The rationalist community seems complicit in dishonesty about trans matters. Historically, I also can’t claim innocence from engaging in the dynamics you’ve mentioned in the post, including towards you, though I think I have become better at not doing that.
Some various comments:
Okay, so trans people aren’t delusional about their developmental sex.
There is a tension between these two different parts of your post, since I think Phil Illy is kind of describing autogynephilia as a delusion in the book he wrote, e.g. “Autogynephilic people initially feel feminine or like a woman for short spurts, often in association with crossdressing. .. Autogynephilia can also create autogynephilic phantom shifts, the sensation of having female-typical phantom anatomy such as breasts, a vulva, wide hips, or long hair … Males who continually feel feminine or see themselves as women can be considered to have a transfeminine gender identity. … This desire to see themselves as feminine also shifted their self-perception in that direction …”.
In one taxon, the “early-onset” type, you have same-sex-attracted males who have been extremely feminine (in social behavior, interests, &c.) since to early childhood, in a way that causes social problems for them—the far tail of effeminate gay men who end up fitting into Society better as straight women.
“Social behavior, interests, &c” is as far as I can tell gender conservative propaganda that Blanchardians made up in order to make it fit with the grand HBD narrative about gender differences. (People-things, Mahalanobis D, etc..) If you look at e.g. the correlation between Richard Lippa’s MF-Occ interests and gender identity among homosexual males, you get basically no correlation:
I think the aspect that actually correlates with gender issues is actually mainly down to appearance, both in terms of innate physique and acquired characteristics like an interest in makeup. (Possibly nonlinearly interacting with androphilia—needs more investigation.) Though I will have to finish up my analysis and get back to you on that.
The task force writing this section was lead by Ken Zucker, who is a personal friend of Blanchard and Bailey, right?
There could be situations in psychology where a good theory (not perfect, but as good as our theories about how to engineer bridges) would be described by (say) a 70-node causal graph, but that some of the more important variables in the graph anti-correlate with each other. Humans who don’t know how to discover the correct 70-node graph, still manage to pattern-match their way to a two-type typology that actually is better, as a first approximation, than pretending not to have a theory. No one matches any particular clinical-profile stereotype exactly, but the world makes more sense when you have language for theoretical abstractions like “comas” or “depression” or “bipolar disorder”—or “autogynephilia”.
I think there’s a bunch of things to be said about categories which doesn’t really get said because people either short-circuit to “correlation-clusters!” or vaguely go ” 🤨 I don’t really think there is One True Essence of the trait you mention”.
I don’t understand depression or bipolar disorder well enough to comment on them, but consider the case of comas. First, part of how to deconfuse GCS is to notice that it makes a lot more sense if you invert it: it doesn’t measure comas, it measures consciousness. Comas are defined as “a prolonged state of deep unconsciousness, caused especially by severe injury or illness”, i.e. roughly the negation of consciousness, and so it makes sense that you would measure comas by a consciousness scale.
And from this perspective, the latent variable/factor model makes a lot of sense. Why does being “oriented to time, place and person” correlate with “obeys commands”? Because they both require you to be conscious. And you don’t have two different consciousnesses in your head, so representing it as a single variable makes sense. Yes, it’s true that mechanistically, consciousness draws on a whole shitton of different parts of your nervous system—but these different parts contribute to a unified computational system, and therefore have essentially identical effects on the symptoms of consciousness.
(Well, most parts probably have fairly limited effects on fairly narrow things. But what I mean is that the reason the GCS works is because there are a set of parts that are sufficiently necessary that they knock you out across a broad range of conscious functions.)
When I emphasize the value of causality in thinking about categories, it’s partly with cases like this where a reasonable yet loose understanding justifies the common-sense categories. But it is also very much in order to recognize the cases where categories don’t work this way—see for example my discussion with ChatGPT on trauma measurement here. By thinking carefully causally about why one creates a category, one can come up with better definitions and measurements.
(Bonus: results from trauma survey. Has strongly upped my estimate for how much of a thing traumas are. Though there’s issues of sampling; I don’t think I have much self-selection going on due to my filtering methods, but plausibly the women who find it worthwhile to answer surveys for money on the internet are going to be non-representative.)
Not if you buy into the masochistic CEO theory, as most people seem to:
I do have a lot of uncertainty about what the True Causal Graph looks like, even if it seems obvious that the two-type taxonomy coarsely approximates it. Gay femininity and autogynephilia are important nodes in the True Graph, but there’s going to be more detail to the whole story: what other factors influence people’s decision to transition, including incentives and cultural factors specific to a given place and time?
