Your point seems to be that it doesn't matter if a model is noisy *if* we handle the noise. The whole point of this chapter is that we *don't* do that - that acting as if there is no noise in the model is, in fact, the source of a whole lot of problems. (Not least IGM - we literally *impose* a binary).

A broader point here is the use of "biology" as a proxy for reinforcing social order.

The reification of sex *as* gender i.e. the idea that there are two classes that you can be born into, and that which one you are in governs your nature, is where Stock moves onto from this. Lots of people have seen problems with this view, hence, broadly, feminism.

Likewise, the insistence that everyone clearly fits into a binary is used to argue that sex is immutable (and that therefore transitioning should be disregarded or disallowed). You can see some broader throughts on that in https://medium.com/@kim-hipwell/on-truly-changing-sex-a7770e903810

You're claiming that "it's absolutely not the case that an octonary, quintary, or unary conception is as good as a binary one" - even though there are societies that use those distinctions. I don't see a genuine case for that here.

As far as I am concerned, the idea that language is/should be shaped by some abstract notion of communication efficiency seems a total red herring. When talking about inflections, you're conflating gender classes with *linguistic* gender (which also varies wildly across language, and blurs with noun classes of which there can be 100s in a language).