Stephen Hsu is a crank who doesn’t have any expertise whatsoever in genomics. His doctorate is in physics and he was hired by the Beijing Genomics Institute based on his experience as an administrator.

He uses his affiliation with BGI to give credibility to what he puts out on his crackpot “race realist/HBD” blog Information Processing, which caters to an audience of mostly neo-Nazis.

Good rule of thumb is that if you’re getting your science from the same place as people who post on Stormfront, it’s probably not very good.

The reality is that we can currently predict roughly 4 percent of variation in IQ based on polygenic score, and that’s probably not going to change anytime soon. While the general heritability of IQ is about .50 based on twin studies, heritability is not fixed and at the lowest socioeconomic status, genetic contribution to intelligence approaches zero (See Turkheimer’s research on gene-environment interaction)

There are also lots of other factors that complicate things. For one, most of the GWAS studies are done on European populations, so the results don’t necessarily transfer over to other populations. (See Danielle Posthuma, the foremost authority on genetic intelligence research).

Right now we only know of about 50 or so candidate genes linked to intelligence, which is highly polygenic. Ballpark estimate is that there are more than 10,000 that play a role, so trying to predict intelligence from just that tiny handful is dangerous and stupid.

I know you want to write an article that sounds exciting about how we’re living in Gattaca and the future is here, but this is serious, so please report responsibly. Thanks.