I’ve been raging against parsimony for a while, so I thought I should mention a notion in favor of parsimony that I have been thinking of for a while:
In some sense there’s nothing wrong with a utility function that gives 9001 utilons for being female if there are an even number of people alive, and 9001 utilons for being male if there are an odd number of people alive. Except this would be kind of impossible to act on and would flip all of the time.
So in practice, utility functions don’t look like this, but instead decompose into e.g. ridges of optimal states along dimensions of indifference, with gradual falloff as one distances oneself from the ridge.
In the case of autogynephilia, this looks like a ridge of being a woman, optimal against wildly varying external environments, with (possibly partly due to GAMP/AGAMP stuff) continuous falloff based on lesser degrees of feminization.
But for a different etiology, the shape of the utility function might look very different. Issues with gender norms might lead to a ridge that passes through being female as optimal in the current environment, but in an environment where the gender norms were flipped, the optimal sex would also be flipped. And an environment with weaker gender norms might be equally desirable to being the opposite sex in the current environment.
I took this as confirmation of my expectation that alleged “autogynephilia” in women is mostly not a thing—that normal women appreciating their own bodies is a qualitatively distinct phenomenon. When she didn’t know what I was talking about, my friend mentioned that she also fantasized about being a hot girl. After I went into more detail (and linked the TVTropes page), she said she didn’t understand why anyone would want boobs. Well, why would she? But I think a lot of a.m.a.b. people understand.
I find this phrasing/situation weird to parse. Taken literally, it sounds confusing as most cis women like having breasts and it’s far from unheard of for them to mention having breasts as one of the best parts of being female. So taken literally this feels like you are overupdating on a single unusual woman.
(Reminds me of the time where I posted quotes from women talking about what they liked about being female in a gender critical subreddit, and the GCs were like “this sounds so autogynephilic!”)
But I think maybe by the context, both you and her are using “want boobs” to mean something like “want to play with your own breasts in private for Personal Enjoyment”? It’s just a confusing way of phrasing it to me. (Do you have more logs/context from this conversation?)
… If I wanted to, I could cherry-pick from my life to weave a more congruent narrative about always having been a girl on the inside. …
… HSTS-taxon boys are identified as effeminate by others. …
I still get kind of bothered by how these discussions of transfeminine femininity are always measured with respect to cis women or HSTSs, rather than with respect to cis men. Like from a political perspective, it makes sense—if you are trying to fit into a category of “woman”, or trying to do some sort of good trans/bad trans thing, then these are the natural points to compare relative to.
But Blanchardians emphasize that their theory is about etiology—which would be the things that distinguish transfems from cis men—rather than politics. Clearly we can see that this is bullshit and they really care about politics, but the fact that they shift the discussion to etiology interferes with having a productive discussion about these things.
Moldbug contends that the triumph of progressivism is bad insofar as the oligarchic theocracy, for all its lofty rhetoric, is structurally incapable of good governance: it’s not a coincidence that all functional non-government organizations are organized as monarchies, with an owner or CEO[14] who has the joint authority and responsibility to hand down sane decisions rather than being hamstrung by the insanity of politics (which, as Moldbug frequently notes, is synonymous with democracy).
I’m not sure this is true. Microsoft has traditionally been known for having a dysfunctional, internally adversarial culture. But it no longer does, and from what I’ve been told, it’s directly related to the takeover by progressive “diversity/emotional intelligence” culture.
I mean of course it still has a CEO and lots of hierarchy and inequality, so in a sense it’s still not progressive (or perhaps rather one should say, still not socialist).
I should say that when I first learned about the Big Five Aspect Scales, I was enthusiastic about them and did a whole bunch of stuff related to them, but over time I came to the conclusion that I was just wading around in meaningless variance-shuffling.
In another post that afternoon, I acknowledged my right-wing influences. You know, you spend nine years reading a lot of ideologically-inconvenient science, all the while thinking, “Oh, this is just interesting science, you know, I’m not going to let myself get morally corrupted by it or anything.” And for the last couple years, you add in some ideologically-inconvenient political thinkers, too.
One gets morally corrupted by right-wing science because right-wingers put in lots of subtle poor design decisions, vague implications, motivated stopping, incorrect epistemology, strange research questions, etc., in ways that are optimized for pushing right-wing views. No one aspect of it is unforgiveable, but the fact that it is a fractal of manipulation is what corrupts you infinitely.
It was as if the part of people that talked didn’t have a problem representing their knowledged using a graph generated from a variable ordering that put “biological sex” closer to last than first. I didn’t think that was what the True Causal Graph looked like.
