Challenges to Yudkowsky’s Pronoun Reform Proposal

Link post

Go, Soul, the body’s guest,
Upon a thankless errand:
Fear not to touch the best;
The truth shall be thy warrant:
Go, since I needs must die,
And give the world the lie.

”The Lie” by Walter Raleigh

Summary

  • In a February 2021 Facebook post, Eliezer Yudkowsky inveighs against English’s system of singular third-person pronouns: as a matter of language design, English’s lack of a gender-neutral singular third-person pronoun is a serious flaw: you shouldn’t be required to commit to a stance on what sex someone is in order to say a grammatical sentence about her or him.

  • This seems fine as a critique of the existing English language. However, Yudkowsky then goes on to proclaim, in connection with pronouns for transgender people, that “the simplest and best protocol is, ‘”He” refers to the set of people who have asked us to use “he”, with a default for those-who-haven’t-asked that goes by gamete size’ and to say that this just is the normative definition. Because it is logically rude, not just socially rude, to try to bake any other more complicated and controversial definition into the very language protocol we are using to communicate.”

  • However, this allegedly “simplest and best” proposal fails to achieve its stated aim of avoiding baking controversial claims into the language grammar. The reason trans people want others to use their designated pronouns is because they’re trying to control their socially-perceived sex category and English speakers interpret she and he as conveying sex-category information. Yudkowsky’s proposed circular redefinition is functionally “hypocritical”: if it were actually true that he simply referred to those who take the pronoun he, then there would be no reason for trans people to care which pronoun people used for them.

  • The “meaning” of language isn’t some epiphenominal extraphysical fact that can be declared or ascertained separately from common usage. The word “dog” means what it does because English speakers use the word that way; if you wanted “dog” to mean something different, you’d need to change the way English speakers behave. Thus, circularly redefining he and she as purportedly referring to pronoun preferences rather than sex doesn’t work, if people are still in practice choosing pronouns on the basis of perceived sex.

  • Given that she and he do in fact convey sex category information to English speakers, some speakers might perceive an interest in refusing demands to use pronouns in a way that contradicts their perception of what sex people are. This does not constitute a philosophical commitment that pronouns can be “lies” as such.

  • In the comments of the Facebook post, Yudkowsky seemingly denies that pronouns convey sex category information to English speakers, claiming, “I do not know what it feels like from the inside to feel like a pronoun is attached to something in your head much more firmly than ‘doesn’t look like an Oliver’ is attached to something in your head.” This self-report is not plausible, as evidenced by previous writings by Yudkowsky that treat sex and pronouns as synonymous.

  • I’m not claiming that Yudkowsky should have a different pronoun usage policy. I agree that misgendering all trans people “on principle” seems very wrong and unappealing. Rather, I’m claiming that policy debates should not appear one-sided: in order to be politically neutral in your analysis of why someone might choose one pronoun usage policy over another, you need to acknowledge the costs and benefits of a policy to different parties. It can simultaneously be the case that pressuring speakers to use pronouns at odds with their perceptions of sex is a cost to those speakers, and that failing to exert such pressure is a cost to trans people. It’s possible and desirable to be honest about that cost–benefit analysis, while ultimately choosing a policy that favors some parties’ interests over others.

  • People with gender dysphoria who are considering whether to transition need factually accurate information about gender-transition interventions: if you have the facts wrong, you might wrongly avoid an intervention that would have benefited you, or wrongly undergo an intervention that harms you. This includes facts about how pronouns work in the existing English language. If it were actually true that the simplest and best convention is that he refers to the set of people who have asked us to use he, then asking for new pronouns despite not physically passing as the corresponding sex wouldn’t be costly. But in fact, it is costly. As someone with a history of gender problems, this is decision-relevant to me. Thus, Yudkowsky is harming a reference class of people that includes me by spreading disinformation about the costs of asking for new pronouns; I’m better off because I don’t trust Eliezer Yudkowsky to tell the truth.


In a February 2021 Facebook post, Eliezer Yudkowsky inveighs against English’s system of singular third-person pronouns. As a matter of clean language design, English’s lack of a gender-neutral singular third-person pronoun is a serious flaw. The function of pronouns is to have a brief way to refer back to entities already mentioned: it’s more concise to be able to say “Katherine put her book on its shelf” rather than “Katherine put Katherine’s book on the book’s shelf”. But then why couple that grammatical function to sex-category membership? You shouldn’t need to take a stance on someone’s sex in order to talk about her or him putting a book on the shelf.

This affects, for example, science-fiction authors writing about AIs or hermaphroditic aliens (which don’t have a sex), or mystery authors writing about a crime suspect whose identity (and therefore, sex) is unknown. In these cases, she or he are inappropriate, but the English language offers no alternative lacking its own downsides: it is understood to refer to non-persons, they gets conjugated as a plural, and neopronouns like ey/​em/​eir—or ve/​ver/​vis, as used in some of Yudkowsky’s juvenilia—are hard to rally adoption for because pronouns are a closed class—not something people are used to new members of being coined, in the way that people are used to seeing unfamiliar nouns, adjectives, or verbs.

It doesn’t have to be this way! If you were fortunate enough to be in the position of intelligently designing a language from scratch, you could just include a singular third-person gender-neutral pronoun (like it, but for persons, or like they but unambiguously singular) in the original closed set of pronouns! If you wanted more pronoun-classes to reduce the probability of collisions (where universal ey or singular they would result in more frequent need to repeat names where a pronoun would be ambiguous), you could devise some other system that doesn’t bake sex into the language while driving the collision rate even lower than that of the sex-based system—like using initials to form pronouns (Katherine put ker book on its shelf?), or an oral or written analogue of spatial referencing in American Sign Language (where a signer associates a name or description with a direction in space, and points in that direction for subsequent references).

(Although—one might speculate that “more classes to reduce collisions” could be part of the historical explanation for grammatical gender, in conjunction with the fact that sex is binary and easy to observe. None of the other most salient features of a human can quite accomplish the same job: age is continuous rather than categorical; race is also largely continuous (clinal) and historically didn’t typically vary within a tribal/​community context.)

If you grew up speaking English, gendered pronouns feel “normal” while gendered noun classes in many other languages (where, e.g., in French, a dog, le chien, is “masculine”, but potatoes, la pommes de terre, are “feminine”) seem strange and unnecessary, but someone who grew up with neither would regard both as strange. If you spoke a language that didn’t already have gendered pronouns, you probably wouldn’t be spontaneously eager to add them.

All this seems fine as a critique of the existing English pronoun system! However, I argue that Yudkowsky’s prescription for English speakers going forward goes badly wrong. First, Yudkowsky argues that it’s bad for stances on complicated empirical issues to be part of the language grammar itself: since people might disagree on who fits into the empirical clusters of “female” and “male”, you don’t want speakers to be forced to make a call on that just in order to be able to use a pronoun.

Fair enough. Sounds like an argument for universal singular they (and eating the cost of increased collisions where it’s ambiguous which subject an instance of they would refer to): if you don’t think pronouns should convey sex-category information, then don’t use pronouns that convey sex-category information! But then, in an unexplained leap, Yudkowsky proclaims:

So it seems to me that the simplest and best protocol is, “‘He’ refers to the set of people who have asked us to use ‘he’, with a default for those-who-haven’t-asked that goes by gamete size” and to say that this just is the normative definition. Because it is logically rude, not just socially rude, to try to bake any other more complicated and controversial definition into the very language protocol we are using to communicate.

