[ben_w]

I remember hearing about the variance thing ages ago. Back when I was young enough and naïve enough to trust statements said in official voices without critically assessing them.

With the caveat that IQ tests scores are now provably something one can learn to be good at (because LLMs do much better on public tests than private ones), was the claim about variably actually justified at the time, or was it nonsense even back then?

[tptacek]

I'm not touching the variability thing with a 10 foot pole except to say that the further out on each extreme of the IQ "scale" you go the less reliable the scores are. The whole idea of using IQ as a ranking of ability rather than a diagnostic tools is bogus. I do think it's clear now though that Summers was simply being a misogynist (you lose the presumption of good faith when you disclose that you'd been lying all along.)

[ben_w]

Oh indeed, on all counts. I'd just like to know if it was purely his own BS, or the reproducibility crisis.

(I don't know why I'd like to know, thinking about it at a meta level…)