This is a pretty good book. It covers a lot of basics, and gets almost everything right. Twins, adoptees, genomics are given ample attention. Mitchell is very interesting in developmental randomness, so this point gets a lot of coverage. Good! Not many people accept this. The book is fine on sex differences too, and even the part on racial classification is decent.
The main bad points relate to his engaging in dumb anti-hereditarian arguments, like those in his article. He seemingly thinks that traits controlled by more variants are harder to evolve group differences for. This may surprise the reader, since one of the best known genetic differences is height, which shows massive international variation, and even larger between sex variation, but is similarly influenced by 20k+ variants. How so? No answer in the book. Nevertheless, it’s a new argument, so I guess it needs a simple simulation study to disprove.
He is also following the typical value-fact, no relationship model, and thinks real differences somehow do not mean using stereotypes is reasonable. Any Bayesian will have to give him ‘the talk’ on priors.
Oddly, his prior writings on epigenetics were not adapted for the book. They are here and here, 2013, so old:
Overall though, this was better than expected. I recommend for beginners.
Some specific commentary, for Twitter, since it is not too serious:
