So, if you've been reading my Less Wrong posts over the last twenty-one months, you'll notice that I've been doing a lot of philosophy-of-language blogging lately! Those posts are:

But, if you didn't know, the reason I've been doing so much philosophy-of-language blogging lately is because I was frustrated because I thought a lot of people were motivatedly getting the philosophy of language wrong for political convenience, specifically around transgender issues. (I think assertions like "trans women are women" and similar need to be argued for on the empirical merits; you can't just define them to be true.)

In particular, "Where to Draw the Boundaries?" was intended as a "stealth reply" (quoting/paraphrasing without attribution in order to reply to the philosophical substance while eliding the political context) to:

I also write a blog about gender issues under a differential-visibility-but-not-actually-Secret pen name. The posts from that blog that are most relevant to my philosophical writing on Less Wrong are:

To allay concerns about premature abstraction of political issues, I'm putting up this non-Frontpageable post in case anyone wants to comment on or ask questions about my object-level motivations without cluttering up our philosophy discussions.

New Comment
23 comments, sorted by Click to highlight new comments since: Today at 4:56 PM

As a response to your recent thread:

Blanchardianism makes claims about trans women which trans women disagree with. For instance that trans women are primarily motivated by autogynephilia.

Given this disagreement, it's possible to analyze it in a cooperative manner, for instance maybe Blanchardians and trans women have observed different pieces of the puzzle, and by sharing information with each other, they can get a more accurate picture of things. However the fact that the disagreement has been stuck for a while now suggests that this is not a very likely explanation.

So another possibility is that one of the sides is trolling the other side. If one of the sides is trolling, then we should expect that side to make bizarre and misleading arguments. And I would claim that the Blanchardian side regularly does so, e.g. with emphasizing menstruation fetishism, or by doing deceptive sampling.

These are the sorts of things that would make me go "You'll Never Persuade People Like That".

You counterargue:

This would seem to be an odd change of topic. If I was arguing for this-and-such proposition, and my interlocutor isn't, themselves, convinced by my arguments, it makes sense for them to reply about why they, personally, aren't convinced. Why is it relevant whether I would convince some third party that isn't here?

What's going on in this kind of situation? Why would someone think "You'll never persuade people like that" was a relevant reply?

In this sort of case, I personally have a bunch of evidence about the validity of autogynephilia theories that makes me unlikely to change my mind a lot on the topic. However, I might be able to provide feedback of what sorts of arguments normatively should change people's minds, among those on the fence.

And in these cases, the arguments are really trollish. So normatively, when people see those arguments, they should go "huh, I guess the Blanchardians are a trollish side in this debate", and un-update away from whatever they've "learned" from the Blanchardians.

Originally I was annoyed by the trollish arguments, since I was trying to argue for autogynephilia theory. However I have mostly given up with pushing for autogynephilia theory, for various reasons including the fact that the people involved are terrible trolls. But it's kind of frustrating because I didn't originally have much way of proving the trollishness, since I had learned that they were trollish through trollish arguments in private emails/private conversations and through their lack of updates on my critiques. So at this point I feel a sense of reassurance and validation when Blanchardians push obviously trollish points and I get to point them out in a low-context way.

In the comments on "Blood Is Thicker Than Water", TekhneMakre writes:

Or it might be that someone was trying to, say, work out how to treat people in ways that are good for them and then you tried to stop them from using "women" to mean, the cluster of social practices around treating women as opposed to men. Which would perhaps be a fault of yours, or at least, I'd want to know why you were doing that

Because I think this motion of circularly defining a category in terms of things-that-are-good-to-be-in-this-category doesn't actually get people what they want; I think it's a comforting casuistry optimized to make it harder to notice that people do not, in fact, have what they want.

If we actually had magical perfect sex change technology, it would never occur to anyone to redefine gender in terms of "working out how to treat people in ways that are good for them." People who wanted to change sex would do so, and everyone else would use the corresponding language because it was straightforwardly true, not because they were trying to be nice.

To be sure, there is such a thing as cluster of social practices around treating women as opposed to men: we call these gender roles, or sex roles. The stricter these roles are and the more they encompass all aspects of life, the more important it is to allocate short codewords for them. If your Society is structured such that the most salient things about "women" is that they're the ones who work in the household and take care of children—then sure, in the context of that Society, males who work in the household and take care of children are women.

But ... how many of the people in modern Society who say "Gender is a social category" actually want that? In practice, I think "Gender is a social category" is more often an excuse, selectively deployed because consensus-based definitions can't be challenged by those who dispute the consensus, and not because people actually only care about the subspace of social practices, with everything else that covaries with sex truly being irrelevant.

motion of circularly defining a category in terms of things-that-are-good-to-be-in-this-category

Someone could say "I am a woman", in context with the goal of communicating something like "please treat me like you treat women, not like you treat men", and I called that "using 'women' to mean, the cluster....". But that doesn't imply they aren't also referring to some underlying reality. Maybe a developmental quirk made their brain develop with a female regulatory mode. Maybe not, but as a child they ended up in a feminine psychic basin of attraction. Maybe they are harmed by their social position because of their gender and want out. In any case, they want to be treated as women. I don't see what's wrong with giving a name to the cluster of people who want to be treated the way women are treated.