The problem with this is that the alleged rationale for the proposal does not support the proposal. If your default pronoun for those-who-haven’t-asked goes by perceived sex (which one presumes is what Yudkowsky means by “gamete size”—we almost never observe people’s gametes), then you’re still baking sex-category information into the language protocol in the form of the default! Moreover, this is clearly an “intended” rather than an accidental effect of the proposal, in the sense that a policy that actually avoided baking sex-category information into the language (like universal singular they, or name-initial- or hair-color-based pronouns) would not have the same appeal to those who support self-chosen pronouns: why is it that some people would want to opt-out of the sex-based default?

Well, it would seem that the motivating example—the causal–historical explanation for why we’re having this conversation about pronoun reform in the first place—is that trans men (female-to-male transsexuals) prefer to be called he, and trans women (male-to-female transsexuals) prefer to be called she. (Transsexuals seem much more common than people who just have principled opinions about pronoun reform without any accompanying desire to change what sex other people perceive them as.)

But the reason trans people want this is because they’re trying to change their socially-perceived sex category (“gender”) and actually-existing English speakers interpret she and he as conveying sex-category information. People who request he/​him pronouns aren’t doing it because they want their subject pronoun to be a two-letter word rather than a three-letter word, or because they hate the voiceless postalveolar fricative (sh) sound. They’re doing it because, in English, those are the pronouns for males. If it were actually true that she and he were just two alternative third-person pronouns that could be used interchangeably with no difference in meaning, with the only function of the distinction being collision-avoidance, then there would be no reason to care which one someone used, as long as the referent was clear. But this doesn’t match people’s behavior: using gender pronouns other than those preferred by the subject is typically responded to as a social attack (as would be predicted by the theory that she and he convey sex-category information and transsexuals don’t want to be perceived as their natal sex), not with, “Oh, it took me an extra second to parse your sentence because you unexpectedly used a pronoun different from the one the subject prefers as per convention, but now I understand what you meant” (as would be predicted by the theory that “he refers to the set of people who have asked us to use he [...] and to say that this just is the normative definition”).

You can’t have it both ways. “That toy is worthless”, says one child to another, “therefore, you should give it to me.” But if the toy were actually worthless, why is the first child demanding it? The problem here is not particularly subtle or hard to understand! If the second child were to appeal to an adult’s authority, and the adult replied, “The toy is worthless, so give it to him,” you would suspect the grown-up of not being impartial.

“Pronouns shouldn’t convey sex-category information, as an apolitical matter of language design,” is a fine motte, but it’s not consistent with the bailey of, “Therefore, when people request that you alter your pronoun usage in order to change the sex-category information being conveyed, you should obey the request.” Even if the situation is an artifact of bad language design, as Yudkowsky argues—that in a saner world, this conflict would have never come up—that doesn’t automatically favor resolving the conflict in favor of the policy of keeping both she and he but asserting that the difference doesn’t mean anything.

This may be clearer to some readers if we consider a distinction less emotionally and politically fraught than sex/​gender in the current year. Many languages have two different second person singular pronouns that distinguish the speaker’s relationship to the listener as being more familiar/​intimate, or more formal/​hierarchical. In Spanish, for example, the familiar pronoun is and the formal pronoun is usted: one would address friends, family members, children, or personal servants as , but strangers or social superiors as usted. Using the wrong pronoun can be the cause of offense or awkwardness. A speaker switching from usted to for an interlocutor who they’re getting along with might ask if it’s okay with ¿Te puedo tutear? (Can I call you ?) or Nos tuteamos, ¿verdad? (We call each other , right?); this is somewhat analogous to an English speaker asking if they may address someone by first name, rather than with a courtesy title or honorific (Ms./​Mr. Lastname, or ma’am/​sir).

One could argue that the /​usted distinction is bad language design for the same reason Yudkowsky opposes the she/​he distinction: you shouldn’t be forced to make a call on how familiar your relationship with someone is just in order to be able to use a pronoun for them. The modern English way is more flexible: you can indicate formality if you want to by saying additional words, but it’s not baked into the grammar itself.

However, if you were going to reform Spanish (or some other language with the second-person formality distinction), you would probably abolish the distinction altogether, and just settle on one second-person singular pronoun. Indeed, that’s what happened in English historically—the formal you took over as the universal second-person pronoun, and the informal singular thou/​thee/​thine has vanished from common usage. (People still recognize it as a second-person pronoun when encountered in old poetry—”The truth shall be thy warrant”, &c.—but most probably aren’t aware of the formality distinction.) You wouldn’t keep both forms, but circularly redefine them as referring only to the referent’s preferred choice of address (?!).

Similarly, when second-wave feminists objected to the convention of Miss or Mrs. forcing speakers to identify a woman’s marital status, the response was to popularize the marriage-agnostic alternative Ms., not to circularly redefine Miss and Mrs.

Or consider how previous generations of public intellectuals considered this exact problem. In 1983, Douglas R. Hofstadter also expressed disapproval of she and he as a matter of language design, and to illustrate the point about how alien and unnecessary gendered language would seem if you weren’t already used to it, wrote a satirical piece, “A Person Paper on Purity in Language”, in the persona of a conservative author in a society with race-based (!) language conventions, including the pronouns whe/​wis for whites and ble/​bler for blacks. In neither the piece itself (during which Hofstadter’s alter-ego brings up and rejects a couple of reform suggestions from the liberals of whis Society, including singular they), nor the Post-Scriptum in its subsequent anthologization, does Hofstadter entertain the idea of redefining he and she (or whe and ble) to refer to the subject’s pronoun preference.

It’s worth asking: why not? The statement of the objective language-design flaw (pronouns shouldn’t denote sex, that’s dumb; why would you define a language that way) was the same in 1983 as it is in 2022. If it’s so clear to Yudkowsky in the current year that self-identification is just the “simplest and best protocol” to repair the objective flaw in English’s design, why didn’t that simplest and best solution occur to Hofstadter in 1983?

Could it, perhaps, be the case that public intellectuals in the current year might have some other motivation to conclude that “he refers to the set of people who have asked us to use he”, that was not present for their analogues in 1983? But if so, they’d tell us that … right?

Really, the circular definition shouldn’t satisfy anyone: people who want someone to call them usted (or ), do so because of the difference in meaning and implied familiarity/​respect, in the existing (pre-reform) language. (Where else could such a preference possibly come from?) From an AI design standpoint, the circular redefinition can be seen as a form of “wireheading”. You want people to respect you as a superior, and if they respected you as a superior, they’d call you usted. That could make a policy of coercing people into calling you usted seem superficially appealing. But the appeal solely rests on confusing the pre-reform meaning (under which the choice of usted implies respect and is therefore desirable) and the post-reform meaning (under which the choice implies nothing). Whether or not the proponent of the change consciously notices the problem, the redefinition is functionally “hypocritical”: it’s only desirable insofar as people aren’t actually using it internally.

