A Generic Ontological Finickiness

On Twitter, Eliezer Yudkowsky wonders why male-to-female transsexualism triggers more reactionary public opprobrium than female-to-male.

Yudkowsky proposes four potential explanatory factors, not mutually exclusive: (1) that modern liberal Society doesn't care about gatekeeping male roles the way its forbearers did, (2) that MtF triggers intuitions in men for the repression of specifically male homosexuality, (3) that FtM transitions are more effective on the merits of passability with existing technology, and (4) that MtF threatens sex-segregation conventions whose purpose is to protect females from males (as of, e.g., bathrooms and chess tournaments).

I agree that to the extent that MtF faces more or different opposition than FtM, all four explanatory factors seem broadly plausible. I seem to disagree with Yudkowsky on the relative importance of the four factors and am cautiously skeptical of his reading of history, but the details are probably better relegated to a footnote.¹

The occasion of the present post is that Yudkowsky goes on to say:

One thing that all of these theories have in common: They do not appeal to the generic hardness of ontological categories producing alarm about generic boundary violations. They do not appeal to a deep generic horror of things being described as other than what they are.

I could only dream of a world where human beings always, inherently, consistently cared that much about exact descriptive accuracy. They would insist on FDA labels actually describing things.

An older era's indignation about a black trying to pass as white, a newer era's indignation about a white trying to pass as black, is tied to each era's notion of what specifically is being stolen—not to generic ontological finickiness. Hence the outrage being asymmetrical.

A theory #0 that people are outraged by things being classified into the wrong intrinsic category—as opposed to people reaching for that afterward as a rationalization—does not predict in advance any asymmetry of indignation about MtF versus FtM.

mike drop

With the important exception of the "as opposed to people reaching for that afterward as a rationalization" clause (as I'll explain momentarily), I agree with each of the sentences here, but I find it puzzling that Yudkowsky thinks this is a "mic drop" moment. Who exactly is being criticized here? Is there anyone who thinks that anti-trans public sentiment is explained by a deep generic horror of things being described as other than what they are? To establish that he's not beating up on a strawman, Yudkowsky would do better to quote some specific representative author actually making the claim that he's refuting.

As someone who has expressed a deep generic horror of things being described as other than what they are and criticized gender identity ideology on those specific grounds, perhaps I should clarify why I don't think Yudkowsky's rationalization accusation holds weight.

At one level, there's a simple answer: different people have different beliefs and motivations, even if they end up expressing roughly or superficially similar opinions on some particular issue; and also, the same person can harbor multiple motivations which all contribute to their opinion on some particular issue.

When gender-critical feminists like Meghan Murphy or Kellie-Jay Keen decry the destruction of women's single-sex spaces, they're standing up for what they see as women's interests, as feminists do. (This is Yudkowsky's theory #4.)

When conservatives like Michael Knowles decry transgenderism as continuous with a broader leftist project to alienate humanity from our embodied nature which was ordained by God, they're opposing radical social change, as conservatives do.

When philosophers like the present writer or Alex Byrne criticize gender identity theories as describing things as other than what they are, they're expressing a devotion to the truth, as philosophers do. (Given his earlier work on human rationality, one might have expected Yudkowsky to sympathize more with this motivation, but I guess those days are over.)

These things can all be true of these different people at the same time. The same person can belong to multiple groups. (Kathleen Stock, for example, is both a philosopher and a gender-critical feminist.) The same person can belong to one group but not others. (I am not a fan of Michael Knowles, and it seems safe to say that Meghan Murphy isn't, either.) There's no contradiction here to explain.

It might seem like there's a contradiction to explain if you project everyone's views down to a one-dimensional "pro-trans"/"anti-trans" subspace, divide the subspace into two buckets, and expect people in one bucket to answer for the views of everyone else in the same bucket. But that's an artifact of how much information you're throwing away by collapsing everyone into two buckets. Eliezer Yudkowsky and Timnit Gebru are both "anti-AI", but it would not make sense to reject a theory that some people are outraged about swiftly impending human extinction on the grounds that that theory predicts indifference to longer-term effects of global warming, which is not what we observe in anti-AI advocates. (Mic drop!)

Yudkowsky acknowledges that his theories #1–4 aren't mutually exclusive. (Theory #2, for example, only tries to explain anti-MtF sentiment in men.) But the same is true of theory #0. Yudkowsky is correct to point out that theory #0 can't be the whole story of public anti-trans sentiment, because that would imply equal amounts of anti-MtF and anti-FtM sentiment, which isn't what we observe. But that only means that the philosophers are outnumbered by more political actors (like the gender-criticals and conservatives), which I don't think anyone would doubt—least of all the philosophers.

The problem here is not just that no one said theory #0 was the whole story. Theory #0 wouldn't be the whole story even if the people described by theory #0 were objectively correct in their views, because it's possible for humans to arrive at correct beliefs for bad reasons. (The correctness of the theory of evolution isn't the whole story of why it receives public support: some people express pro-evolution sentiments to express resentment of their repressive religious upbringing. But evolution still actually happened.)

It's not clear why Yudkowsky would argue against the position that theory #0 by itself explains public anti-trans sentiment (without pointing to any examples of anyone who thinks that!), unless his goal were to minimize the gender-political relevance of the of the philosophy of language—to tar anyone expressing alarm about inaccurate descriptions as rationalizing bigotries held for less high-minded reasons.