This circles back to my point about causality and categories I mentioned earlier in my comment. And it also gets at some issues I have with some people’s characterization of the g factor. And with my issue with the sequences on words.
Within psychometrics, this debate is termed “reflective vs formative”. Basically, while the Sequences acknowledge that the underlying distribution matters for how to define words, they still treat category membership as caused by having the properties that we use to exemplify the category, not as causing it.
In IQ research, this becomes treating g as a composite of abilities measured by IQ tests. And in sex differences this becomes putting biological sex at the end of the causal network, rather than the beginning.
And like of course there are various senses in which this view is not totally false. g/biological sex is an abstraction of some more complicated underlying dynamics. But at least in the case of sex (and I suspect also for g) when you dig into what is abstracted over, then structurally in many ways it still satisfies the causal property of coming at the beginning of the causal network, rather than the end.
So basically your objection seems right but it gets at the core issue of how formativity is generally accepted as the way to create categories but actually reflectivity is usually a more natural view.
The debate between formativity and reflectivity usually seems kind of academic, partly because psychometricians are already wading around in highly abstracted data. I didn’t properly get its importance when I first got introduced to it, only later as I thought through it carefully in lots of different contexts. I still don’t have a great way to communicate the underlying insights because most ways of talking about it just sound too weird. I guess to some extent it’s also that people rely so much on their intuitive categorization that they don’t bother thinking through the rules for how categories work in practice.
I added that the worst part was that the “trans women are disgusting blokes in dresses who are invading my female spaces for nefarious purposes” view was basically correct. It was phrased in a really dismissive manner. But words don’t matter! Only predictions matter!
There’s a thing in some GC spaces where they say that what autogynephilia is really about is getting off to transgressing boundaries. I usually interpret this to be a claim that autogynephilia is a member of the courtship disorders factor, a group of sexual interests that includes e.g. exhibitionism, voyeurism, and paraphilic rape. I think this prediction is outright false.
Human psychology is a very high-dimensional vector space. If you’ve bought into an ideology that says everyone is equal and that sex differences must therefore be small to nonexistent, then you can selectively ignore the dimensions along which sex differences are relatively large, focusing your attention on a subspace in which individual personality differences really do swamp sex differences.
I think an underrated issue is that it is not gender differences per se that are critical, but also how they add up.
Like take sociosexuality. Men are much more attracted to having sex with many different strangers than women are. But there is overlap in the distributions of sociosexual desire, yet my impression is that despite this overlap, men’s and women’s experience with sociosexuality is still totally different. Why?
Supply/demand and conflicts. If you are in the middle—an especially sociosexual woman or a not that sociosexual man—then you might want some casual sex, but the feasibility of this is gonna depend massively on your environment. The man is probably gonna do better by focusing on his comparative advantage, credibly promising intimacy to a long-term partner, whereas the woman can get as much sex as she wants, though possibly triggering all sorts of social dynamics and resentment as a result.
Blanchardians are IME incapable of reasoning about this sort of thing, because whenever the superficial outcomes look different, they jump to the conclusion that the underlying causes must be distinct, rather than that continuous variation can add up to give highly distinct results.
In “Interpersonal Entanglement” (in the Fun Theory Sequence back in ’aught-nine), Yudkowsky had speculated that gay couples might have better relationships than straights, since gays don’t have to deal with the mismatch in desires across sexes. The noted real-life tendency for AGP trans women to pair up with each other is probably partially due to this effect[22]: the appeal of getting along with someone like you, of having an appropriately sexed romantic partner who behaves like a same-sex friend. The T4T phenomenon is a real-life analogue of “Failed Utopia #4-2″, a tantalizing substitute for actual opposite-sex relationships.
I feel like one can’t discuss this without bringing up gynandromorphophilia as a correlate of autogynephilia.
The rationalist community seems complicit in dishonesty about trans matters. Historically, I also can’t claim innocence from engaging in the dynamics you’ve mentioned in the post, including towards you, though I think I have become better at not doing that.
Some various comments:
There is a tension between these two different parts of your post, since I think Phil Illy is kind of describing autogynephilia as a delusion in the book he wrote, e.g. “Autogynephilic people initially feel feminine or like a woman for short spurts, often in association with crossdressing. .. Autogynephilia can also create autogynephilic phantom shifts, the sensation of having female-typical phantom anatomy such as breasts, a vulva, wide hips, or long hair … Males who continually feel feminine or see themselves as women can be considered to have a transfeminine gender identity. … This desire to see themselves as feminine also shifted their self-perception in that direction …”.