Indeed, when I look at what contemporary trans activists write, I don’t see them approving of this idea that pronoun choices don’t mean anything. In the words of one Twitter user:

misgendering sucks, but what feels even more violent is when people get my pronouns right and i can tell they still perceive me as a man

In the words of another:

a lot of cis people use ‘learning someone’s pronoun’ as a copout from doing the important internal work of actually reconsidering their impression of the person’s gender

like let’s be real—the reason you have a hard time “remembering” her pronoun is because you don’t really think of her as a her. if you practiced thinking of her as a her, her pronoun would just come. and then you wouldn’t be privately betraying her in your head all the time.

These authors are to be commended for making their view so clear and explicit: in order to not betray your trans friends (according to this view), you need to think of them as the gender that they say they are. Mere verbal pronoun compliance in the absence of underlying belief is insufficient and possibly treacherous.

This point that pronoun changes are desired precisely because of what they do imply about sex categories in the existing English language is a pretty basic one, that one would think should scarcely need to be explained. And yet Yudkowsky steadfastly ignores the role of existing meanings in this debate, bizarrely writing as if we were defining a conlang from scratch:

It is Shenanigans to try to bake your stance on how clustered things are and how appropriate it is to discretely cluster them using various criteria, into the pronoun system of a language and interpretation convention that you insist everybody use!

There are a couple of problems with this. First of all, the “that you insist everybody use” part is a pretty blatant DARVO in the current political environment around Yudkowsky’s social sphere. A lot of the opposition to self-chosen pronouns is about opposition to compelled speech: people who don’t think some trans person’s transition should “count”—however cruel or capricious that might be—don’t want to be coerced into legitimizing it with the pronoun choices in their own speech. That’s different from insisting that others use sex-based non-subject-preferred pronouns, which is not something I see much of outside of gender-critical (“TERF”) forums. That is, in the world I see, the pronouns-by-self-identity faction is overwhelmingly the one “insist[ing] everybody use” their preferred convention. Characterizing the issue as being about “freedom of pronouns”, as Yudkowsky does in the comment section, elides the fact that freedom to specify how other people talk about you is in direct conflict with the freedom of speech of speakers! No matter which side of the conflict one supports, it seems wrong to characterize the self-ID pronoun side as being “pro-freedom”, as if there weren’t any “freedom” concerns on the other side.

If you actually believed it was Shenanigans to bake a stance on how clustered things are into a pronoun system and insist that everyone else use it, then it should be equally Shenanigans independently of whether the insisted-on clusters are those of sex or those of gender identity—if you’re going to be consistent, you should condemn them both. And yet somehow, people who insist on sex-based pronouns are the target of Yudkowsky’s condescension, whereas people who insist on gender-identity-based pronouns get both a free pass, and endorsement of their preferred convention (albeit for a different stated reason)? The one-sidedness here is pretty shameless!

Perhaps more important than the speaker-freedom vs. subject-freedom issue, however, is that in discussing how to reform English, we’re not actually in the position of defining a language from scratch. Even if you think the cultural evolution of English involved Shenanigans, it’s not fair to attribute the Shenanigans to native speakers accurately describing their native language. Certainly, language can evolve; words can change meaning over time; if you can get the people in some community to start using language differently, then you have ipso facto changed their language. But when we consider language as an information-processing system, we see that in order to change the meaning associated with a word, you actually do have to somehow get people to change their usage. You can advocate for your new meaning and use it in your own speech, but you can’t just declare your preferred new meaning and claim that it applies to the language as actually spoken, without speakers actually changing their behavior. As a result, Yudkowsky’s proposal “to say that this just is the normative definition” doesn’t work.

To be clear, when I say that the proposal doesn’t work, I’m not even saying I disagree with it. I mean that it literally, factually doesn’t work! Let me explain.

The “meaning” of language isn’t some epiphenomenal extraphysical fact that can be declared or ascertained separately from common usage. We can only say that the English word “dog” means these-and-such four-legged furry creatures, because English speakers actually use the word that way. The meaning “lives” in the systematic correspondence between things in the world and what communication signals are sent.

There’s nothing magical about the particular word/​symbol/​phoneme-sequence “dog”, of course. In German, they say Hund; in Finnish, they say koira; in Korean, they say . Germans and Finns and Koreans (and their dogs) seem to be getting along just as well as we Anglophones.

Nevertheless, it is a fact about contemporary English that “dog” means dog. If you thought this was bad for whatever reason, and you wanted to change that fact, you’d have to change the behavior of actually-existing English speakers. If you tried to stipulate on your Facebook wall that the word “dog” should mean tree now, and all of your Facebook friends nodded in agreement at your clever argument and then continued to call dogs “dogs” and trees “trees” in their everyday life just like they always had, then your language reform attempt would have, in fact, failed—even if the fact that it failed would be less obvious if you only looked at the Facebook thread full of people nodding in agreement.

Or suppose I wrote a Facebook post arguing that it’s bad language design that “billion” means 1,000,000,000 instead of 2001. You see, the etymology comes from the prefix bi- (meaning two, from the Latin bis), combined with mille (Latin for 1000), combined with the augmentive suffix -one. How do you get 10⁹ from that, huh? (It turns out there’s an explanation, but I don’t find it intuitive.) Clearly, it’s better language design if the meaning of number words straightforwardly reflects their parts, so “billion” should mean 2001 (bi-, mille, -one; 2, times 1000, plus 1).

Even if you found this argument compelling from an theoretical language-design perspective after it had been presented to you, if I were to subsequently go around calling myself a billionaire (and condescendingly Tweeting about how anyone objecting to this usage is ontologically confused), you would probably suspect that I had some other reason to come up with this particular theoretical language-design argument—probably a reason having to do with what “billion” already means in the usage of actually-existing English speakers, even if you honestly think the existing English language is poorly designed in that aspect.

The inseparability of meaning from behavior-and-usage may be clearer if considered in a context other than that of natural language. Take computer programs. Sometimes programmers make bad design decisions. For example, in the C programming language, it’s standard to represent strings (textual data) in memory with a sequence of bytes ending in a zero (null) character; the machine only knows where the string stops when it reaches the null at the end. This convention has a lot of disadvantages relative to the alternative of prefixing the string data with the length; a missing or misplaced null character could cause the machine to erroneously read or write data in adjacent memory, causing serious bugs or security vulnerabilities.

Given the existence of strong arguments for the length-prefixed string convention, replacing old software that uses null-terminated strings with new software that uses length-prefixed strings, sounds like a good idea! But the thing is, you do have to upgrade or replace the old software. If you just start sending data in a new format to the old software that doesn’t understand the new format, your code is not going to yield the expected results. It would be convenient if you could just declare a new semantics for your existing data on your Facebook wall and be done, but that just doesn’t work if you’re still using the old software, which is programmed to behave according to the old data-interpretation convention. This continues to be true even if the convention you’re trying to retire is very bad (like null-terminated strings), and if the old software is widely deployed and would be very expensive to systematically replace. The backwards-compatibility trap is real and can’t be defied away even if it’s very unpleasant.