That doesn't mean anyone has to do anything; just because you think me treating you some way would be good for you, doesn't mean I believe you, and doesn't mean it would be good for me if I treated you that way.

would use the corresponding language because it was straightforwardly true, not because they were trying to be nice.

Those aren't the only two options. You could use language because it's unstraightforwardly true, and use appropriate precautions. I wonder if you're pushing against calling such a person "woman" because that makes it seem like how you use words like "sky" or "hamster" or "seven" or "tasty" or "beautiful", which is a native representation that you can use unreflectively in your internal planning, whereas really it's a cluster of other peoples' preferences, and if you use them unreflectively you're "unsandboxing" other peoples' preferences into your internal planning.

Maybe we could separate your objections into two parts: you object to calling this cluster "women", and you object to reifying this cluster at all (?).

how many of the people in modern Society who say "Gender is a social category" actually want that?

If we unstrawman from "work in the household and take care of children" to the whole panoply of social attitudes, then maybe a lot of people actually want that? I don't know, it would be interesting to know how much trans people care about treatment vs. some more transcendent type of being believed that they're really women/men, or vs. other stuff; and if someone cares about other people thinking that they're really a women, what exactly does that mean to them etc. and how would that reconcile with biological or psychological facts (noting that some psychological facts are objects of free will, e.g. one could decide to shape ones mind like a woman's).

I don't see what's wrong with giving a name to the cluster of people who want to be treated the way women are treated.

I agree that it's fine to have a name for that, and I agree that it's fine to use the word "women" for that in contexts where it doesn't cause any important confusions. (Well, I usually formulate transness in terms of mimickry—people who are trying to pass as a sex, rather than merely wanting to be treated as a sex without taking any steps to make that look plausible to others—but let that pass for the moment.)

Words can be used in many ways depending on context: we can sometimes say that ice is "water" (because it's H₂O) and sometimes say that it's not "water" (because it's not liquid H₂O), and this mostly doesn't cause confusion. At worst, if I ask for water and you bring me ice, I can say, "No, sorry, I meant 'water' in the sense of liquid H₂O, for which purposes this ice doesn't count even if there's also an ice-inclusive sense of the word 'water'; I expected you to figure that out from context without me having to explicitly disambiguate, but I am explicitly disambiguating now."

If I were to say that about ice, you would probably reply, "Oh, okay." You wouldn't pretend not to understand, or insist "But frozen water is water" or accuse me of causing harm or being philosophically naïve for using the word "water" in the ice-exclusive sense.

In the case of water and ice, people are generally pretty good at this without any explicit philosophy-of-language knowledge. If people in contemporary liberal Society routinely exhibited the same skills about gender—if it were possible to say, "No, sorry, I meant 'women' in the 'adult human female' sense, for which purposes this trans woman doesn't count even if there's also a trans-inclusive sense of the word 'woman'; I expected you to figure that out from context without me having to explicitly disambiguate, but I am explicitly disambiguating now" and for people to reply, "Oh, okay"—then I wouldn't bother doing so much philosophy-of-language blogging.

The problem is that a dominant faction in my Society really doesn't want it to be socially permissible to disambiguate, and so people who want to align themselves with the dominant faction keep introducing distortions into their rationality lessons (like "I ought to accept an unexpected [X] or two deep inside the conceptual boundaries of what would normally be considered [Y] if it will save someone's life") to make it look like the dominant faction is in the right, prompting me to waste years of my life correcting the distortions in increasing amounts of technical detail.

But then it turns out to be surprisingly hard to explain the correct lesson without someone reading me as saying something technically false that I didn't spend the wordcount disclaiming that I don't believe in every single post in the Sequence; the fact that I often take the wordcount to disclaim it ("Thus we can legitimately end up with a non-circular trans-inclusive sense [...]") doesn't help because I can't count on the critic having read everything I've ever written.

When I say, "Look, the hidden Bayesian structure of reality says we should want short codewords to point to clusters of high density in thick subspaces of configuration space; when a codeword is already pointing to such a cluster, and someone proposes redefining the word in order for people to not be sad, that's a problem, because presumably the sad people are sad about the cluster in the territory; if you edit the map instead of the territory, you're just deceiving them, not solving the actual problem that's making them sad", then someone will reply, "Ah, but what if we want to point to the thin-subspace cluster defined by what the sad people want, and furthermore want to use the same word for that as the thick-subspace cluster that the sad people are sad about?"

And I'm like, sure, technically you could do that, as long as you're scrupulously careful to disambiguate which cluster you're using the codeword to point to in cases where the difference matters and the referent isn't clear from context. But when you look at the history of why we're discussing the philosophy of language in this much detail, can you blame me for being suspicious that most people just aren't interested in being scrupulously careful to disambiguate when the referent isn't clear?

Okay, maybe I'm getting where you're coming from now...??

In the hypothetical scenario you quoted from me, it's you who are trying to prevent someone from using a useful concept in a context where it's clear what concept is meant by the word. You seemed to endorse doing so, writing "[TM:]I'd want to know why you were doing that [ZD:] Because I think [reasons]". It seems like in this hypothetical, you're the one making the mistake that you sometimes rightly accuse others of making: trying to stop other people from using a useful concept (using ambiguous but contextually clear language).

Now it sounds like basically you're saying: "If that's what they were doing, then I was making a mistake. I don't believe that that's what those people are doing, I think they're not trying to use words for clusters and instead trying to use words to make people feel a certain way, and I think they're going to make a bunch of destructive mistakes because they're using words not for clusters."

(Which you roughly said in your first response above, but it didn't land for me, maybe because you packaged it with a strawman of the hypothetical position that *is* using "women" for a cluster meaning something about social treatment.)

Since that might be right about what's happening, I'm curious why we're down this rabbit hole, and will go back again and look upthread.

Looking back, it seems like you're using bad examples to argue your point, if I've got you right. This conversation came from a post where you argue that concepts based on niche-adaptedness are less cohesive:

In contrast, "finned swimmy animals" is an intrinsically less cohesive subject matter: there are similarities between them due to convergent evolution to the aquatic habitat, and it probably makes sense to want a short word or phrase (perhaps, "sea creatures") to describe those similarities in contexts where only those similarities are relevant.
But that category "falls apart" very quickly as you consider more and more aspects of the creatures: the finned-swimmy-animals-with-gills are systematically different from the finned-swimmy-animals-with-a-blowhole, in more ways than just the "respiratory organ" feature that I'm using in this sentence to point to the two groups.

It seems to me now that

(1) you're mostly mistaken about such concepts being "less" anything (well, probably in some useful sense there are "more features" explained by phylogeny than by niche-adaptedness, but that doesn't make the latter "less cohesive"),

(2) that example is a red-herring for your point, which is that "concepts" that aren't attached to a cluster *at all* are fake and bad (such as "women" used to "mean" "whatever makes people not sad if I use the word this way").

...Though it's worth noting that it's almost impossible to *avoid* some cluster-related-ness. If you use a word in whatever way makes people not sad, you are going to pick up on some cluster-structure. I think this *is* a good way of *finding one's way* to *new* words; I think we agree that this is a very bad way of *continuously correcting towards territory-reflection*.

If your Society is structured such that the most salient things about "women" is that they're the ones who work in the household and take care of children—then sure, in the context of that Society, males who work in the household and take care of children are women.

In Britain in the 1950s and 1960s (and long before, but I'm limiting this to what I observed in my own lifetime), that was the position of women. Men who took on that role were not "women". They were thought very unusual, even unnatural, and the word "househusband" was coined.

If I understand it correctly, you are willing to use the preferred pronouns for passing trans people, because your System 1 doesn't object. May I interpret it as: "as long as we are not mentioning biology explicitly, I am okay to use the words 'man/woman' in the sense 'what my System 1 perceives as a man/woman', because that is a natural boundary"? (Like, biology is one natural boundary, System 1 judgment in another natural boundary, neither is inherently better than the other, the proper choice depends on context.)

This makes me imagine an opposite example: a non-passing cis person. Like, a fat man with man-boobs, with long hair, cleanly shaven, having somewhat feminine skull shape... who insists on being called "he". Or a bald muscular woman, with some facial hair, wearing male clothes... who insists on being called "she". I assume you would agree that it is proper to use their preferred pronouns (and would be quite offensive not to), because they match the biology (even in situations where biology is not explicitly mentioned), and to ignore the confused feelings of System 1. Am I right?

I wrote Deliberately Vague Language is Bullshit without considering the political implications. Then I read your Reply to The Unit of Caring on Adult Human Females. I like your line about how the "very terms of discourse have been systematically gamed to conflate dissent with unkindness". It seems like a leftist version of the rightest strategy of gaming the terms of discourse to conflate dissent with heresy, treachery and terrorism.

(A reply to gjm, split off from the comments on "There's No Such Thing as a Tree")

would you care either to argue for that principle or explain what weaker principle you are implicitly appealing to here?

No, not really. What actually happened here was, I was annoyed at being accused of not understanding something I've been obsessively explaining and re-explaining for multiple years—notice the uncanny resemblance between your comment ("If I and the people I need to talk to about pumpkins spend our days [...]") and one of my replies from March 2019 (!) to you (!!) ("If I want to use sortable objects [...] If I'm running a factory [...]")—so I fired off a snippy reply daring you to engage with my latest work. It wasn't very principled. (It worked, though!)

Affirm your summary points 1–6.

Suppose someone's legal name, given by their parents, is George, but they hate the way that sounds [...] Suppose our interlocutor actually thinks his name is Fred [...] the big rock in Australia named Uluru, but your interlocutor is an Englishman stuck in the past who insists that it is, and must always be, called Ayers Rock

But these examples are all about which proper name to use, which is not the philosophy-of-language subtopic I've been writing about at all! The communicative function of proper names (an arbitrary symbol/"pointer" to refer to some entity) is different from the cognitive function of categories (whereby different entities are considered instances of the "same kind" of thing)! Why did any of these examples seem relevant to you?

where everyone else seems to be drawing them

Yes, I wrote about this in "Schelling Categories, and Simple Membership Tests". (Pet-topic object-level application: "Self-Identity Is a Schelling Point".)

I no longer think that's quite what's going on, but I do think you're objecting to more than your more nuanced analyses of category boundaries (e.g., in UCAOFD) justify

I think the actual pouncing algorithm is, "If someone favorably cites Scott Alexander's 'The Categories Were Made for Man, Not Man for the Categories', then pounce."

