That's probably because I wasn't writing a rebuttal per se, but a clarification. The distinction is important because, although he's incorrect to say that we have no means of identifying if somebody is a women besides them honestly self identifying, we also don't know if we have found all the different means by which a person may legitimately be considered a women. We can positively ID a person as a certain gender, but we can't negatively ID them as not a certain gender.
So I guess the direct answer to the question about if we can identify a woman outside of a person self identifying is "sometimes". Certainly, allowing people to self identify is easier than forcing them to take a bunch of tests and MRI scans only to get results ranging from a "yes" to "maybe"
Ah, I think I see where the confusion is.
The "positive" or "negative" identification is in relation to what the person claims. So if a person claims to be a woman, we can use science to determine either "yes this person is definitely a woman" or "maybe this person is a woman." What we can't do is say "no this person definitely isn't a woman" because it's possible there is some factor we haven't identified or discovered yet which would validate their identity.
Edit to add: actually, I can think of ONE test to prove that somebody who says they're a woman but isn't: gender transition to the gender they claim to identify as. Cisgender people usually get severe gender dysphoria if they attempt gender transition. I would consider that proof positive that they aren't the gender they claim to be. However, subjecting somebody to such an experiment without fully informing them if the risks and/or against their will is massively unethical which, imo, disqualifies it for the purposes of this conversation. But technically it's an option.