Natural language faces a similar backwards-compatibility trap. The English language, as “software”, is already “deployed” to 370 million brains as native speakers, and another 980 million second-language speakers. And among those hundreds of millions of speakers, there is already a very firmly entrenched convention that she refers to females and he refers to males, such that if you say, “I met a stranger in the park; she was nice”, the listener is going to assume the the stranger was (or appeared to be) female, even if you didn’t say “The stranger was female” as a separate sentence. If the listener later gets the chance to meet the stranger and the stranger turns out to be (or appear to be) male, the listener is going to be surprised: your pronoun choice induced them to mis-anticipate their experiences.

Bad language design? I mean, maybe! You could argue that! You could probably get a lot of Likes on Facebook arguing that! But if 370 million native English speakers including you and virtually everyone who Liked your post are going to continue automatically noticing what sex people are (or appear to be) and using the corresponding pronouns without consciously thinking about it (in accordance with the “default for those-who-haven’t-asked” clause of your reform proposal), then the criticism seems kind of idle!

The “default for those-who-haven’t-asked [going] by gamete size” part of Yudkowsky’s proposal is trying to deal with the backwards-compatibility problem by being backwards-compatible—prescribing the same behavior in the vast majority of cases—but in doing so, it fails to accomplish its stated purpose of de-gendering the language.

To actually de-gender English while keeping she and he (as contrasted to coordinating a jump to universal singular they, or ve), you’d need to actually shatter the correlation between pronouns and sex/​gender, such that a person’s pronouns were just an arbitrary extra piece of data that you couldn’t deduce from secondary sex characteristics and just needed to remember in the same way you have to remember people’s names and can’t deduce them from their appearances. But as far as I can tell, no one wants this. When’s the last time you heard someone request pronouns for non-gender-related reasons? (“My pronouns are she/​her—but note, that’s just because I prefer the aesthetics of how the pronouns sound; I’m not in any way claiming that you should believe that I’m in any sense female, which isn’t true.”) Me neither.

But given that pronouns do convey sex-category information, as a fact about how the brains of actually-existing English speakers in fact process language (whether or not this means that English is terribly designed), some actually-existing English speakers might have reason to object when pressured to use pronouns in a way that contradicts their perception of what sex people are.

In an article titled “Pronouns are Rohypnol”, Barra Kerr compares preferred pronouns to the famous Stroop effect. When color words are printed in text of a different color and people are asked to name the color of the text, they’re slow to respond: the meaning of the word interferes with our ability to name the color in front of our eyes.

Kerr suggests that preferred pronouns have a similar effect, that “a conflict between what we see and know to be true, and what we are expected to say, affects us.” As an exercise, she suggests (privately!) translating sentences about transgender people to use natal-sex-based pronouns.

Unfortunately, I don’t have a study with objective measurements on hand, but I think most native English speakers who try this exercise and introspect—especially using examples where the trans person exhibits features or behavior typical of their natal sex, with things like “she ejaculated” or “he gave birth” being the starkest examples—will agree with Kerr’s assessment: “You can know perfectly the actual sex of a male person, and yet you will still react differently if someone calls them she instead of he.”

Let’s relate this to Yudkowsky’s specialty of artificial intelligence. In a post on “Multimodal Neurons in Artificial Neural Networks”, Gabriel Goh et al. explore the capabilities and biases of the CLIP neural network trained on textual and image data.

There are some striking parallels between CLIP’s behavior, and phenomena observed in neuroscience. Neurons in the human brain have been observed to respond to the same concept represented in different modalities; for example, Quiroga et al. observed a neuron in one patient that responded to photos and sketches of actress Halle Berry, as well as the text string “Halle Berry”. It turns out that CLIP neurons also exhibit this multi-modal responsiveness. Furthermore, CLIP is vulnerable to a Stroop-like effect where its image-classification capabilities can be fooled by “typographic attacks”—a dog with instances of the text “$$$” superimposed over it gets classified as a piggy bank, an apple with a handwritten sign saying “LIBRARY” gets classified as a library. The network knows perfectly what dogs and apples look like, and yet still reacts differently if adjacent text calls them something else.

I conjecture that the appeal of subject-chosen pronouns lies precisely in how they exert Stroop-like effects on speakers’ and listeners’ cognition. (Once again, if it were actually true that she and he had no difference in meaning, there would be no reason to care.) Pronoun badges are, quite literally, a typographic attack against English speakers’ brains.

Note, I mean this as a value-free description of how the convention actually functions in the real world, not a condemnation. One could consistently hold that these “attacks” are morally good. (Analogously, supernormal stimuli like chocolate or pornography are “attacks” against the brain’s evolved nutrition and reproductive-opportunity detectors, but most people are fine with this, because our goals are not evolution’s.)

Is susceptibility to Stroop-like effects an indication of bad mind design? I mean, probably! One would expect that an intelligently-designed agent (as contrasted to messy human brains coughed up by blind evolution or lucky neural networks found by gradient descent) could easily bind and re-bind symbols on the fly, such that a sane AI from the future could use whatever pronouns without dredging up any inapplicable mental associations. But it seems kind of idle to criticize humans for not having a capability (natural language fluency without Stroop-like effects) that we don’t even know how to implement in a computer program.

Back to Kerr’s article—importantly, Kerr is explicitly appealing to psychological effects of different pronoun conventions. She is absolutely not claiming that the use of preferred pronouns is itself a “lie” about some testable proposition. She writes:

I’ve heard many people tell me they don’t mind doing this, as a courtesy, although it takes some effort to keep up the mental gymnastics of perceiving one sex, but consistently using pronouns for the other. That’s a personal choice, and I respect the reasons why some people make it.

I’ve also heard many people declaring that anyone who won’t comply (usually directed at a woman) is obnoxious, mean, hostile, and unpleasant. ‘Misgendering’ is hate speech. They say.

But I refuse to use female pronouns for anyone male.

Note the wording: “That’s a personal choice”, “I refuse”. Kerr knows perfectly well that people who use gender-identity-based pronouns aren’t making a false claim that trans men produce sperm, &c.! Rather, she’s saying that a pronoun convention that groups together females, and a minority of males who wish they were female, affects our cognition about that minority of males in a way that’s disadvantageous to Kerr’s interests (because she wants to be especially alert to threats posed by males), such that Kerr refuses to comply with that convention in her own speech. (Compare to how a Spanish speaker might refuse to address someone they disrespected as usted because of its connotations, without thereby claiming that using usted would make the sentence literally false.)

Relatedly, critics of this blog sometimes refer to me as she, reflecting their belief that I’m a trans woman in denial, even though I think of myself of a man (adult human male not trying to appear otherwise). I never correct them—not just because it’s kind of flattering, and not just because I don’t think of myself as having the right to dictate how other people talk about me—but because “she” is the correct pronoun to convey the meaning they’re trying to express, whether or not I agree with it.

I take pains to emphasize that pronouns can have meaningful semantics without being denotative statements that can be straightforwardly “false”, because Yudkowsky misrepresents what his political opponents are typically claiming, repeatedly trying to frame the matter of dispute as to whether pronouns can be “lies” (to which Yudkowsky says, No, that would be ontologically confused)—whereas if you actually read what the people on the other side of the policy debate are saying, they’re largely not claiming that “pronouns are lies”!

