Extraordinary claims require extraordinary evidence. But claims of sexual harassment, abuse, assault, and rape are not extraordinary. They are depressingly ordinary. So the level of evidence we should need to believe a claim about sexual harassment, abuse, assault, or rape is substantially lower than the level of evidence we should need to believe a claim about, say, Bigfoot.
This is straight Bayes — since the prior for rape is higher than the prior for Bigfoot, it requires less evidence to raise our credence above 0.5 in any given case of a claimed occurrence. In the comments, one person points out the connection to Bayes, in part remarking:
“Bayesian updating” is a good method for using evidence rationally to change your mind. If someone requires extraordinary evidence to believe a depressingly common event, they are not being rational.
In response, another commenter, apparently triggered by the mention of Bayes, goes on a tirade about Michael Anissimov and Less Wrong being misogynistic. This commenter selectively quotes Anissimov regarding IQ, Larry Summers, and “political correctness” — a quote that (at least, out of context) sounds pretty damning, as silly as it would be to infer from Anissimov to Less Wrong. When I read this comment, I winced; my reaction could be stated something like this: “Aw, jeez. LW does not need a squabble with the FTB folks and the progressive-feminist end of the skeptic movement. Hardly anyone can speak both groups’ languages. If a conflict happened, both groups would be worsened by the polarization.”
That is, I was (for just a moment) ① willing to take the tirade-poster as representative of “the FTB folks” and ② predicting the tirade-poster to be a catalyst of an intertribal conflict between two groups I’d prefer to see reconciled.
But I kept looking … and it turned out that the tirade-poster was a troll, or at least a crank, on FTB and was readily recognized as such by the folks there. In other words, my initial expectation of a brewing political clash was flat wrong — and I had (albeit momentarily) taken the words of a deviant, undesired member of a group as indicative of that group!
(Trigger warnings: mention of rape, harassment, and hostile criticism of Less Wrong.)
A lesson on politics as mindkiller —
There’s a thread on Greta Christina’s FTB blog about standards of evidence in discussions of rape and harassment. One of her arguments:
This is straight Bayes — since the prior for rape is higher than the prior for Bigfoot, it requires less evidence to raise our credence above 0.5 in any given case of a claimed occurrence. In the comments, one person points out the connection to Bayes, in part remarking:
In response, another commenter, apparently triggered by the mention of Bayes, goes on a tirade about Michael Anissimov and Less Wrong being misogynistic. This commenter selectively quotes Anissimov regarding IQ, Larry Summers, and “political correctness” — a quote that (at least, out of context) sounds pretty damning, as silly as it would be to infer from Anissimov to Less Wrong. When I read this comment, I winced; my reaction could be stated something like this: “Aw, jeez. LW does not need a squabble with the FTB folks and the progressive-feminist end of the skeptic movement. Hardly anyone can speak both groups’ languages. If a conflict happened, both groups would be worsened by the polarization.”
That is, I was (for just a moment) ① willing to take the tirade-poster as representative of “the FTB folks” and ② predicting the tirade-poster to be a catalyst of an intertribal conflict between two groups I’d prefer to see reconciled.
But I kept looking … and it turned out that the tirade-poster was a troll, or at least a crank, on FTB and was readily recognized as such by the folks there. In other words, my initial expectation of a brewing political clash was flat wrong — and I had (albeit momentarily) taken the words of a deviant, undesired member of a group as indicative of that group!
Congratulations, you avoided stepping on a landmine!
Is there a name for the bias “if a person A is commenting on a forum X, then person A is a representative of the forum X”?