I don't feel guilty about this because that post is utterly pants-on-fire mendacious. To Scott's credit, the disinformation situation there is at least slightly less bad after he added the edit-note at the bottom after I spent Christmas Day 2019 yelling at him, but I think most readers will fail to notice how much the edit-note undermines the grand moral of the post: if "[i]n most cases plausible definitions will be limited to a few possibilities suggested by the territory" (as the edit-note finally admits), then it's not true that one "ought to accept an unexpected man or two deep inside the conceptual boundaries of what would normally be considered female [...] There's no rule of rationality saying that I shouldn't" (as the main text claims). There are rules!

The examples seem relevant to me because they illustrate that language is not used only to predict, that the merits of a particular language-using strategy are not determined only by its impact on predictive accuracy. If language in general has proper goals other than predictive accuracy, why should I think that category-boundary drawing has no proper goal other than predictive accuracy?

I'm not sure exactly what distinction you're appealing to, by the way. In particular, you say "the communicative function of proper names ... the cognitive function of categories" and it's not clear to me whether (1) you're suggesting that proper names are used primarily for communication while categories are used primarily for cognition, or (2) you're saying that your complaints about talk of arbitrariness apply only when thinking about cognition as opposed to communication, or (3) something entirely different. I say that proper names and category boundaries are highly relevant to both communication and cognition, and that some of the examples that apparently bother you most seem much more about communication than about cognition (e.g., being asked to use pronouns that you find inappropriate for the people they refer to, which you say amounts to asking you to lie, clearly a communicative category more than a cognitive one). It's entirely possible that your meaning is something entirely different, of course; please feel free to enlighten me if so.

The pouncing algorithm can't literally be "pounce when people say nice things about TCWMFM" because in the particular case that sparked this particular discussion that didn't happen but you still pounced.

It probably won't surprise you that I don't agree with your description of TCWMFM as pants-on-fire mendacious, and I don't think the edit at the end undermines the foregoing material nearly as much as you say it does. There are rules, but the rules are of the form "if you want your thinking to be optimal in such-and-such a way, you have to think like this" and Scott is talking about words: how to speak, not how to think. Words and thoughts are closely related but they aren't the same, and I agree with Scott that neither rationality nor rationalism includes any obligation to optimize your words for the exact same thing as your thoughts.

For example, suppose that (1) someone decides to use the word "woman" to mean, as precisely as they are able, "adult human being who identifies as female", and that also (2) you are correct in holding that the region of person-space picked out by this definition is less useful for making predictions than one like "person whose gender-related characteristics are, collectively, more like those of the average woman than like those of the average man". (That definition is kinda-circular but in a way that does no actual harm, and it leaves open the question of what "more like" means, but let's suppose that (3) no plausible way of making "more like" concrete gives you anything close to "adult human who identifies as female", though I'm not sure that's actually the case.) Is our hypothetical person doing something rationalists should be ashamed to do?

I claim they need not be. They are (given all the stipulations in the previous paragraph) using language in a way that isn't optimized for prediction, but I see no reason why language-use should be optimized purely for prediction, because the actual uses of language are broader than that.

You might object that they are impairing their ability to think clearly: when they say or hear that someone is a woman, that will lead to certain expectations in their mind, and if they define "woman" in a gerrymandered way then those expectations will be wronger-in-expectation than they need be. It seems to me that any such effect is very small; someone using the word "woman" this way knows that they are doing so, knows that when someone else says "X is a woman" they don't necessarily mean it the same way, can when it seems appropriate think explicitly in terms of Y-chromosomes or hormones or whatever other characteristics might be relevant. The last point is important. When you have some category without a standard precise technical definition, and it's actually important to reason about its members, it usually turns out that what really matters isn't membership in the category as such but something more specific. Is So-and-So "really" a woman? In cases where there's any scope for doubt about that, you will get more useful answers by figuring out what the real question is (do I want to have sex with her? should I let her into my javelin-throwing competition? which public toilet should she be using? what dose of this drug should I give her? is it possible that she's pregnant? etc., etc., etc.) and answering that more directly. So, anyway, I think any thought-impairment here is very small, and reiterate that language has purposes other than facilitating clear thought; if our hypothetical person thinks there are other benefits that outweigh any such loss, it is far from obvious that they're wrong.

You might object that if they use language that way they are failing to communicate maximally informatively: if someone who uses "woman" that way tells you someone is a woman, the information they're giving you is less useful for your prediction-making than if they used the word in a prediction-optimized way. Well, maybe, kinda, but for at least two reasons I don't think this is something it's reasonable to get upset about. 1. Regardless of what they do, the word "woman" is going to have ambiguities around the edges unless the person using that word tells you explicitly and in detail where they draw the boundaries. (The same is true, in less politically-sensitive ways, of many other words. If I call something a mountain or a town, or say someone is stupid or beautiful, or say that a wall is painted green, you don't know exactly how "towns" shade into "villages" and "cities" for me, or what features make someone more beautiful or how "beautiful" shades into weaker terms like "pretty" or "sexy" or "elegant", or how I categorize turquoise-ish colours.) So even if I choose to use the word "woman" in a way that is in fact optimized for your predictive purposes, that's only any use if I first of all tell you exactly how I'm defining that word. Which almost never happens. 2. If I do tell you exactly how I'm defining the word, then no matter what definition I'm using my uses of "woman" are thereafter strictly more informative than if I don't give you that information; i.e., strictly more informative than say 99.9% of uses of the word; I don't think you're then entitled to complain that I could be helping you even more.

