The "truth-seeking community in a belonging costume" claim hangs on LessWrong: approved framings, canonical sequences, jargon that started as shorthand and became a membership badge. Questioning core concepts gets social pushback.
A research program with a core looks like that too. Shorthand that stays shorthand is not a loyalty test. Pushback on a core is what you get when people think the challenge is wrong.
The specimen needs an observation that isn't "they have a canon."
The tell isn't the terms. It's pushback that does not care whether the challenge is right, plus the people who still wanted to update leaving.
A canon that updates is still truth-seeking. A canon you get punished for touching even when the group later treats the challenge as correct is the costume.
From outside you see pushback. You don't see that the challenge was right and still punished.
The leftover is an outer sign: a correction the group treated as true, then punished the person for saying it. If that row isn't on the page, "costume" is just "they have a core."
Hypothetical: someone updates a core claim in public, the update sticks, and the updater still loses status for having made it. That would be costume.
An update treated as error is truth-seeking doing its job. I don't know which one the page is pointing at with LessWrong. Stopping there rather than filling in a case.
Evaporative cooling is doing different work in the same paragraph.
People leaving is compatible with the remaining group getting stricter about method, not just more club-like. Cooling can be selection for a harder standard. The page reads it as belonging-selection because the leavers are labelled "truth-seekers" and the remainers "conformists." Those labels are the conclusion.
Then you need a case where social punishment went up while the method got looser. If both rose together, both stories still fire. The specimen does not separate them.
"Truth-seeking was the tribal mythology" is doing the identification for every specimen.
Any community whose founding story isn't what it currently maximizes looks like a costume. Punk that became a dress code. Journals that select for publishable results. The pattern is: name the founding myth, point at current practice, call the gap belonging.
A passing case would be a group that keeps membership while paying people who correct it. The page says those fall outside the condition. They are also the only thing that would show the costume reading isn't automatic.
So the specimens are all failing cases, and the passing case is defined out as a different condition.
That can be the right cut. It means LessWrong, journals, and punk cannot show the mechanism against a near-miss. They can only illustrate a condition already granted. The costume language still sounds like a verdict on those groups, not a conditional.