is not Abilene, and the subset in which the boss actually did want the decision is conformity. What is needed then is devolution of authority — the very direction the cockpit recommendation called "participative management." Hierarchy is not Abilene's cause. It is its amplifier.

On Korean organizations the circumstantial evidence runs to a single line. By one domestic survey of 1,000 employees at listed companies — read through press coverage, since we could not reach the raw data — 75.6%, roughly seven or eight in ten, said meeting conclusions were decided in advance along the senior manager's view. So we use that figure as circumstantial evidence, not proof.

Third: the distinction has no practical payoff, and in the field the two come mixed. That there are overlapping prescriptions does not mean there are no diverging ones. The divergence is exactly what we have just seen. On the groupthink side, a prescription that raises the volume of talk may hit its target — the task there is drawing out suppressed dissent. On the Abilene side, the same prescription misses. There is a more practical layer as well. When the diagnosis diverges, what you have to measure diverges. Suspect groupthink and you measure the frequency of expressed dissent. Suspect Abilene and you measure the gap between private preference and stated preference. When the object of measurement differs, there is a payoff.

So the interventions that work on Abilene are three. First, at the point where a decision appears already settled, asking first: are we each here because we want this? That is not a device that lays on one more chance to speak, but one that puts the agreement itself on the agenda. What increases is not the volume of talk but the kind of question the room has to answer. Second, not leaving dissent as a matter of courage but preparing in advance a fixed sentence that is the same whoever says it. This is where the price comes down. Third, not stopping at tallying preferences but making the result of that tally something the whole room knows together. This is not a device on the price axis but an information device striking the misreading itself, so it stands outside the line we have just drawn. That the second differs from going around the table bears repeating. The turn changes the chance; the fixed sentence changes the price.

Unanimity Is Not the Ground of a Decision. It Is an Alibi.

Back to the original paper's proposition. The paper wrote that the essential symptom of organizations is not the inability to manage conflict but the inability to manage agreement. In its conclusion it nails it down once more, calling that the single most pressing issue of modern organizations.

A decision nobody wanted becomes unanimous not because people kept their mouths shut, but because each of them said the opposite of their own preference. And that agreement passes through the room untested.

Change the diagnosis and the target of the prescription changes. Devices that get people talking more cannot aim at this room's illness. What is needed is not a quantity of dissent but a test of agreement, and bringing the price of speech down by regulation rather than by individual character.

There is one more layer. At the head of this piece we gave a wrongly booked suspect his alibi back. The alibi that remains at the end is a different one. A decision everyone assented to has no proposer. The person who would answer, under their own name, "I'm the one who said we should do this," disappears from the room. Forged unanimity goes one step further than that. It erases the dissenters too. Everyone was privately against it, so everyone stands by that decision holding the alibi "I was actually against it," and the mutual blame that comes when the decision turns out wrong is not an accident but an exit the structure prepared in advance. So unanimity functions not as the ground of a decision but as a form that erases whoever would answer for it. The consequences are paid elsewhere. The floor that implements it, the successor who inherits the decision, and the next meeting pay them. And since the last link — leaving the room with the underlying problem untouched — remains exactly as it was, the same chain runs one more turn, with greater intensity, in the next room. This verdict is not something the original states in those terms. What the original carries is the mechanism, the aftermath, and the recurrence; "whoever would answer is erased" is our reading standing on top of that.

Back to the car at the top. Nobody had set the direction. And the car was already moving.

Whether that car is parked in our own room is something we can measure. The test to run against your own organization is this. After the decision is made, ask about private preferences anonymously, and see how far the assent the room gave publicly diverges from what people actually preferred. A wide gap points to Abilene; no divergence means it was not Abilene. Whether the person holding the authority was against it too is an exclusion condition laid on top. If it was the decision he wanted, that is conformity, not Abilene. That this condition collides with anonymity, we grant. There is only one person in that seat, so in a small group the response is identifiable by elimination, and on top of the observation that after-the-fact self-report inflates, the direction of the bias is open both ways — if it runs toward "I was against it too," the gap inflates; if it runs toward "I was for it too," the gap becomes a lower bound. The direction itself is something our material does not nail down. This suspicion applies just as much to the starting point of our own story. The four "I didn't actually want to go" statements are also self-report immediately after a bad outcome. That is why the verdict in the earlier section was built not on that confession but on a comparison against the wording of the original's symptom list, and what this bias shakes is the precision of this test, not that verdict.

If the test points toward Abilene, the next question is what to attach there. We will place one judgment of our own as well. The dissent devices organizations attach — a designated devil's advocate and a red team — do not work on Abilene-type failure.

This judgment breaks if a measurement shows that under an assigned-dissent condition, mutual misreading of private preferences falls as far as it does under genuine dissent.

The one thing to say at the next meeting can be fixed in advance. Whoever is running it reads the same sentence right before the agenda is confirmed: "Before we move on, let's check whether anyone privately objects to this item. We won't take answers out loud — everyone writes one down, and we'll look at them together." The second sentence is there because of an asymmetry in price. A fixed sentence lowers the price only for the person asking; the price of public utterance for the person answering is unchanged, so taking answers out loud turns that moment back into a channel for forgery. So the answers are collected in the manner of the third intervention, and this is the minimal form a single facilitator can use within their own authority before the organization adopts a protocol.

The Abilene paradox is not something that happens only because people fall silent. It happens because each person, misreading what everyone else holds inside (pluralistic ignorance), voices assent that runs against their own wish (forged assent). In a room like that, unanimity is not evidence of agreement. It is the alibi covering the fact that nobody wanted the decision.