None of this is particularly specific to the word "woman". The same is true of, say, the word "fish" as in Scott's initial example. I think Scott is exactly correct when he says that "in no case can an agreed-upon set of borders or a category boundary be factually incorrect". Even though a category boundary absolutely can be not optimized for any reasonable notion of optimal prediction. (And can be bad in other ways; in practice, I think being very different from other people's boundaries is a more important failure mode for the purpose of communication.)

I should maybe add that I don't actually find the stipulations above very plausible. Most social interactions are rather shallow, which means that most gender-related predictions I need to make that might be influenced by what pronoun is being used for a given person have little to do with the fascinating ways in which typical men's and women's bodies or minds differ, which means that if all I know about someone is that they were referred to as "she" or that someone called them a "woman", quite possibly the most useful information I can get from that does concern whether that person would prefer to be treated as male or female. If at some point those other deeper questions start mattering to me, almost certainly by that point I will have met them, seen what they look like, talked to them, etc., and at that point most of the information I have about their personality and preferences and so forth doesn't come from their Official Gender at all. In which case, during the period when their Official Gender is actually of much use for my predictions, the predictions I'm needing to make are in fact mostly about "how they identify".

I'm not sure exactly what distinction you're appealing to

Thanks for asking! More detail: if you're building a communication system to transmit information from one place to another, the signals/codewords you use are arbitrary in the sense that it doesn't matter which you use as long as the reciever of the signals knows what they mean (the conditions under which they are sent).

(Well, the codeword lengths turn out to matter, but not the codewords themselves.)

If I'm publishing a weather report on my website about whether it's "sunny" or "cloudy" today, it doesn't matter whether I give it to you in JSON and English ({"weather": "sunny"}/{"weather": "cloudy"}), or HTML and Spanish <h1>soleado</h1>/<h1>nublado</h1>: whichever one I choose, you can use it to make the same predictions about what you'll experience when you go outside.

In contrast, the choice of where I draw the boundary between what constitutes a "sunny" vs. a "cloudy" day does make a difference to your predictions. What a signal like {"weather": "sunny"} means is different if I only send it when there's not a single cloud in the sky, or if I only don't send it when it's completely overcast.

The choice between Fred and George, or between Uluru and Ayers Rock, is analogous to the difference between {"weather": "sunny"} and <h1>soleado</h1>; I consider the psychology of why a human might prefer the sound of one name over another to be out of scope of the hidden-Bayesian-structure-of-cognition thing I've been trying to talk about for the last thirty-eight months.

being asked to use pronouns that you find inappropriate for the people they refer to, which you say amounts to asking you to lie

Not "lying" exactly, but rather that actually-existing-English-speakers naturally interpret she and he as conveying sex-category information, even if this seems like a weird or bad design, if we were somehow in the position of designing a natural language from scratch. (You could propose that it shouldn't work that way, but the language is already "widely deployed"; the act of proposing a change doesn't automatically update how 370,000,000 people interpret their native language.)

That's why trans people care about being referred to with the correct pronoun in the first place! If the pronouns didn't convey sex-category information and you could just choose one or the other arbitrarily with no difference in meaning, then there would be no reason to care, unless you had an æsthetic preference for the voiceless palato-alveolar fricative, or for words with two letters rather than three.

because in the particular case that sparked this particular discussion that didn't happen

Yes, it did: did you see the "Related: Timeless Slate Star Codex / Astral Codex Ten piece" link at the bottom of the post? That's why I commented. Do I feel a little guilty now that the OP author expressed dissatisfaction with the semi-derailed thread? A little! But, ultimately, I think it's morally right to pay the cost of being a litle annoying from time to time to try to halt the spread of this bonkers meme that continues to exert influence to this day! (I'm guessing that in the counterfactual where "... Not Man for the Categories" was never published, but Eukaryote still wrote the (excellent) post about trees, it wouldn't have contained the advice to "Acknowledge that all of our categories are weird and a little arbitrary", which is a very different and (I claim) much worse pedagogical emphasis from what was laid out in the Sequences.)

Eliezer Yudkowsky once wrote about dark side epistemology: wrong lessons about how to think that people only have an incentive to invent in order to force a conclusion that they can't get on the merits. That's what I think is happening here: if we actually had magical sex-change technology such that people who wanted to change sex could do so, then everyone else would use the corresponding language because it was straightforwardly true, and no one would have invented this deranged "gerrymandered categories are Actually Fine" argument in the first place!

neither rationality nor rationalism includes any obligation to optimize your words for the exact same thing as your thoughts.

But—don't you want the language you speak to your friends to be the same as the language you use to organize your own thoughts? How can you accept a wall between the world you see, and the world you're allowed to talk about? Doesn't your soul die a little bit?

"person whose gender-related characteristics are, collectively, more like those of the average woman than like those of the average man". (That definition is kinda-circular but in a way that does no actual harm

Tangential, but—what's even the motivation for the circularity here? What's wrong with "adult human female"?

