Why Smart People Agree With Things They Don’t Believe
WHY SMART PEOPLE AGREE WITH THINGS THEY DON’T BELIEVE
Picture a meeting. Doesn’t matter what kind.
Somebody floats an idea. A couple of people nod. And you feel your own hand start to go up — even though thirty seconds ago you weren’t sold on it at all.
Nobody pressured you. Nobody said anything to you. You folded anyway, and you got to watch yourself do it in real time.
There’s a name for that, a number attached to it, and — the part I can’t stop thinking about — a few moments in history where the exact same wiring took over entire towns.
THE STUDY THAT CAUGHT IT
In 1951, Psychologist Solomon Asch puts one real participant in a room with seven actors. Everyone’s told it’s a vision test.
The task: here’s a line, here are three other lines, which one matches.
It’s not hard. It’s not close. There’s nothing to debate.
Everyone answers out loud, in order, and the real participant sits near the end of the row. First couple of rounds go fine. Then one of the actors gives an answer that’s wrong. Not subtly wrong — visibly, obviously wrong.
The next actor says the same wrong thing. Then the next. Then the next.
Then it’s your turn.
On the rounds that counted, people went with the group’s wrong answer about a third of the time. Roughly three in four caved at least once across the full set. About one in twenty went along every single time.
And the control group — same lines, no audience, just write it down — got it wrong less than one percent of the time.
Less than one percent. That’s the actual difficulty of the task. Everything above that number is the room.
Here’s what gets me, though. Asch talked to his participants afterward, and they’re not describing confusion. They weren’t fooled. They could see the lines perfectly well. What they describe is closer to panic — that something has clearly gone wrong with either the room or with them, and in the two seconds they have, they can’t work out which.
So they split the difference and said the wrong number out loud.
Two details that are better than the headline
It doesn’t take many people. Asch messed with the group size, and the pressure climbs fast and then just flattens out. Three or four people saying a thing lands about as hard as a dozen. You don’t need a mob. You need a small table.
And unanimity is the active ingredient — not accuracy. This is the whole essay, so stay with me.
When Asch added one dissenter who gave the correct answer, conformity dropped from about 32 percent to around 5.5 percent. One person. Not a speech, not a coalition. One person saying the obvious thing out loud.
Then he ran the better version. He had a dissenter give a different wrong answer. Still wrong. No help to you at all on the actual question of which line is longer.
Conformity still fell — to around nine percent.
Read that again, because it means the thing holding that room together was never the correct answer and never the strength of anyone’s argument. It was the unbroken surface. The second one person says something different — anything different — the room stops being a wall and turns back into a handful of individual people with individual opinions. And your opinion is suddenly just one more in a pile instead of the lone defect.
You don’t need to be right.
You just need to stop the room from being unanimous.
THE HISTORY THAT GOT THERE FIRST
Dancing mania is not a legend. It’s a documented, recurring pattern across medieval Europe — outbreaks in Erfurt, Aachen, Trier, and most famously Strasbourg in 1518, when a woman known as Frau Troffea walked into the street and started dancing and didn’t stop.
Within about a month, roughly four hundred people had joined her. Some danced themselves to death.
Historian John Waller’s explanation is stress-induced mass hysteria, and the setup has two parts. One: a population already wrecked by famine and disease. Two — and this is the part people skip — a regional superstition already sitting in every head in the city, about a saint who could curse you with compulsive dancing.
Nobody in Strasbourg looked at Frau Troffea and saw a woman having a breakdown. They saw the thing. The known thing. Happening on their street, to someone they knew.
Which made it available to them too.
And then the city did the single best-worst thing in this entire story. They decided the cure was to let the affliction run its course, so they hired musicians to help it along. Cleared space. Built a stage.
It got worse. Obviously it got worse. They took something that was spreading by being visible and gave it better lighting and a band. The outbreak only faded after officials banned public dancing outright.
Hold onto that stage. It comes back.
Get Sketches Of The Mind’s stories in your inbox
Join Medium for free to get updates from this writer.
Salem ran the same thing through a courtroom
- A handful of girls’ accusations spread until roughly 150 people were sitting in jail. Nineteen were hanged.
The reason it was nearly unstoppable is that the court accepted “spectral evidence” — testimony that someone’s spirit appeared in a dream or a vision. No physical evidence required.
Think about defending yourself against that. What’s your alibi? You were asleep in your own bed with witnesses, and that isn’t even a contradiction, because the whole claim is that the part of you doing it wasn’t in the bed.
The accusation was built so that denying it looked like more evidence.