“Social behavior, interests, &c” is as far as I can tell gender conservative propaganda that Blanchardians made up in order to make it fit with the grand HBD narrative about gender differences. (People-things, Mahalanobis D, etc..) If you look at e.g. the correlation between Richard Lippa’s MF-Occ interests and gender identity among homosexual males, you get basically no correlation:
I think the aspect that actually correlates with gender issues is actually mainly down to appearance, both in terms of innate physique and acquired characteristics like an interest in makeup. (Possibly nonlinearly interacting with androphilia—needs more investigation.) Though I will have to finish up my analysis and get back to you on that.
The task force writing this section was lead by Ken Zucker, who is a personal friend of Blanchard and Bailey, right?
I think there’s a bunch of things to be said about categories which doesn’t really get said because people either short-circuit to “correlation-clusters!” or vaguely go ” 🤨 I don’t really think there is One True Essence of the trait you mention”.
I don’t understand depression or bipolar disorder well enough to comment on them, but consider the case of comas. First, part of how to deconfuse GCS is to notice that it makes a lot more sense if you invert it: it doesn’t measure comas, it measures consciousness. Comas are defined as “a prolonged state of deep unconsciousness, caused especially by severe injury or illness”, i.e. roughly the negation of consciousness, and so it makes sense that you would measure comas by a consciousness scale.
And from this perspective, the latent variable/factor model makes a lot of sense. Why does being “oriented to time, place and person” correlate with “obeys commands”? Because they both require you to be conscious. And you don’t have two different consciousnesses in your head, so representing it as a single variable makes sense. Yes, it’s true that mechanistically, consciousness draws on a whole shitton of different parts of your nervous system—but these different parts contribute to a unified computational system, and therefore have essentially identical effects on the symptoms of consciousness.
(Well, most parts probably have fairly limited effects on fairly narrow things. But what I mean is that the reason the GCS works is because there are a set of parts that are sufficiently necessary that they knock you out across a broad range of conscious functions.)
When I emphasize the value of causality in thinking about categories, it’s partly with cases like this where a reasonable yet loose understanding justifies the common-sense categories. But it is also very much in order to recognize the cases where categories don’t work this way—see for example my discussion with ChatGPT on trauma measurement here. By thinking carefully causally about why one creates a category, one can come up with better definitions and measurements.
(Bonus: results from trauma survey. Has strongly upped my estimate for how much of a thing traumas are. Though there’s issues of sampling; I don’t think I have much self-selection going on due to my filtering methods, but plausibly the women who find it worthwhile to answer surveys for money on the internet are going to be non-representative.)
Not if you buy into the masochistic CEO theory, as most people seem to:
I’ve been raging against parsimony for a while, so I thought I should mention a notion in favor of parsimony that I have been thinking of for a while:
We might represent gender issues as a utility difference between being a woman vs being a man. Except, this is a type error; generally utility functions are not meant to take in predicates, but instead meant to take in states/world-trajectories/???.
In some sense there’s nothing wrong with a utility function that gives 9001 utilons for being female if there are an even number of people alive, and 9001 utilons for being male if there are an odd number of people alive. Except this would be kind of impossible to act on and would flip all of the time.
So in practice, utility functions don’t look like this, but instead decompose into e.g. ridges of optimal states along dimensions of indifference, with gradual falloff as one distances oneself from the ridge.
In the case of autogynephilia, this looks like a ridge of being a woman, optimal against wildly varying external environments, with (possibly partly due to GAMP/AGAMP stuff) continuous falloff based on lesser degrees of feminization.
But for a different etiology, the shape of the utility function might look very different. Issues with gender norms might lead to a ridge that passes through being female as optimal in the current environment, but in an environment where the gender norms were flipped, the optimal sex would also be flipped. And an environment with weaker gender norms might be equally desirable to being the opposite sex in the current environment.
(This is somewhat complicated by caching.)
I find this phrasing/situation weird to parse. Taken literally, it sounds confusing as most cis women like having breasts and it’s far from unheard of for them to mention having breasts as one of the best parts of being female. So taken literally this feels like you are overupdating on a single unusual woman.
(Reminds me of the time where I posted quotes from women talking about what they liked about being female in a gender critical subreddit, and the GCs were like “this sounds so autogynephilic!”)
But I think maybe by the context, both you and her are using “want boobs” to mean something like “want to play with your own breasts in private for Personal Enjoyment”? It’s just a confusing way of phrasing it to me. (Do you have more logs/context from this conversation?)
