There's no doubt Strumia did a dumb, and while he may kind of have a point regarding the variability hypothesis, there's actually more to it.
It comes down to chromosomes. XY individuals have a single X chromosome while XX individuals have two (duh).
The X and Y chromosomes have a small section that pairs one another (the pseudo-homologous region), but most of their chromatin is different.
The Y is mostly made of genes required for the architecture of the testicle, whereas the non-homologous part of the X contains genes that are implicated in many cellular processes. Humans need at least one chromosome to be viable.
In XX individuals, to avoid the over-expression of the dedicated X genes (those that are not part of the pseudo-homologous region), one X chromosome is deactivated. This happens at random, during the early embryonic life.
As a consequence XX individuals have roughly one half of their cells that expresses the alleles of their father's X and the other half expresses those of their mother's. XY individuals only express the alleles of their mother's X.
Depending on what an allele does and how an allele pair interacts, this can have different consequences.
The canonical example is color vision where XY individuals are more often affected by color perception issues than at a much higher frequency than XX individuals. Also, some XX individuals exhibit tetrachromacy, i.e. for them color is a 4d space rather than the RGB 3d space most of us are used to.
For other parameters like body size, having distinct alleles results in a phenotype that is in between. After normalization, XY individuals have a larger variance in body size than XX folks, since the latter can fall into more intermediate buckets.
Sex chromosomes in birds are the other way around. Males have a WW pair while females have a WZ pair. In birds, the variance for body size is larger in females.
This leads us, at long last to IQ, whose variance is larger in the XY group than in the XX one (with an identical mean in both groups). This may lead one to think that it makes it normal to have more XY folks in academia for mathy/brainy topics like physics (since there will be more of them in the upper tail of the distribution), but it ignores several things:
- IQ is an imperfect measure, and we may be ignoring people whose faculties are, like tetrachromacy, out of reach for XY folks, who designed the tests.
- XY folks in the upper tail of the distribution for a given parameter (say, IQ) have nevertheless a higher probability to be catastrophic in other dimensions (when compared to XX folks). This is relevant because science is becoming more and more of a collective endavour where social abilities matter as much as raw intelligence. It looks like Strumia sucks at that game.