And look at who got named first. Tituba, a woman enslaved in the household where it started. Sarah Good, poor enough to be begging door to door. Sarah Osborne, elderly, unwell, already in a property dispute with a prominent family and already skipping church.
Not a cross-section of the village. The cheapest possible people to accuse — the names where saying it out loud cost you nothing, because nobody powerful was going to push back and nobody was going to make it awkward on Sunday.
Once those names went through the process and the process held, the circle widened. By the end it reached a former minister and the wife of a wealthy merchant, because the machinery had a track record by then.
It took the clergyman Increase Mather publicly arguing that spectral evidence shouldn’t count on its own to stop it. He wrote that it were better ten suspected witches escape than one innocent person be condemned.
WHY THE ROOM BREAKS, EVEN A GOOD ONE
Irving Janis named this failure mode groupthink in 1972: a tight group’s drive toward agreement quietly outrunning its drive to get the answer right.
His best contribution is a job nobody applies for. He called them mindguards — the person who, with completely good intentions, filters the inconvenient thing out before it reaches whoever’s deciding. Not out of malice. Protectively. Morale’s fragile, the boss has already committed, so the doubt gets softened, or delayed, or just not forwarded.
You’ve met this person. You’ve probably been this person. I have, this month, and the worst part was that it felt like being helpful.
His case studies were real disasters — the Bay of Pigs, the warnings ignored before Pearl Harbor — and the 1986 Challenger launch fits the same shape decades later. Engineers warned explicitly that launching in the cold was untested and unsafe. They were overridden under schedule pressure. The information was in the building. The room had developed an immune system against it.
One honest correction, because this gets flattened a lot: a major meta-analysis found that group cohesion on its own doesn’t reliably predict bad decisions. What mattered was the type of cohesion. Groups bonded through personal liking did worse under pressure than groups bonded around shared pride in the work. It got worse as groups got bigger and worse again when a directive leader was already steering the outcome.
So it isn’t closeness. It’s what the closeness is loyal to. When the relationships are the thing being protected, every piece of bad news is also a small betrayal. And it didn’t stay in the boardroom.
A 2022 study on cyberbullying found bystanders join social-media pile-ons for two specific reasons: losing their sense of individual, accountable identity inside an anonymous crowd, and a mental move that reclassifies piling on as fine because everyone else is doing it.
Same wiring as the line study. Same wiring as the mindguards. Same wiring as four hundred people dancing themselves to exhaustion in 1518.
Different outfit. Identical mechanism.
SO WHAT ACTUALLY HELPS
Not “think harder.” Asch’s most conforming participants weren’t stupid, and the Challenger engineers weren’t silent.
Here’s the frame that made all of this click for me. Those participants in that chair weren’t confused about the lines. They were doing math — fast, unconscious, and mostly correct math about cost. Being right in that room was worth almost nothing. Nobody hands out prizes for line accuracy. And being the one person who contradicts five strangers out loud costs something real: the next twenty minutes of being looked at.
So the room isn’t voting on what’s true. The room is voting on what’s expensive to say.
Which re-reads everything above. In Strasbourg, dancing was cheap — it was the expected thing a cursed body did. In Salem, accusing was cheap and doubting was lethal. In a 2022 comment section, joining in is free and standing apart is not.
Nobody in any of those rooms was voting on reality. They were reading the price tag.
And that math runs a little wrong in one specific direction. It overcharges you for speaking and undercharges you for going along, because it prices a social cost it can feel and ignores a real cost it can’t.
The one variable that breaks this across every example here is a single voice willing to disagree out loud while it’s still cheap to do it. And it doesn’t have to be the correct voice. Nine percent, remember.
If you’re the one running the room, there’s an even cheaper version, and it costs you one sentence.
Stop asking “everyone good with this?”
That question asks for hands. It makes agreement the default and turns disagreement into something you have to interrupt the room to produce — you’ve built the unanimous surface with the phrasing, before anyone’s thought about the decision.
Ask for the objection instead. What’s the strongest case against this? Who sees it differently? What are we going to feel stupid about in three months?
Same room, same people, same information. But now speaking up is the thing you were asked for instead of the thing you’re risking.
That’s the whole trick. Strasbourg built a stage and made it spread. You can build the opposite thing with one question.
If you want the fuller version — the complete list of groupthink’s eight symptoms, more on how the Salem trials actually ended, and the group polarization research on why discussion makes opinions more extreme instead of less — the full episode is live now on Sketches of the Mind, wherever you get your podcasts, with an expanded companion piece on Substack.
New teaser blog every Wednesday
Listen on Spotify: Sketches Of The Mind Podcast
Read the full blog on Substack: Sketches of the Mind