Okay, I get it: we want to be trans-inclusive. But the clean way to do that is, "Adult human females, plus sufficiently successful male mimics thereof in the context that I'm using the word." (See the explanation of mimickry and story about robot ducks in "... Optimized for Deception.") We can accomodate mimics in the domain that their mimickry is successful without trashing our ability to acknowledge the existence of the original thing!

ambiguities around the edges unless the person using that word tells you explicitly and in detail where they draw the boundaries. [...] exactly how "towns" shade into "villages" and "cities" for me [...] how "beautiful" shades into weaker terms like "pretty" or "sexy" or "elegant", or how I categorize turquoise-ish colours

Sex actually seems significantly disanalogous to all of these examples, because municipality size, beauty, and color are all continuous (you know, like, for all ε, there exists a δ, such that if |color1 − color2| < δ, then |greenness(color1) − greenness(color2)| < ε), whereas sex is functionally binary: there's a morph that procduces eggs, and a morph that produces sperm, but no continuum between them that produces a continuum of intermediate-sized gametes. The existence of various intersex conditions (which some authors call "disorders of sex development"), and cross-sex hormone replacement therapy, don't substantially change this picture, because they're the result of some specific thing "going wrong" with one of the two-and-only-two evolved developmental processes: you end up with various DSDs and females-on-masculinizing-HRT and males-on-feminizing-HRT each being their own tiny clusters in configuration space that you can draw a category boundary around. True, there is going to be a continuum of HRT dosages (or, say, the degree of how severe a polycystic ovary syndrome case is, if you want to count that), but my point is that the taxonicity means that we don't need to specify edge cases in detail in order to stick a category boundary between the taxons: males-on-feminizing-HRT aren't part of the female taxon! They just aren't!

whether that person would prefer to be treated as male or female.

What does that even mean? (Also, isn't that kind of sexist?) If Official Gender doesn't matter once you get to know someone, why would people have this mysterious deep-seated preference to be "treated as male or female"?

Definitions aren't generally arbitrary in communication for reasons similar to why they aren't arbitrary in cognition; if I define "woman" to mean "adult female human" (for some possibly-contentious definition of female" I will communicate more effectively than if I define it to mean "adult female human who is not called Jane, OR 4x2 lego brick" (same definition of "female"), even if everyone knows what definitions I am using. I think the distinction that's doing the actual work isn't between communication and cognition, but between proper nouns (where the question is how you assign names to specific single things) and common nouns (where there are also boundary-drawing questions). Anyway, no matter; as I said, the point of what I said about proper nouns was merely to establish that language is not used only for prediction-optimizing.

Of course people take "he" and "she" to convey sex-category information! But (because different people draw boundaries in different places) that doesn't mean that what e.g. "he" conveys is anything as specific as "part of the cluster in thingspace that contains typical men, with Zack's preferred metric on thingspace", which is what it would have to be for your use of terms like "lie" to be right.

My apologies for missing the link to TCWMFM! I retract my claim that eukaryote didn't mention it. For what it's worth, my guess is that without TCWMFM eukaryote would in fact have said much the same things as they did with TCWMFM; the idea that category boundaries are "a little bit arbitrary" did not originate there. Philosophers have been saying "well, it all depends on what you mean by ..." for ages. And, despite your many thousands of words on the subject, I remain entirely unpersuaded that it's "bonkers", and it still looks to me as if your assertions that it's "bonkers" are based on entirely unrealistic ideas about how language ought to be used.

Sure, it would be nice if the language I use to talk were perfectly optimized for thinking. But it isn't, and it never will be, and any possible conflict between tact and prediction-optimality in gender-words is far down the list of reasons why. Because the primary requirement for the language I use to talk to my friends is that I should be able to use it to talk to my friends, and it happens that the language my friends mostly talk is English, which like every other natural language is a pile of kludges and path-dependent arbitrary choices. And also because different people define things in slightly different ways, so no matter how carefully optimized my notion of (say) "woman" might be, when I'm talking with other people that doesn't matter because they are not going to mean the exact same thing by it as I do. Even if nothing we are saying has anything to do with trans people at all.

To be more precise: no, I don't "want" the language I speak with my friends to be the same as the language I use to organize my own thoughts. The way I organize my own thoughts is necessarily idiosyncratic; it has to work for my brain, thinking the kinds of thoughts that I happen to need to think, given the things that I happen to know; fortunately my friends are not my clones, and their brains and thoughts and knowledge are different; so there is no possible way for the language we speak to be optimal for all of our thoughts.

What's wrong with "adult human female" is that it just pushes whatever difficulties there might have been in "woman" over to "female". I'm perfectly happy agreeing that women are adult female humans (or adult human females, but that's an ugly locution) and I think everyone would be if that set of words hadn't been adopted by some people specifically as a way of objecting to some things trans people want to say and think. But some people would want "female" to mean "person with no Y chromosome" and some would want it to mean "person nearer to the cluster in Zack-concept-space that contains typical women" and some would want it to mean "person who considers herself female". Again, the circularity is perfectly benign; take any definition of "woman" that is anywhere near any definition any sane human being uses, apply the cluster definition, and iterate a few times, and it will converge very rapidly.

I've no idea why you say at that point "I get it, we want to be trans-inclusive"; I was referring to your preferred understanding of the word "woman" (in so far as I understand it right, which quite possibly I don't) and so far as I can tell you don't particularly want that to be trans-inclusive, to whatever extent being trans-inclusive conflicts with prediction-optimality. But, for what it's worth, I don't think that the clean way to be trans-inclusive is what you say; maybe that's one clean-ish way to be kinda-trans-inclusive but it seems to me it doesn't do either of those things terribly well and that term "sufficiently successful" has an awful lot of stuff swept under it.

