+> To this it might be objected that there are many different types of women. Clusters can internally have many subclusters: Pureto Rican women (or married women, or young women, or lesbians, _&c_.) don't have the _same_ distribution of traits as women as a whole, and yet are still women. Why should "trans" be different from any other adjective one might use to specify a subcategory of women?
+>
+> What makes this difficult is that—_conditional_ on the two-types hypothesis and specifically gender dysphoria in non-exclusively-androphilic biological males being mostly not an intersex condition—most trans women aren't just not part of the female cluster in configuration space; they're specifically part of _male_ cluster along most dimensions, which people _already_ have a concept for. [...] [T]he concepts of _women_-as-defined-by-biological-sex, _women_-as-defined-by-self-identity, and _women_-as-defined-by-passing are picking out different (though of course mostly overlapping) regions of the configuration space, which has inescapable logical [consequences](http://lesswrong.com/lw/nx/categorizing_has_consequences/) on the kinds of inferences that can be made using each concept.
+
+If that wasn't sufficiently clear, perhaps I have _failed as a writer_, and I can only beg that Ozy and our joint readership permit me the chance to try again.
+
+I don't want to _define_ gender based on psychology. ([Definitions are overrated](https://www.lesswrong.com/posts/cFzC996D7Jjds3vS9/arguing-by-definition), anyway.)[ref]Also, on a personal note, can I remark on how _weird_ and _uncomfortable_ it is that defending psychological sex differences has now apparently become my thing? I'm an individualist/egalitarian androgyny fan, not a [complementarian](https://en.wikipedia.org/wiki/Complementarianism); I [_don't want_](http://unremediatedgender.space/2017/Dec/theres-a-land-that-i-see-or-the-spirit-of-intervention/) women and men to have incommensurable souls. But faced with an intellectual climate where brilliant, kind, otherwise-sane people seem to feel morally obligated to _destroy our collective ability to reason about sex using natural language_, I feel morally obligated to not let them get away with it. Not for love of the territory in its current state, but for the love of that property of maps that _reflect_ the territory.[/ref] I do think that biological sex is almost as close as you can get to being a _natural category_[ref]The [chemical elements](https://en.wikipedia.org/wiki/List_of_chemical_elements) would be an example of an even more robustly natural category. Atoms with more protons than nitrogen but fewer than oxygen _do not exist_, and thus there is no analogue in chemistry to the "Well, what about intersex conditions?" [challenge](http://geekfeminism.wikia.com/wiki/Derailment) to the concept of sex or the "Well, what about [ring species](https://en.wikipedia.org/wiki/Ring_species)?" challenge to the concept of species.[/ref] in something like the following sense. If we imagine a distribution of artificial intelligences studying life on Earth and humans in particular, but lacking any preconceived concept of _sex_,[ref]It would be more traditional to put aliens rather than AIs in the observer role of this genre of thought experiment, but evolved aliens probably _would_ already know about sex![/ref] different AIs would each invent [different concepts](https://www.lesswrong.com/posts/XeHYXXTGRuDrhk5XL/unnatural-categories) in order to model the aspects of reality relevant to their own individual values, but most of them would be forced to reinvent the category of _sex_ sooner or later, because sex category membership [makes predictions about _many_](https://en.wikipedia.org/wiki/Sex_differences_in_humans) different [dimensions of observation](https://thingofthings.wordpress.com/2017/05/05/the-cluster-structure-of-genderspace/)—although some with much larger [or smaller](https://www.apa.org/pubs/journals/releases/amp-606581.pdf) effect sizes than others—at least _some_ of which are likely to be relevant to the interests of any particular AI that's paying any attention to animal life at _all_.
+
+[TODO: rephrase along the lines of, there might be situation where you might be quantitatively less surprised to see trans women behave in a way that is more common in males, than to see the same behavior in cis women, even if behavior never defines anyone's sex, because sex categories are attributed based on genetalia, secondary sex characteristics, &c.]
+
+[TODO: mention the rhetorical problem where I bring up psychology in discussions, because I can predict that if I said "You're not a woman because of your penis", ppl will say, "Oh, that's not what we meant"]
+
+That categories are clusters in a _high-dimensional_ space is relevant because of a statistical phenomenon perhaps most famously elucidated in [A. W. F. Edwards's critique of Richard Lewontin's critique of the concept of _race_](https://en.wikipedia.org/wiki/Human_Genetic_Diversity:_Lewontin's_Fallacy): groups that overlap along any _one_ particular measurement might be much more clearly distinguishable when you look at the conjunction of many different measurements.