But the philosophy of language clearly is relevant to what humans are doing when they argue about gender politics. The cognitive function of categorization is to group decision-relevantly similar things into the same category in order to make similar decisions about them. As a matter of AI theory, that's what human brains are doing when we use words, whether or not anyone knows it. (In the time of Aristotle, people didn't know it.)

In deriding the "generic hardness of ontological categories producing alarm about generic boundary violations" as something that "people reach[ ] for [...] afterward as a rationalization", Yudkowsky seems to portray philosophical arguments about accuracy of representations as somehow opposed to culture-specific sensitivities to specific misrepresentations—as if objecting to some lies more than others implies that it's hypocritical to have a philosophical account of lying.

But that's wrong. Policy arguments and epistemic arguments are deeply intertwined, because a policy presupposes the existence of some reality that the policy is about. A generic theory of deception is compatible with a policy that doesn't punish all deceptions equally.² Indeed, it's commonplace for punishments of deception to depend on who was misled about what. If you try to sell pyrite (a.k.a. "fool's gold") as gold, people will get angry at the attempted fraud. If you try to sell gold as pyrite, people will be confused but not angry.

Nevertheless, pyrite is not gold.

People who point out that pyrite is not gold are telling the truth.

People who point out that pyrite is not gold are still telling the truth even if their motivation for speaking up is that they think gold is valuable and pyrite is not, rather than a "generic ontological finickiness."

People who claim that pyrite is gold because they define gold as whatever sellers identify as gold, are lying or confused.

All of this continues to be true if gold used to be cheap and pyrite used to be expensive.

If someone develops a new chemical treatment that alters pyrite to have some but not all of the properties of gold, and they trademark the altered substance as "trans gold" (two words), then someone could validly make a case on the empirical merits that there should be a sense of the word gold such that trans gold is gold: "Make it yellower, increase the density, decrease the Mohs hardness, and I think most metallurgists would shrug and say 'good enough.'"

Fine. But it has to actually be good enough. The shrug has to be genuine, not coerced: a shrug of "I can't be bothered to pay attention to this", rather than of accepting a social convention that it's none of your business.³

I write what I do because when I look at the world I see and compare it to how I'm expected to describe it in polite liberal Society, I am not shrugging. If you, the reader, genuinely can't tell and don't care about the difference between males and females-on-testosterone, or between females and males-on-estrogen, then it makes sense for you to shrug. But I can often tell. (No, not always, and yes, I know about the obvious selection effect.)

The other week I was in Seattle for RustConf, which is the kind of event that has a trans-woman-to-cis-woman ratio higher than Less Online. At one point, I was sitting at a table with nine people, five or six of whom were trans women.

Why? The joint gender and assigned-sex-at-birth composition of that table can't possibly be a coincidence. Am I supposed to pretend not to notice?

In the comments on Twitter, a user proposed a theory #5 to Yudkowsky's puzzle, that FtMs are "indistinguishable from guys in the personality and mood and thought process and demeanor department. not so with mtf, they are more like a third different gender". Yudkowsky replied, "That's #3."

(Theory #3 verbatim was that "[m]odern medtech makes it easier to go FtM, and the lack of FtM panic in modern times compared to earlier times is because in modern times the aspiring FtM can use T to actually grow a beard. (And acquire male psychology (anger), but this is not as visible.) That is: The uncanny valley does not need to be symmetrical or fair with respect to perceptual cues or current medical technology. Maybe if you grow a beard, cut the hair, and remove the breasts, that just works to cross the valley.")

But theory #3's focus on visibility and perceptual cues is missing the commenter's point. If the medtech got much better, if it were cheap and easy to swap into a real female body, I would absolutely transition—and then that table at RustConf would have six or seven trans women. Even if there were no uncanny-valley effect, even if nothing looked visibly out of ordinary to the naked eye, The joint gender and assigned-sex-at-birth composition of that table would still not be a coincidence. Again, why?

It still feels ugly to say even now—it's not something I want to be true—but if I'm being honest, I have to put a decent chunk of probability on the hypothesis that women on average have less of whatever cognitive repertoires make people to want to spend their lives writing Rust code.⁴

If we're going to fantasize about improved transition medtech, should the tech do something about that? Alter the brain/mind as well as the body to give a more female-typical pattern of behavior, such that tables of six trans women and three cis men would not be a naturally occurring phenomenon in groups that are not explicitly selecting on gender and assigned-sex-at-birth status?⁵

If the goal is to be a true sex change, then it should, right?

But I don't think most trans women actually want that! I think a lot of us like being the kind of status-seeking gynephilic systematizers that we often are. Without prematurely committing myself to any particular policy decision, I think this has a lot of implications for how we should plan our lives and how the rest of Society should relate to us!

Is this a "generic ontological finickiness", "a deep generic horror of things being described as other than what they are"?

It depends on what you mean by generic. I certainly don't spend an equal amount of time blogging about all possible misrepresentations-via-suboptimal-categorization. That's obviously not possible and it's not even clear what it would mean. I selfishly focus on the topics that are important to me.

In another sense, I suppose it is "generic" in the sense that I do think there are general laws of cognition governing language and thought, which I seek to understand and apply in my work in the hopes of improving the accuracy of shared maps. The methodology is general even if the choice of topic is idiosyncratic and personal.

Yudkowsky writes that he could only dream of a world where human beings cared that much about exact descriptive accuracy.

But why only dream? Why not care, yourself, and exhort others to care, in public, on Twitter? Maybe sometime in the last sixteen years, Yudkowsky has abandoned his sense that more is possible.

But I haven't. Maybe human beings don't always, inherently, consistently care about exact descriptive accuracy—but rationalists should.