Four Walls Protecting Your Common Sense

When a whole room denies what people can plainly see, those who answer in private mostly hold out: the eyes were never the problem. Gaslighting and bullshit make it too costly to say what you see. Our defence is structure: four walls, the last a culture where speaking truth beats fitting in.

By Scott Volk · · 9 min read

Four Walls Protecting Your Common Sense

One of the first people to look at our investor deck put it to us straight: “Fine, cognitive warfare is a threat to Canada. So how exactly will Missionloops help Canadians regain their common sense? How do you teach someone to detect gaslighting and bullshit?” He asked important questions; this briefing is our answer.1

His second question has the more surprising answer: put the facts in front of people and most of them already see through the bullshit. What fails under pressure is the nerve to say what they see, out loud, against a unanimous crowd. That failure is among the most replicated findings in psychology, and it comes with instructions.

Gaslighting has a laboratory model. Psychologists reserve the word for the long, intimate campaign of making someone doubt their own mind; in 1951 Solomon Asch built the sixty-second version. A subject sits with seven or eight others and judges which of three lines matches a target line. The task is easy: alone, people get it right more than 99 times in 100. But everyone else in the room is an actor, answering aloud, confidently, and wrong. The whole room insists on something the subject can plainly see is false. That is gaslighting, distilled: a confident consensus against your own eyes.

WHICH MATCHES? Target A B C

Gaslighting works. In the original study, 32 percent of subjects’ answers went with the plainly wrong room. A later, larger study found 36.8 percent, and about three quarters of people went along at least once.2 Asch also varied how many people stood against the subject. One wrong stranger barely moved anyone. Three delivered almost the whole effect. Fifteen added nothing that three had not already done. In Asch’s rooms the crowd’s gaslighting power topped out at three people: the pressure comes from the appearance that everyone present, independently, agrees, and that appearance is complete at three. Your feed can show you a unanimous room at close to zero cost. Manufacturing the appearance of independent consensus has never been cheaper.

Your eyes were never the problem. The most important variation is the least famous one: the private answer. When Asch let people write their answers down instead of announcing them, errors fell from 36.8 to 12.5 percent in the same later study. Same room, same lies, same eyes. The judgment was intact the whole time; what people could not afford was saying it out loud, alone, against everyone. Interviewed afterwards, most subjects underestimated how often they had folded. They did not experience themselves as giving in. That is what gaslighting does: it does not blind you, it makes your own eyes feel unaffordable.

The eyes survive only when there is something to see. Asch’s task had a right answer sitting in plain sight; that is why the eyes survived and only the voice gave way. Muzafer Sherif ran a different experiment in 1935, one with no right answer in it: a stationary point of light in a dark room, which appears to drift when you stare at it. Asked how far it moved, groups settled on shared numbers, and each person quietly took the group’s number as their own, keeping it even when judging alone later. Where reality gives no answer, the group does not just pressure your answer; it becomes your answer. So the room’s power depends on the question: where you can see for yourself, it captures your voice; where you cannot, it captures your eyes. Wall three, below, is built for that blind spot. Real stakes can turn a question you cannot see for yourself into one you can.

So common sense is neither knowledge nor cleverness; it is the practical ability to keep trusting what you can plainly see when a room disagrees. It is not lost. It is outvoted. And every lever the lab found for getting it back is structural: privacy, an ally, a norm. We engineered our defence the same way: four walls, built in depth.3 Each faces a different attack, and what slips past one meets the next.

Wall one: write down what you see, before the room votes. A judgment committed in private is a different object from an impression floating in your head. Asch measured the private half: answers written down instead of spoken aloud largely escaped the room. Sherif measured the sequence half: people who sized up his ambiguous light on their own first held their own line measurably longer than people who met it inside the group. On our platform this is the mission file: the operator, the person working the problem, writes their own read of their situation, in their own words, before anyone else weighs in. That written read is the anchor, ground to argue from. An unwritten read is negotiable, and the room will negotiate it.

Wall two: one real witness. Asch’s strongest result was a single partner who answered correctly: surrender collapsed from the original study’s 32 percent to 5.5 percent.4 One genuine ally against a wrong room hands most people their eyes back. The everyday version is a question everyone already knows: are you seeing what I am seeing? The room’s spell rarely survives it. On our platform the person the operator asks is the guide: someone they already trust, working the situation beside them.

Where does one ally’s power come from? Asch found that even a dissenter who was wrong in a different direction than the room freed people almost as fully as the partner who answered correctly. So the base of the power is independence itself, a genuine break in the room’s unanimity. The witness’s standing, their visible ability to judge, multiplies it. Allen and Levine measured by how much, in one of psychology’s stranger follow-ups. They gave subjects an ally who said all the right things but had been theatrically discredited in advance, with bottle-thick glasses and a failed vision test. Even that wreck of a witness still cut the surrender, at about half the strength of a credible ally. Dissent is hard to kill: strip a witness of standing and half the protection survives.