This misrepresentation is a serious problem because, as Yudkowsky pointed out in 2007, “To argue against an idea honestly, you should argue against the best arguments of the strongest advocates. Arguing against weaker advocates proves nothing, because even the strongest idea will attract weak advocates.” By selectively drawing attention to the weaker form of the argument, Yudkowsky is likely to leave readers who trusted him to be fair with an unrealistic picture of what people on the other side of the issue actually believe. (Kerr’s article seems representative of gender-critical (“TERF”) concerns; I’ve seen the post linked in those circles more than once, and it’s cited in embattled former University of Sussex professor Kathleen Stock’s book Material Girls.)

Anyway, given these reasons why the existing meanings of she and he are relevant to the question of pronoun reform, what is Yudkowsky’s response?

Apparently, to play dumb. In the comments of the Facebook post, Yudkowsky mentions encountering exotic pronouns on LambdaMOO at age 13 and no one thinking anything of them, and goes on to claim:

I do not know what it feels like from the inside to feel like a pronoun is attached to something in your head much more firmly than “doesn’t look like an Oliver” is attached to something in your head.

...

I’m sorry, but I can’t take this self-report literally. I certainly don’t think Yudkowsky was consciously lying when he wrote that. (When speaking or writing quickly without taking the time to scrupulously check every sentence, it’s common for little untruths and distortions to slip into one’s speech. Everyone does it, and if you think you don’t, then you’re lying.)

Nevertheless, I am incredibly skeptical that Yudkowsky actually doesn’t know what it feels like from the inside to feel like a pronoun is attached to sex categories more firmly than a given name is attached to someone’s appearance.

I realize this must seem impossibly rude, presumptuous, and uncharitable of me. Yudkowsky said he doesn’t know what it feels like from the inside! That’s a report out his own mental state, which he has privileged introspective access to, and I don’t! What grounds could I possibly, possibly have to think he’s not telling the truth about his own mind?

It’s a good question. And my answer is, even without mind-reading technology, people’s minds are still part of the same cause-and-effect physical universe that I can (must) make probabilistic inferences about, and verbal self-reports aren’t my only source of evidence about someone’s mind. In particular, if someone’s verbal self-report mis-predicts what we know about their behavior, it’s far from clear that we should trust the report more than our senses.

And the thing is, Eliezer Yudkowsky is a native English speaker born in 1979. As a native English speaker born in 1987, I have a pretty good mental model of how native English speakers born in the late 20th century use language. And one of the things native English speakers born in the late 20th century are very good at doing, is noticing what sex people are and using the corresponding pronouns without consciously thinking about it, because the pronouns are attached to the concept of sex in their heads more firmly than proper names are attached to something in their heads.

I would bet at very generous odds that at some point in his four decades on Earth, Eliezer Yudkowsky has used she or he on the basis of perceived sex to refer to someone whose name he didn’t know. Because all native English speakers do this. Moreover, we can say something about the cognitive algorithm underlying how they do this. People can recognize sex from facial photos alone (hair covered, males clean-shaven) at 96% accuracy. In naturalistic settings where we can see and hear more secondary sex characteristics than just someone’s face (build, height, breasts, voice, gait, &c.), accuracy would be even greater. It’s not a mystery why people can get sex-based pronouns “right” the vast majority of the time without having to be told or remember specific people’s pronouns.

Conversely, I would also bet at very generous odds that in his four decades on Earth, Eliezer Yudkowsky has very rarely if ever assumed what someone’s name is on the basis of their appearance without being told. Because no native English speakers do this (seriously, rather than as a joke or a troll). Now, it’s true that the “doesn’t look like an Oliver” example was introduced into the discussion by another commenter, who recounts once having called someone Bill who had introduced himself as Oliver:

It did feel a little weird calling him Oliver, but everyone present knew what I was doing was being a jerk and teenagers are horrible. The “feels like lying” principle seems like it lets me keep calling him Bill, now righteously. I just can’t even really bring myself to play in that sandbox in good faith.

But the “everyone present knew what I was doing was being a jerk” characterization seems to agree that the motivation was joking/​trolling. How did everyone present know? Because it’s absurd to infer a particular name from someone’s appearance.

It’s true that there are name–feature correlations that observers can pick up on. For example, a “Juan” is likely to be Latino, a “Gertrude” in the current year is likely to be old; a non-Hispanic white Juan or a young Gertrude may indeed be likely to provoke a “Doesn’t look like an X” reaction (which may also be sensitive to even subtler features). But while probabilistic inferences from features to low likelihood of a particular name are valid, an inference from features to a particular name is absolutely not, because the function of a name is to be an opaque “pointer” to a particular individual. A Latino family choosing a name for their male baby may be somewhat more likely to choose “Juan” rather than “Oliver” (or “Gertrude”), but they could just as easily choose “Luis” or “Miguel” or “Alejandro” for the very same child, and there’s no plausible physical mechanism by which a horrible teenager thirty years later could tell the difference.

Thus, I reject the commenter’s claim that “feels like lying” intuitions about pronouns and sex would have let her “keep calling him Bill, now righteously”. What algorithm you would use to infer that someone’s name is “Bill” based on how he looks? What are the “secondary Oliver characteristics”, specifically? People for whom it was actually true that names map to appearances the way pronouns map to sex, should not have trouble answering these questions!

If there were a substantial contingent of native English speakers who don’t interpret pronouns as conveying sex category information, one would expect this to show up in our cultural corpus more often—and yet, I’m actually not aware of any notable examples of this. In contrast, it’s very easy to find instances of speakers treating pronouns and sex as synonymous. As an arbitrarily chosen example, in one episode of the animated series The Amazing World of Gumball featuring the ravenous spawn of our protagonists’ evil pet turtle, the anthropomorphic-rabbit Bumbling Dad character says, “Who’s to say this pregnant turtle is a her?” and everyone gives him a look.

The joke, you see, is that bunny-father is unthinkingly applying the stock question “Who’s to say X is a he/​she?” (which makes sense when X is, e.g., “the nurse”) in a context where there’s an obvious answer—namely, that the referents of “her” pronouns are female and only females get pregnant—but the character is too stupid to notice this, and we enjoy a laugh at his expense.

The Amazing World of Gumball is rated TV-Y7 and the episode in question came out in 2016. This is not a particularly foreign or distant cultural context, nor one that is expected to tax the cognitive abilities of a seven-year-old child! Is … is Yudkowsky claiming not to get the joke?

Posed that way, one would imagine not—but if Yudkowsky does get the joke, then I don’t think he can simultaneously honestly claim to “not know what it feels like from the inside to feel like a pronoun is attached to something in your head much more firmly than ‘doesn’t look like an Oliver’ is attached to something in your head.” In order to get the joke in real time, your brain has to quickly make a multi-step logical inference that depends on the idea that pronouns imply sex. (The turtle is a “her” iff female, not-female implies not-pregnant, so if the turtle is pregnant, it must be a “her”.) This would seem, pretty straightforwardly, to be a sense in which “a pronoun is attached to something in your head much more firmly than ‘doesn’t look like an Oliver’ is attached to something in your head.” How else am I supposed to interpret those words?

Perhaps it’s not justified to question Yudkowsky’s “I do not know what it feels like [...]” self-report based on generalizations about English speakers in general? Maybe his mind works differently, but dint of unusual neurodiversity or training in LambdaMOO? But if so, one would perhaps expect some evidence of this in his publicly observable writing? And yet, on the contrary, looking over his works, we can see instances of Yudkowsky treating pronouns as synonymous with sex (just as one would expect a native English speaker born in 1979 to do), contrary to his 2021 self-report of not knowing what this feels like from the inside.