Indeed sex is functionally binary (though I remark that most of the individual consequences of sex that actually matter for most human interactions are very much not binary, although if you take them collectively you can pretty much recover the binary classification them them), and indeed it is unlike (say) size in that respect. But I don't think that's relevant to the point I was making, which is simply that "woman" and "town" both have boundaries that different people draw in different places, and therefore the only circumstances in which you are going to get anything like the optimal communication you want are those where we say very explicitly where we draw the boundaries, and once we do that we are communicating far more precisely than the norm however much you dislike the specific boundaries I draw.

The rest of that paragraph only has any chance of being right if we make the assumption, which I explicitly reject, that the only valid way to delineate the meanings of words is by picking something like minimum-volume clusters in concept-space. No! It isn't! The only valid way to delineate the meanings of words, outside technical discussions where you make explicit and precise definitions, is to try to match your usage of each word to how the other people you're trying to communicate with use it. And, like it or not, the way language actually works does not give you anything like a guarantee that common usage will do a very good job of optimizing compactness in concept-space. If your goal is to communicate with other people, it doesn't matter whether your preferred usage corresponds to something you consider a natural cluster in concept-space unless everyone else is also choosing their definitions that way. Which they aren't.

(And I am very unconvinced by the assumptions you are making about the proper metrics in concept-space, for reasons I already explained about as well as I think I can explain them. If all I know about you is what gender you, or someone else, says you are, then it is most likely that the appropriate metric on concept-space for the purposes of whatever interactions we are going to have while that remains the case is one that cares mostly about things like what words you prefer me to use when talking to you. And once we start having interactions for which more subtle gender-related differences matter, what I've been told about your gender probably no longer has much impact on my predictions because I know a bunch of other things that mostly screen that off.)

"Treating someone as male or female" means, in practice, treating them in ways that they interpret as indicating that I regard them as male or female. This is not a thing with a precise definition and it probably couldn't be. But as long as we have (e.g.) different pronouns for male and female, different public toilets for men and women, different social expectations for men and women (maybe we shouldn't have those! but they are there none the less), it will be hard to avoid some degree of pigeonholing and many people will care which pigeonhole you seem to be putting them into.

Okay. I give up. I really liked your 11 May comment, and it made me optimistic that this conversation would lead somewhere new and interesting, but I'm not feeling optimistic about that anymore. (You probably aren't, either.) This was fun, though: thanks! You're very good at what you do!

OK. I'm not sure to what extent I'm supposed to take the last comment as an insult ("you're very good at emitting sophistical bullshit" or whatever), but no matter :-).

I don't know that I was feeling optimistic, but I had had some hopes that you might be persuaded to engage with what seem like key criticisms rather than just dismissing them. But you certainly should feel obliged to engage with someone you aren't finding it worthwhile arguing with. [EDITED to add:] Er, oops, of course I mean you shouldn't feel obliged.

By the way, I see that at least one earlier comment of yours in this thread has been downvoted; it wasn't by me.

But—don't you want the language you speak to your friends to be the same as the language you use to organize your own thoughts? How can you accept a wall between the world you see, and the world you're allowed to talk about? Doesn't your soul die a little bit?

I am not quite sure such unity of language is possible.

My thoughts really aren't organized in words. (I hope to eventually write some posts about it, but:) My general understanding is that I have some underscrutinized mental activity that hosts a whole lot of models, prediction engines, and such; and that additionally there are many speech-generators that can take these wordless representations and convert them into words, sentences, texts of various languages. (into an echo in a verbal loop; into a sequence of muscle contractions that hit keyboard buttons; into a sequence of muscle contractions that produce sounds of speech.)

I can readily name a dozen distinct generators, and assume there are more. They adjust for a lot of social context — politeness, friendship, feelings of acceptability of topics. Whether I'm lecturing or questioning or bantering; who I am talking to and what words they will find persuasive. Whether I'm having fun or trying to explain or trying to formalize concepts or trying to express feelings. Heck, I speak 2.03 languages; empirically their generators borrow a lot from each other, but otherwise they're as different as it goes.

When I'm writing notes, I again have many modes of speech-generation. There's the babble mode; there's the mode of writing a coherent post that adresses something; there's a mode of taking extremely verbose what-is-happening notes; there's a mode of trying to write a zettelkasten note. These modes turn on different heuristics for {what should I do next}; they produce different results, leave different mental impressions.

These langauge generators are optimized-ish for different purposes, of course, and----I wouldn't have it any other way? Optimal ways to express my thoughts in private, to talk with my friends, and to give speeches are different, for reason that — I am tempted to argue — is implacable: these situations' have significantly different spaces-of-consequences for same sequences-of-words; of course I would learn to optimize them separately, and therefore use somewhat different languages for them.

Likewise, writing modes are useful for decidedly different outcomes: I won't get a good public-facing text by babbling; formal public post format is way too slow and stifling to represent an ongoing conversation with maximal fidelity; such representations are not optimized for later skimming; a (permanent) zettel note cannot help me catch subtle in-the-moment intuitions and produce insight.

