If someone looks like a man, and sounds like a man, and I can model him as a man in the range of circumstances I interact with him without making any grievous prediction errors, then I call him a man even if he's technically a female on testosterone—not as a favor, but because it genuinely seems like the best word to describe what I'm seeing. Futhermore, if I don't perceive someone as a man, but everyone else does (or speaks as if they do), I'll usually go along with the majority's language usage for game-theoretic reasons (explanation: http://unremediatedgender.space/2019/Oct/self-identity-is-a-schelling-point/). But if he further demands that I'm not even allowed to use the phrase "female on testosterone" to describe what I think is going on in this situation, then we have a problem.
+Trump—
+> devastating in their percentage and power of destruction
+
+Like I said, it would seem the disagreement centers around how robust the trait-clusters are, and how useful it is to be able to use simple language to talk about the overwhelming-majority case, and I'm wary of these standards being selectively varied for political reasons that, however well-intentioned, actually in-practice interfere with our collective ability to make sense of the world.
+
+When I say "selectively varied", I mean that the same standards aren't being applied evenhandedly to different topic-areas. For example, when talking about the anatomy of the hands, you might end up offhandedly saying something like, "Humans have ten fingers." I think it's good for people to be able to say "Humans have ten fingers" in most contexts without it being construed as a denial of the facts that polydactyly exists and that some people lose fingers in accidents. (It's true that polydactyly exists! And it's true that some people lose fingers in accidents! I enthusiastically affirm both of those facts! But I think it's also true that ... you know what I meant.)
+
+This is also my rationale for wanting "Men are the ones with penises" to be considered "true" in most contexts. The claim isn't that trans men don't exist; the claim is that adult human males are the prototypical thing that trans men are imitating (to the limits of available technology), such that it's useful for many unmarked usages of "man"/"men" to be understood to refer to the prototype unless there's some contextual reason to think otherwise.