For example, in Yudkowsky’s 2001 Creating Friendly AI: The Analysis and Design of Benevolent Goal Architectures, the text “If a human really hates someone, she” is followed by footnote 16: “I flip a coin to determine whether a given human is male or female.” Note, “is male or female”, not “which pronoun to use.” The text would seem to reflect the common understanding that she and he do imply sex specifically (and not some other thing, like being named Oliver), even if flipping a coin (and drawing attention to having done so) reflects annoyance that English requires a choice.

A perhaps starker example comes in the comments to Yudkowsky’s 2009 short story “The Hero With A Thousand Chances”. A commenter (in the guise of a decision theory thought experiment) inquired whether Yudkowsky flipped a coin to determine the protagonist’s gender, to which Yudkowsky replied (bolding mine):

Ha! I tried doing that, the generator came up female … and I realized that I couldn’t make Aerhien a man, and that having two “hers” and “shes” would make the dialogue harder to track.

Sometimes a random number generator only tells you what you already know.

But the text of the story doesn’t say Aerhien isn’t a “man”; it merely refers to her with she/​her pronouns! If Yudkowsky “couldn’t make [the character] a man”, but the only unambiguous in-text consequence of this is that the character takes she/​her pronouns, that would seem to be treating sex and pronouns as synonymous; the comment only makes sense if Yudkowsky thinks the difference between she and he is semantically meaningful. (It’s possible that he changed his mind about this between 2009 and 2021, but if so, you’d expect the 2021 Facebook discussion to explain why he changed his mind, rather than claiming that he “do[es] not know what it feels like from the inside” to hold the position implied by his 2009 comments.)

In the Facebook comments, Yudkowsky continues:

My current policy stance is that anybody who does feel that way needs to get some perspective about how it can be less firmly attached in other people’s heads; and how their feelings don’t get to control everybody’s language protocol or accuse non-protocol users of lying; especially when different people with firm attachments have different firm attachments and we can’t make them all be protocol.

The sheer chutzpah here is jaw-dropping. Someone’s feelings don’t get to control everybody’s language protocol, huh? But—the causal–historical reason we’re discussing pronoun reform at all is precisely to let trans people’s feelings control everybody’s language protocol! The original post is very explicit about this! It says:

Even before considering all gender issues, there is some sense in which somebody saying “help help pronouns attacking” sounds to me like a sympathetic innocent asking to get out from under a bad system, not like a law-deuniversalizer asking for exceptions from a good system.

In terms of important things? Those would be all the things I’ve read—from friends, from strangers on the Internet, above all from human beings who are people—describing reasons someone does not like to be tossed into a Male Bucket or Female Bucket, as it would be assigned by their birth certificate, or perhaps at all.

Okay, so Yudkowsky never thought sex-based pronouns were a good idea in the first place. But the important thing, he says, is that some people (“who are people”, Yudkowsky pleonastically clarifies, as if anyone had doubted this) don’t want other people to use language that refers to what sex they are.

Personally, I have a lot of sympathy for this, because in an earlier stage of my ideological evolution, I was one of those people. (I tried to use an ostensibly gender-neutral nickname and byline for a while in the late ’aughts, and while I never asked for new pronouns, this is probably a matter of Overton window placement rather than any underlying difference in sentiments; it seems pretty likely that my analogue growing up in the current year’s ideological environment would be a trans woman.)

But it’s important to not use sympathy as an excuse to blur together different rationales, or obfuscate our analysis of the costs and benefits to different parties of different policies. “Systematically de-gender English because that’s a superior language design” and “Don’t misgender trans people because trans people are sympathetic” are different political projects with different victory conditions: victory for the de-genderers would mean singular they or similar for everyone (as a matter of language design, no idiosyncratic personal exceptions), which is very different from the ask-and-share-pronouns norms championed by contemporary trans rights activists.

Perhaps it might make sense for adherents of a “degender English” movement to strategically ally with the trans rights movement: to latch on to gender-dysphoric people’s pain as a political weapon to destabilize what the English-degenderers think of as a bad pronoun system for other reasons. Fine.

But if that’s the play you want to make, you forfeit the right to honestly claim that your stance is that “feelings don’t get to control everybody’s language protocol”. If you piously proclaim that the “important thing” is trans people’s feelings of “not lik[ing] to be tossed into a Male Bucket or Female Bucket, as it would be assigned by their birth certificate”, that would seem, pretty straightforwardly, to be participating in an attempt to make it so that “[someone’s] feelings [...] get to control everybody’s language protocol”! Again, how else am I supposed to interpret those words?

There’s nothing inconsistent about believing that trans people’s feelings matter, and that the feelings of people who resent the Stroop-like effect of having to speak in a way that contradicts their own sex-category perceptions, don’t matter. (Or don’t matter as much, quantitatively, under the utilitarian calculus.) But if that were your position, the intellectually honest thing to tell people like Barra Kerr is, “Sorry, I’m participating in a political coalition that believes that trans people’s feelings are more important than yours with respect to this policy question; sucks to be you”, rather than haughtily implying that people like Kerr are making an elementary philosophy mistake that they are clearly not making if you actually read what they write.

(In general, an honest “sucks to be you” from someone whose political incentives lead them to oppose your goals, is much less cruel than the opponent distorting your position to make you look bad to their followers.)

All this having been said, Yudkowsky is indeed correct to note that “when different people with firm attachments have different firm attachments [...] we can’t make them all be protocol”. It’s possible for observers to disagree about what sex category they see someone as belonging to, and it would be awkward at best for different speakers in a conversation to use different pronouns to refer to the same subject.

As it happens, I think this is an important consideration in favor of self-identity pronouns! When different parties disagree about what category something should belong to, but want to coordinate to use the same category, they tend to find some mutually-salient Schelling point to settle the matter. In the case of disagreements about a person’s social sex category, in the absence of a trusted central authority to break the symmetry among third parties’ judgments (like a priest or rabbi in a tight-knit religious community, or a medical bureaucracy with the social power to diagnose who is “legitimately” transsexual), the most obvious Schelling point is to defer to the person themselves. I wrote about this argument in a previous post, “Self-Identity Is a Schelling Point”.

But crucially, the fact that the self-identity convention is a Schelling point, doesn’t mean we have a one-sided policy debate where it’s in everyone’s interests to support this “simplest and best protocol”, with no downsides or trade-offs for anyone. The thing where she and he (which we don’t know how to coordinate a jump away from) imply sex category inferences to actually-existing English speakers is still true! The Schelling point argument just means that the setup of the social-choice problem that we face happens to grant a structural advantage to those who favor the self-identity convention.

Although they’re not the only ones with an structural advantage: a social order whose gender convention was “Biological sex only; transsexualism isn’t a thing; sucks to be you if you want people to believe that you’re the sex that you aren’t” would also be a Schelling point. (Trans people’s developmental sex is not really in dispute.) It’s the moderates who want to be nice to trans people without destroying the public concept of sex who are in trouble!