A serious attacker learns the same lesson from the same data: smearing the witness only cuts the protection in half; to finish the job the room has to remove the witness. Asch measured both ways of removing one. When the partner merely left the room, the protection partly outlasted him. When the partner switched sides and joined the majority, errors leapt to 28.5 percent, nearly as bad as facing the room alone. A turned witness does more damage than a missing one. The platform’s answer is that it never assigns the witness: the operator chooses their own guide, someone whose judgment and character they already trust.

Wall three: let reality vote. The first two walls still depend on people, and people can be worked on. Reality is the one witness you never have to vet; it cannot be an actor, and it does not care what the room thinks. And this is the wall built for bullshit, the other half of the question. Bullshit, in Frankfurt’s strict sense, is not lying: the bullshitter is not tracking truth at all, and what they say need not even be false. That is why no lie-detector fixes it; there is often nothing false to catch. What bullshit struggles to survive is cost. It thrives exactly where being wrong costs nothing, and real stakes make it expensive, fast. Sherif’s drifting light had no answer to give; a real problem does. Move the question onto ground where reality can answer, and it starts answering.

How do you make reality vote? Pick a problem from your own life, one where you are close to the evidence and have skin in the game. That grounding is the point. On your own ground your eyes work, and nobody can outvote you about what you can check for yourself. Then act on the problem, and notice you are already in a loop: you read the situation, form a picture, act on the picture, and the world answers. Boyd named that cycle the OODA loop, and the missionloops training is built on tightening it. Each pass around the loop throws your picture of the situation against what actually happened, and each mismatch is a chance to fix the broken part. Loop by loop, the picture gets closer to how your local reality actually works. Boyd’s school has a blunt rule for this: bad news is the only kind that does you any good. A failed plan is a correction no room got to vote on.

But the picture that made the plan will fight to survive its correction. Hearing it hurts, and the reflex is to explain the failure away. The practice treats that reflex as a target. A major part of the training is the habit of catching it in yourself and reacting to it on the spot. And the guide stands against that reflex from outside, a witness at exactly the painful moment a failed plan goes looking for a comfortable explanation.

Wall four: a second room. Asch’s room worked because the subject had no way to check who was real; every voice but his belonged to an actor. The feed is that room at scale, and you cannot fix it from your seat. A crowd cannot be vetted, and the manipulation is not going to stop. So stop arguing with the rigged room and build a second one, with a door you control. Keep it small: you and one other person, someone you vetted yourself. Not an actor, and not someone the crowd has already captured. Inside it, the whole practice lives for as long as the problem does: your own read, one real witness, the loop testing your picture against reality. On our platform the second room is the operator and the guide they vetted for themselves. A room of two can go wrong together. That is why wall three lives inside it: reality never joins the consensus.

A second room does not have to stay a room of two. An operator can have more than one guide, a guide can work with more than one operator, and the rooms overlap into a network. What a network of rooms shares is a culture: a way of working built around the loop and the habit of hearing a plan fail. And a culture is the strongest wall in the set. The evidence is an anomaly the conformity literature still argues about: in 1980, British students from engineering, chemistry and physics faced Asch’s rigged room and surrendered once in 396 trials.5 The interviews found the same pressure and the same doubt as 1951; what changed was the direction the pressure pointed. For those students, going along with an answer they believed wrong was what would make them look foolish. Our read is that they answered to another culture: the truth-seeking one the sciences grow by accident. It walked into the rigged room with them. We are building that culture on purpose.

The claim we intend to test. One prediction here is ours and unproven: people who consistently make hard decisions under real pressure, with these four walls in place, should become measurably harder to move with manufactured consensus. The Asch paradigm is cheap and standardized, which makes this a testable claim, and testing claims like it is what our research programme is for.

So how will we help Canadians regain their common sense? Not with a detector. With a room of their own where their eyes get the first word, one witness who can actually judge, problems real enough that the world itself keeps score, and a culture where speaking truth beats fitting in. Common sense was never lost. It was outvoted. Our walls give it back a room where its vote counts.

The research behind this

  1. This briefing reads like research on purpose. The questions deserve the evidence, and since we are building a defence research instrument, the discipline comes with the work.

  2. The lab numbers are not constants of nature: Asch himself moved the rate between 26 and 53 percent by changing the trial structure, so we make no population forecasts from them.

  3. The walls are maintained, not installed: witnesses can be discredited, rooms decay, written anchors go stale.

  4. A witness restores independence, not correctness: in a recent study of people advised by AI panels, a single dissenting adviser roughly halved deference to the majority without improving anyone’s accuracy. Freed judgment still has to be good judgment, which is what the training itself is for.

  5. One cohort of 33, so we hold it as our reading, not a law. The same paper recovered Asch-level surrender in two other 1980 groups for whom dissent carried real cost; that sharpens the lesson rather than blunting it: the rate follows the social conditions in the room, whatever the calendar says.

You might also like...