+
+[TODO: the standard diagram]
+
+[I use the "bimodal multivariate distribution" frame a lot—it's even in the URL—but it's actually worse: sex-specific adaptations—functional adaptations and not just shifted distributions—are a thing ]
+
+only two types of gametes
+
+[if you have to do definitions, you go by physiology, because that's the part that's truly almost-completely-binary]
+
+When discussing whether a proposed recreational basketball association[ref]I'm somewhat reluctant to choose a sports example, because sporting is such a comparatively small and unimportant part of life—at least from the perspective of non-athletes—but it's a good place to start pedagogically, because merely physical sex differences are easy to measure and relatively uncontroversial, and it's important to avoid the distraction of unnecessarily contentious issues in the presentation of a topic that's already so prone to [motivated misunderstandings](https://en.wikipedia.org/wiki/Straw_man).[/ref] should be sex-segregated or not, one fact that might come up during the discussion is that the sex difference in human height has a magnitude of Cohen's _d_≈1.7, which is relevant because it means that insofar as selecting for good basketball players implies some degree of selection for tall people, it also implies some degree of selection for men, which would detract from the goal of creating an atmosphere where people are socially rewarded for excelling at the [high challenge](https://www.lesswrong.com/posts/29vqqmGNxNRGzffEj/high-challenge) of their chosen sport rather than for the (preëxisting, uninteresting, mostly immutable[ref]Given current technology.[/ref]) brute fact of their sex.
+
+[TODO: https://www.lesswrong.com/posts/neQ7eXuaXpiYw7SBy/the-least-convenient-possible-world where you do care about sports]
+
+So is the discussant who brings up height thereby claiming that _tall women aren't actually women_?
+
+Well, no. That would be stupid. Tall women might be more male-typical than female-typical _in the one particular aspect of their height_—and to some extent correlated variables like "weight"—but they are going to be more female-typical than male-typical in the _conjunction_ of all the _other_ measurements that are predicted from or used to assign sex categorizations—some of which measurements might _also_ be relevant to basketball.
+
+Of course, just because we plausibly want to separate our basketball league into divisions in the service of creating atmospheres of fair competition, sportsmanship, high challenge, _&c._, doesn't mean we have to do it by _sex_. If height were the only relevant major criterion,[ref]Which probably isn't going to be the case for basketball: consider that the sex difference in muscle mass is _d_≈2.6.[/ref] we would want height classes, just as boxers have [weight classes](https://en.wikipedia.org/wiki/Weight_class_(boxing)).[ref]Although it's worth noting that boxing weight classes are divisions _within_ an already otherwise single-sex competition.[/ref]
+
+Similar considerations apply to other social groups or events where some people think sex might be a relevant criterion of inclusion or exclusion. Ozy enumerates some ways in which they and our mutual friend, the author of the (again, highly recommended!) blog [_The Unit of Caring_](https://theunitofcaring.tumblr.com/), would be a poor fit for declared women-only social events. Ozy writes:
+
+> The actual category they should be using is not "cis women." The actual category they should be using is "people who would be contribute to the atmosphere you made this a woman-only event for."
+
+_In all philosophical strictness_, I think I agree. (And I wouldn't want to attend a men-only event.)[ref]But mostly for ideological and gender-dysphoria-related reasons, rather than because I _obviously wouldn't belong_. I've historically been inclined to cultivate a _self-image_ of being "not like the other guys", but self-images [aren't necessarily veridical](/2016/Sep/psychology-is-about-invalidating-peoples-identities/). If my self-perceived unmasculinity isn't reflected in other people's assessments of my unaffected personality and social behavior, it would be somewhat unreflective of me to protest, "But _I'm_ not gender-conforming—I have a _ponytail!_"[/ref] Outside of a few _relatively_ narrow domains of life (medicine, intercourse, family planning), I find it hard to think of good reasons to care about sex _per se_, as opposed to characteristics which might correlate with sex at some nonzero but certainly-not-so-huge-as-to-be-effectively-binary effect size. Ozy and me and Scott Alexander are all in agreement that categories are in the map, not the territory. There aren't ontologically-fundamental <span style="font-family: monospace;"><sex value="F"/></span> XML tags attached to people's souls—and moreover, we wouldn't have any reason to care if there _were_.
+
+The problem is that people don't always _have_ the detailed individual information that they would need to act in all philosophical strictness, at least not in an explicit, communicable form. If you're having a private get-together with some your friends who you know very well, you can pick and choose who to invite based on your individual knowledge of each individual, and you don't need to communicate (much less justify) your decision criteria to anyone else. If you don't like Brian, you can just not-invite-Brian, even if you're bad at introspection and don't even _know for yourself_ why you don't like Brian.
+
+In contrast, imagine telling the organizer of your local Women-in-Your-Favorite-Hobby meetup group, "The actual category you should be using is not _women_ in our-favorite-hobby. The actual category you should be using is, 'people who would contribute to the atmosphere you made this a woman-only event for.'"