Still, I think most people reading this post are “moderates” in this sense. Schelling points are powerful. If we’re not culturally-genocidal extremists who want to exclude transsexuals from Society (and therefore reject the “pronouns = sex, no exceptions” Schelling point), isn’t it reasonable that we end up at the self-identity Schelling point—at least as far as the trivial courtesy of pronouns is concerned, even if some of the moderates want to bargain for the right to use natal-sex categories in some contexts?

Sure. Yes. And indeed, I don’t misgender people! (In public. Only rarely in private, when someone’s transition doesn’t seem legitimate or serious to me, and the person I’m talking to doesn’t seem liable to object.) I’m not arguing that Yudkowsky should misgender people! The purpose of this post is not to argue with Yudkowsky’s pronoun usage, but rather to argue with the offered usage rationale that “the simplest and best protocol is, ‘”He” refers to the set of people who have asked us to use “he”, with a default for those-who-haven’t-asked that goes by gamete size’ and to say that this just is the normative definition.”

As I have explained at length, this rationale doesn’t work and isn’t true (even if better rationales, like sincere belief in gender identity, or the Schelling point argument, can end up recommending the same behavior). No one actually believes (as contrasted to believing that they believe) that she and he aren’t attached to gender in people’s heads, despite Yudkowsky’s sneering claim in the comments that he “would not know how to write a different viewpoint as a sympathetic character.”

Again, without attributing to Yudkowsky any conscious, deliberative intent to deceive (because of the tragic human tendency to unconsciously introduce distortions in the heat of a rapid argument), the pants-on-fire audacity of this ludicrous claim to ignorance still beggars belief. As the author of one of the world’s most popular Harry Potter fanfictions, Yudkowsky clearly knows something about about how to simulate alternative perspectives (includes ones he disagrees with) and portray them sympathetically. And he claims to be unable to do this for … the idea that pronouns imply sex, and that using the pronouns that imply someone is the sex that they are not feels analogous to lying? Really?!

Well, I’m not a popular fiction author with thousands of obsessive fans who pour over my every word, but if Yudkowsky claims not to be up to this writing challenge, I’m happy to give him a hand and show him how it might be done—


A cis woman is testifying in court about a brutal rape that horrifically traumatized her. The rapist has since transitioned.

“And then—” says the victim, reliving those awful moments, “and then, he took his erect penis—”

“Objection!” says the defense lawyer. “The witness misgendering my client is prejudicial.”

“Sustained,” says the judge. Then, to the victim: “Her erect penis.”

“Wh—what?” says the victim.

“You will refer to the defendant with the correct pronoun, or I’ll hold you in contempt of court.”

“Oh. O–okay. And then—then, she took her—” The victim breaks down crying. “I’m sorry, Your Honor; I can’t do it. I’m under oath; I have to tell the story the way it happened to me. In my memories, the person who did those things to me was a man. A—”

She hesitates, sobs a few more times. In this moment, almost more than the memories of the rape, she is very conscious of having never gone to college. The judge and the defense lawyer are smarter and more educated than her, and they believe that the man who raped her is now (or perhaps, always had been) a woman. It had never made any sense to her—but how could she explain to an authority figure who she had no hope of out-arguing, if she was even allowed to argue?

“And by ‘man’, I mean—a male. The way I was raised, men—males—get called he and him. If I say she, it doesn’t feel true to the memory in my head. It—it feels like lying, Your Honor.”

The judge scoffs. “You are ontologically confused,” he sneers. “At age 13 I was programming on LambdaMOO where people had their choice of exotic pronouns and nobody thought anything of it,” says the judge. “Denied.”

“O-okay,” says the victim. She doesn’t know what ontologically means, or what a LambdaMOO is. “So then—then sh-she took her erect penis and she—”

She breaks down crying again. “Your Honor, I can’t! I can’t do it! It’s not true! It’s not—” She senses that the judge will imply she’s stupid for saying it’s not true. She gropes for some way of explaining. “I mean—the Court allows people to testify in Spanish or Chinese with the help of a translator, right? Can’t you treat my testimony like that? Let me say what happened to me in the words that seem true to me, even if the court does its business using words in a different way?”

“You’re in contempt,” says the judge. “Bailiff! Take her away!”


Not a sympathetic character? Not even a little bit?

I suspect some readers will have an intuition that my choice of scenario is loaded, unfair, or unrealistic. To be sure, I chose it an unusually clear-cut case for why someone might have a need to use pronouns to imply sex in their own speech. (If the scenario was just talking about someone borrowing a vacuum cleaner, fewer readers would have any sympathy for someone not wanting to concede the trivial courtesy of preferred pronouns.)

But what, specifically, is unrealistic about it? Is it the idea that a trans woman could have raped someone before transitioning? Of course most trans women are not sex offenders—just as most non-transsexual males are not sex offenders—but instances of trans women committing the kinds of sex crimes that are overwhelmingly the provenance of men are a documented thing.

Is it the idea that the legal system would penalize someone for pronoun non-compliance? But this is also an occasionally documented thing, as in one case where a Canadian father was jailed for violating a court order not to refer to his natal-female child with she/​her pronouns. As liberal intellectuals debating optimal communication policies, we usually hope to govern by consensus: we want people to use preferred pronouns voluntarily, rather than being forced. But maintaining a collective norm in the face of those who have their own reasons to object to it, does ultimately require some sort of enforcement. In the vignette above, given the defense lawyer’s objection, the judge does face a forced choice to Sustain or Overrule, and that choice has consequences either way.

In the comments, Yudkowsky continues:

This is not the woke position. The woke position is that when you call somebody “she” because she requested “she”, you’re validating her gender preference. I may SEPARATELY be happy to validate somebody’s gender preference by using the more complex language feature of NOUN PHRASES to construct an actual SENTENCE that refers to her ON PURPOSE as a “woman”, but when it comes to PRONOUNS I am not even validating anyone.

Right, it’s not the woke position. It’s an incoherent position that’s optimized to concede to the woke the behavior that they want for a different stated reason in order to make the concession appear “neutral” and not “politically” motivated. She requested “she” because acceding to the request validates her gender preference in the minds of all native English speakers who are listening, even if Eliezer Yudkowsky has some clever casuistry for why it magically doesn’t mean that when he says it.

Again, I’m not saying that Yudkowsky should have a different pronoun usage policy. (I agree that misgendering all trans people “on principle” seems very wrong and unappealing.) Rather, I’m saying that in order to actually be politically neutral in your analysis of why someone might choose one pronoun usage policy over another, you need to acknowledge the costs and benefits of a policy to different parties, and face the unhappy fact that sometimes there are cases where there is no “neutral” policy, because all available policies impose costs on someone and there’s no solution that everyone is happy with. (Rational agents can hope to reach some point on the Pareto frontier, but non-identical agents are necessarily going to fight about which point, even if most of the fighting hopefully takes place in non-realized counterfactual possible worlds rather than exerting costs in reality.)

Policy debates should not appear one-sided. Exerting social pressure on (for example) a native-English-speaking rape victim to refer to her male rapist with she/​her pronouns is a cost to her. And, simultaneously, not exerting that pressure is a cost to many trans people, by making recognition of their social gender conditional on some standard of good behavior, rather than an unconditional fact that doesn’t need to be “earned” or justified in any way.