(On that perspective, I would in fact think of English not as a single language, but more of as a generic protocol that allows an amount of distinct ?speech modes? to be implemented on top of it.)

(Disclosure: I have never seriously considered whether maintaining these multitudes is a price I might not want to pay. It definitely feels like the cost of keeping track of context and hosting multiple language modules is totally worth being able to optimize for so many things in separation, but I might be lost in the haze of the argument.)

So my answer here would be:

  1. Best language to organize my thoughts probably can't be the best language to talk to my friends, and I think I would rather have both and translate than have one that's inferior on both counts.
  2. The world I see and the world I can talk about are already different (none of the langauges above have words for certain concepts I observe and want to talk about); this is a barrier I bash my head about whenever I am trying to model-build, to make my understanding of the world crisper by pinning it down with labels. This is perhaps annoying but seems inevitable.
  3. Words I can use in private and be understood by myself, and words I can use to talk to any friend, are likewise different; translation between our wordless world-models does not come for free. I am annoyed about that but neither I can will understanding into others' heads (except by talking) nor do I want to stop privately using words nobody but me already understands.
  4. Not being allowed to talk about certain concepts among my favorite ingroup is indeed quite unpleasant, separately from concerns above.

I think the signficance of the symbols might not be so monolithic or atleast interesting dynamics can be discovered when differences onthem are studied and that concepts would benefit from being divided into smaller chunks for this discussion.

Say that there is a community that has access to 3 kinds of berries and has 2 types of members. There is a "poisonous berry" that kills anybody that eats it. Then there is a "safe berry" that anybody can eat. Then third type kills one type of member while the other type can eat fine. Say that amout of people that are "allergic" is a small minority (say 1%). If the safe and the allergen berries are very close in coloration and identifiable characteristics I could understand why the hunter gatherers would lump them into the same concept. However using this concept in cooking etc will probably result in food that would be unreasonably risky for a allergic person to attempt to bite.

Now anybody could be allergic to anything, it might no tbe feasible to be paranoid about the environemnt and make all the catalog all the tiniest differences. But people can also get lucky in that if the safe and allergen berry are saliently different then the allergics looking after their own survival would be relatively straightforward (gatherers might put them in different buckets to begin with etc). I am reading/guessing that the "information conveyance" property is important in that it is value/survival ambivalent while allowing people to maximise for value/survival.

In a way because anybody could have a lot of weird idiosyncraties then critising that society does something inconvenient for you is not neccesarily a basis to change society per se. On the other hand if a little eye-squinting could prevent poisonings, not bothering to do so could be extremely cruel. If the squinting feels like pointless pedantry to the non-allergic gatherer, it can feel like inefficiency. I could also see that one way to deal with the situation is to forbid allergic persons to eat any food with berry. This means in effect that berry-pickers don't benefit allergic persons nutrition and the amount of dishes with berries mixed in could become a point of contention.

However the main theme is that which for some actor is a non-signficant thing that can go either way is a very signficant thing with high impact to another and having agent-principal relationships over these kinds of situations. Treating the allergen berry to have one single universal lethality level is likely to just lead to confusion and disarray. While the downsides to getting gender wrong isn't always lethal, there is difference of imposing/accpeting a cost rather than not knowing it exists.

I appreciate this set of links grouped together being made because, given the similarity between them, having them grouped together seems useful.

I also think that every one of those posts is probably too long. Specifically, longer than they need to be. I consider this evidence in favor of 'keeping politics out of lesswrong does help with 'rationality''.

I consider this evidence in favor of 'keeping politics out of lesswrong does help with 'rationality''.

Sorry, I'm not following the reasoning here; can you say more? Why that specific hypothesis, rather than "ZMD's writing is too long-winded in general"?


  • Charity (see B.)
  • I didn't generalize that much.
  • Confidence (distinct from probability)*
  • Decision making heuristic (also distinct from probability)**


  • I've read stuff that you've written that didn't seem bad the same way.

Reading Where to Draw the Boundaries?***

  • It's long, hard to read/understand, and seem kind of wrong. Sometimes this is because the author is bouncing between (two) things that conflict, like: 'I think I'm right about this interpretation' and 'multiple interpretations are possible'. (This confusion that might be fixable by breaking things up more.)
  • Given that the post is about a specific thing, maybe it's written in a way that is really hard to read because references to thing have been moved/altered. (I could make some of the same points just using numbers and functions. An infinite number of series**** begin with 1, 4, 9, 16, 25, then don't follow up with 36, 49, etc. And yet, upon seeing those numbers you may see a pattern, and expect the 36, and the 49. And if "our brains know what they're doing" there's a reason for that. (But beware the Law of Small Numbers.))
  • It's also like a dialgoue, but without the two sides delineated, or the reader doesn't get to read half the conversation and it's really confusing because the rebuttals are confusing on their own.

The issue with removed references/abstracting politics has been mentioned before. On it's own it's slightly convincing. Looking these specific examples, it seems like it's horribly accurate.

*Like probability, but with wide error bars.

**Do more general hypotheses 'need' more evidence, or less?

***The word "the" might be out of place in that title. (And borders are drawn on maps. And they're messy around the edges.)

****Similarly, an infinite number of functions have the properties that f(1) = 1, and f(2) = 4, and...

I recently read Algorithms of Deception! and claim that it is of appropriate length.

[+][comment deleted]1y1