I still get kind of bothered by how these discussions of transfeminine femininity are always measured with respect to cis women or HSTSs, rather than with respect to cis men. Like from a political perspective, it makes sense—if you are trying to fit into a category of “woman”, or trying to do some sort of good trans/bad trans thing, then these are the natural points to compare relative to.
But Blanchardians emphasize that their theory is about etiology—which would be the things that distinguish transfems from cis men—rather than politics. Clearly we can see that this is bullshit and they really care about politics, but the fact that they shift the discussion to etiology interferes with having a productive discussion about these things.
I’m not sure this is true. Microsoft has traditionally been known for having a dysfunctional, internally adversarial culture. But it no longer does, and from what I’ve been told, it’s directly related to the takeover by progressive “diversity/emotional intelligence” culture.
I mean of course it still has a CEO and lots of hierarchy and inequality, so in a sense it’s still not progressive (or perhaps rather one should say, still not socialist).
I should say that when I first learned about the Big Five Aspect Scales, I was enthusiastic about them and did a whole bunch of stuff related to them, but over time I came to the conclusion that I was just wading around in meaningless variance-shuffling.
One gets morally corrupted by right-wing science because right-wingers put in lots of subtle poor design decisions, vague implications, motivated stopping, incorrect epistemology, strange research questions, etc., in ways that are optimized for pushing right-wing views. No one aspect of it is unforgiveable, but the fact that it is a fractal of manipulation is what corrupts you infinitely.
This circles back to my point about causality and categories I mentioned earlier in my comment. And it also gets at some issues I have with some people’s characterization of the g factor. And with my issue with the sequences on words.
Within psychometrics, this debate is termed “reflective vs formative”. Basically, while the Sequences acknowledge that the underlying distribution matters for how to define words, they still treat category membership as caused by having the properties that we use to exemplify the category, not as causing it.
In IQ research, this becomes treating g as a composite of abilities measured by IQ tests. And in sex differences this becomes putting biological sex at the end of the causal network, rather than the beginning.
And like of course there are various senses in which this view is not totally false. g/biological sex is an abstraction of some more complicated underlying dynamics. But at least in the case of sex (and I suspect also for g) when you dig into what is abstracted over, then structurally in many ways it still satisfies the causal property of coming at the beginning of the causal network, rather than the end.
So basically your objection seems right but it gets at the core issue of how formativity is generally accepted as the way to create categories but actually reflectivity is usually a more natural view.
The debate between formativity and reflectivity usually seems kind of academic, partly because psychometricians are already wading around in highly abstracted data. I didn’t properly get its importance when I first got introduced to it, only later as I thought through it carefully in lots of different contexts. I still don’t have a great way to communicate the underlying insights because most ways of talking about it just sound too weird. I guess to some extent it’s also that people rely so much on their intuitive categorization that they don’t bother thinking through the rules for how categories work in practice.
There’s a thing in some GC spaces where they say that what autogynephilia is really about is getting off to transgressing boundaries. I usually interpret this to be a claim that autogynephilia is a member of the courtship disorders factor, a group of sexual interests that includes e.g. exhibitionism, voyeurism, and paraphilic rape. I think this prediction is outright false.
(But there are variants of autogynephilia where it probably holds—Kevin Hsu wrote that arousal to going to the women’s bathroom or locker rooms correlates with paraphilic interests such as voyeurism, exhibitionism and frotteurism. But many autogynephiles don’t report an attraction to this. Though I suppose some would argue that is just social desirability bias.)
I think an underrated issue is that it is not gender differences per se that are critical, but also how they add up.
Like take sociosexuality. Men are much more attracted to having sex with many different strangers than women are. But there is overlap in the distributions of sociosexual desire, yet my impression is that despite this overlap, men’s and women’s experience with sociosexuality is still totally different. Why?
Supply/demand and conflicts. If you are in the middle—an especially sociosexual woman or a not that sociosexual man—then you might want some casual sex, but the feasibility of this is gonna depend massively on your environment. The man is probably gonna do better by focusing on his comparative advantage, credibly promising intimacy to a long-term partner, whereas the woman can get as much sex as she wants, though possibly triggering all sorts of social dynamics and resentment as a result.
Blanchardians are IME incapable of reasoning about this sort of thing, because whenever the superficial outcomes look different, they jump to the conclusion that the underlying causes must be distinct, rather than that continuous variation can add up to give highly distinct results.
I feel like one can’t discuss this without bringing up gynandromorphophilia as a correlate of autogynephilia.