You might think the cost of making the rape victim say she is worth it, because you want to make it easy for gender-dysphoric people to socially transition, or because you think it’s dumb that pronouns imply sex in the actually-existing English language and you see the self-identity convention as an incremental step towards degendering the language.

Fine. That’s a perfectly coherent position. But if that’s your position and you care about being intellectually honest, you need to acknowledge that your position exerts costs on some actually-existing English speakers who have a use-case for using pronouns to imply sex. You need to be able to look that rape victim in the eye and say, “Sorry, I’m participating in a political coalition that believes that trans people’s feelings are more important than yours with respect to this policy question; sucks to be you.”

And of course—it should be needless to say—this applies symmetrically. If you think speakers should be able to misgender according to their judgment and you care about being intellectually honest, you need to be able to look a trans person in the eye and say, “Sorry, I’m participating in a political coalition that believes the freedom of speech of speakers is more important than your gender being recognized; sucks to be you.”

Or if you have more important things to worry about (like the fate of a hundred thousand galaxies depending on the exact preferences built into the first artificial superintelligence) and don’t want the distraction of taking a position on controversial contemporary social issues, fine: use whatever pronoun convention happens to be dominant in your local social environment, and, if questioned, say, “I’m using the pronoun convention that happens to be dominant in my local social environment.” You don’t have to invent absurd lies to make it look like the convention that happens to be dominant in your local social environment has no costs.

Really, “I do not know what it feels like from the inside to feel like a pronoun is attached to something in your head much more firmly than ‘doesn’t look like an Oliver’”? Any seven-year-old in 2016 could have told you that that’s just factually not true; if you grew up speaking English in the late 20th century, you absolutely goddamned well do know what it feels like. Did the elephant in Yudkowsky’s brain really expect to get away with that? How dumb does he think we are?!


At this point, some readers may be puzzled as to the mood of the present post. I agree with Yudkowsky’s analysis of the design flaw in English’s pronoun system. I also agree that not misgendering trans people is a completely reasonable thing to do, which I also do. I’m only disputing the part where Yudkowsky jumps to declaring his proposed “simplest and best protocol” without acknowledging the ways in which it’s not simple and not unambiguously the best.

Many observers would consider this a very minor disagreement, not something anyone would want to spend 12,000 words prosecuting with as much vitriolic rhetoric as the target audience is likely to tolerate. If I agree with the problem statement (pronouns shouldn’t denote sex, that’s dumb; why would you define a language that way), and I don’t disagree with the proposed policy solution (don’t misgender trans people in public), why get so hung up on the exact arguments?

(I mean, besides the fact that it’s arguments that matter rather than conclusions, as a completely general principle of correct cognition.)

I guess for me, the issue is that this is a question where I need the correct reasoning in order to make extremely impactful social and medical decisions. Let me explain.

This debate looks very different depending on whether you’re coming into it as someone being told that you need to change your pronoun usage for the sake of someone who will be very hurt if you don’t—or whether you’re in the position of wondering whether it makes sense to make such a request of others.

As a good cis ally, you’re told that trans people know who they are and you need to respect that on pain of being responsible for someone’s suicide. While politically convenient for people who have already transitioned and don’t want anyone second-guessing their identity, I think this view is actually false. Humans don’t have an atomic “gender identity” that they just know, which has no particular properties other than it being worse than death for it to not be recognized by others. Rather, there are a variety of reasons why someone might feel sad about being the sex that they are, and wish they could be the other sex instead, which is called “gender dysphoria.”

Fortunately, our Society has interventions available to approximate changing sex as best we can with existing technology: you can get hormone replacement therapy (HRT), genital surgery, ask people to call you by a different name, ask people to refer to you with different pronouns, get new clothes, get other relevant cosmetic surgeries, &c. In principle, it’s possible to pick and choose some of these interventions piecemeal—I actually tried just HRT for five months in 2017—but it’s more common for people to “transition”, to undergo a correlated bundle of these interventions to approximate a sex change.

On this view, there’s not a pre-existing fact of the matter as to whether someone “is trans” as an atomic identity. Rather, gender-dysphoric people have the option to become trans by means of undergoing the bundle of interventions that constitute transitioning, if they think it will make their life better. But in order for a gender-dysphoric person to decide whether transitioning is a good idea with benefits that exceed the costs, they need factually accurate information about the nature of their dysphoria and each of the component interventions.

If people in a position of intellectual authority provide inaccurate information about transitioning interventions, that’s making the lives of gender-dysphoric people worse, because agents with less accurate information make worse decisions (in expectation): if you have the facts wrong, you might wrongly avoid an intervention that would have benefited you, or wrongly undergo an intervention that harms you.

For example, I think my five-month HRT experiment was a good decision—I benefited from the experience and I’m very glad I did it, even though I didn’t end up staying on HRT long term. The benefits (satisfied curiosity about the experience, breast tissue) exceeded the costs (a small insurance co-pay, sitting through some gatekeeping sessions, the inconvenience of wearing a patch or taking a pill, various slight medical risks including to future fertility).

If someone I trusted as an intellectual authority had falsely told me that HRT makes you go blind and lose the ability to hear music, and I were dumb enough to believe them, then I wouldn’t have done it, and I would have missed out on something that benefited me. Such an authority figure would be harming me by means of giving me bad information; I’d be better off if I hadn’t trusted them to tell the truth.

In contrast, I think asking everyone in my life to use she/​her pronouns for me would be an obviously incredibly bad decision. Because—notwithstanding my clean-shavenness and beautiful–beautiful ponytail and slight gynecomastia from that HRT experiment five years ago—anyone who looks at me can see at a glance that I’m male (as a fact about the real world, however I feel about it). People would comply because they felt obligated to (and apologize profusely when they slipped up), but it wouldn’t come naturally, and strangers would always get it wrong without being told—in accordance with the “default for those-who-haven’t-asked that goes by gamete size” clause of Yudkowsky’s reform proposal, but really because pronouns are firmly attached to sex in their heads. The costs (this tremendous awkwardness and fakeness suffusing all future social interactions involving me) would exceed the benefits (I actually do feel happier about the word she).

I used to trust Yudkowsky as an intellectual authority; his Sequences from the late ‘aughts were so life-alteringly great that I built up a trust that if Eliezer Yudkowsky said something, that thing was probably so, even if I didn’t immediately understand why. But these days, Yudkowsky is telling me that ‘she’ normatively refers to the set of people who have asked us to use ‘she’, and that those who disagree are engaging in logically rude Shenanigans. However, as I have just explained at length, this is bullshit. (Declaring a “normative” meaning on your Facebook wall doesn’t rewrite the actual meaning encoded in the brains of 370 million English speakers.) If I were dumb enough to believe him, I might ask people for new pronouns, which would obviously be an incredibly bad decision. (It might be a less bad decision if done in conjunction with a serious gender transition effort, but Yudkowsky’s pronoun reform proposal doesn’t say “she” is the pronoun for fully-transitioned trans women; it just says you have to ask.) Thus, Yudkowsky is harming a reference class of people that includes more naïve versions of me by giving them bad information; I’m better off because I don’t trust Eliezer Yudkowsky to tell the truth.

(I guess I can’t say I wasn’t warned.)