Thanks for the detailed comment. I omitted details in order to keep my post short, and get the main point across.

I believe that the IQ tests that Terman and Hollingworth were using were effectively scaled differently from modern IQ tests. They may have corresponded to "mental age" as opposed to "standard deviations. In particular, they discuss IQ scores of 180, and there definitely aren't enough people who are 5+ SD above the mean to get reliable scores in that range.

Putting that aside, there are genetic factors other than IQ alone that play a role in intellectual and emotional development See my discussion of aesthetic discernment here: it hasn't been established as a valid psychometric construct, but I have very high confidence that that's simply because psychology researchers haven't investigated it carefully. If one is 2.5+ SD above the mean in each of IQ and aesthetic discernment, one is going to be extremely isolated. I think that that's what one is seeing with someone like Scott Alexander.

Relatedly, Benbow and collaborators also found that children who scored high on verbal and not math have greater social maladjustment than those who score high on math and not verbal (don't have the references immediately on hand, can dig them up later if you want.)