From d7b688ae10a904fbce0b2e4a69a1c4666f7fa617 Mon Sep 17 00:00:00 2001 From: "M. Taylor Saotome-Westlake" Date: Sat, 1 Oct 2022 17:33:21 -0700 Subject: [PATCH] memoir: trans kids on the margin This is going to need some editing and is definitely subject to prior constraint. But I think it's much stronger to have this part, rather than to say "not my story to tell" and move on without a hint. --- ...-hill-of-validity-in-defense-of-meaning.md | 34 ++++++++++++++++--- notes/a-hill-of-validity-sections.md | 2 ++ 2 files changed, 32 insertions(+), 4 deletions(-) diff --git a/content/drafts/a-hill-of-validity-in-defense-of-meaning.md b/content/drafts/a-hill-of-validity-in-defense-of-meaning.md index 7e3c778..e26d782 100644 --- a/content/drafts/a-hill-of-validity-in-defense-of-meaning.md +++ b/content/drafts/a-hill-of-validity-in-defense-of-meaning.md @@ -637,21 +637,47 @@ motivation deflates after Christmas victory ] +------- + There's another extremely important part of the story that _would_ fit around here chronologically, but I again find myself constrained by privacy norms: everyone's common sense of decency (this time, even including my own) screams that it's not my story to tell. -Adherence to norms is fundamentally fraught for the same reason as AI alignment is. That is, in [rich domains](https://arbital.com/p/rich_domain/), attempts to regulate behavior with explicit constraints face a lot of adversarial pressure from optimizers bumping up against the constraint, and finding the [nearest unblocked strategies](https://arbital.greaterwrong.com/p/nearest_unblocked) that circumvent the constraint. The intent of privacy norms restricting what things you're allowed to say, is to conceal information. But _information_ in Shannon's sense is about what states of the world can be inferred given the states of communication signals; it's much more expansive than the denotative meaning of a text, what we would colloquially think of as the explicit "content" of a message. +Adherence to norms is fundamentally fraught for the same reason AI alignment is. That is, in [rich domains](https://arbital.com/p/rich_domain/), attempts to regulate behavior with explicit constraints face a lot of adversarial pressure from optimizers bumping up against the constraint and finding the [nearest unblocked strategies](https://arbital.greaterwrong.com/p/nearest_unblocked) that circumvent the constraint. The intent of privacy norms restricting what things you're allowed to say, is to conceal information. But _information_ in Shannon's sense is about what states of the world can be inferred given the states of communication signals; it's much more expansive than the denotative meaning of a text, what we would colloquially think of as the explicit "content" of a message. -If norms can only regulate the denotative meaning of a text (because trying to regulate subtext is too subjective for a norm-enforcing coalition to coordinate on), someone who would prefer to reveal private information, but also wants to comply with privacy norms, has an incentive to leak everything they possibly can as subtext—to imply it, and hope to escape punishment on grounds of not having "really said it." And if there's some sufficiently egregious letter-complying-but-spirit-violating evasion of the norm, that a coalition _can_ coordinate on enforcing, the info-revealer has an incentive to stay _just_ shy of being that egregious. +If norms can only regulate the denotative meaning of a text (because trying to regulate subtext is too subjective for a norm-enforcing coalition to coordinate on), someone who would prefer to reveal private information, but also wants to comply with privacy norms, has an incentive to leak everything they possibly can as subtext—to imply it, and hope to escape punishment on grounds of not having "really said it." And if there's some sufficiently egregious letter-complying-but-spirit-violating evasion of the norm, that a coalition _can_ coordinate on enforcing, the whistleblower has an incentive to stay _just_ shy of being that egregious. -Thus, it's unclear how much mere adherence to norms helps, when people's wills are actually misaligned. If I'm furious at Yudkowsky for prevaricating about my Something to Protect, and am in fact _more_ furious rather than less that he mostly managed to do it without technically "lying", I should not be so debased as to think myself innocent for not having "really said it." +Thus, it's unclear how much mere adherence to norms helps, when people's wills are actually misaligned. If I'm furious at Yudkowsky for prevaricating about my Something to Protect, and am in fact _more_ furious rather than less that he managed to do it without violating the norm against "lying", I should not be so foolish as to think myself innocent and beyond reproach for not having "really said it." Having considered all this, here's what I think I can say: I spent many hours in the first half of 2020 working on a private Document about a disturbing hypothesis that had occured to me. Previously, I had _already_ thought it was nuts that trans ideology was exerting influence the rearing of gender-non-conforming children, that is, children who are far outside the typical norm of _behavior_ (_e.g._, social play styles) for their sex: very tomboyish girls and very feminine boys. Under recent historical conditions in the West, these kids were mostly "pre-gay" rather than trans. (The stereotype about lesbians being masculine and gay men being feminine is, like most stereotypes, basically true: sex-atypical childhood behavior between gay and straight adults [has been meta-analyzed at _d_ ≈ 1.31 for men and _d_ ≈ 0.96 for women](/papers/bailey-zucker-childhood_sex-typed_behavior_and_sexual_orientation.pdf).) A solid supermajority of children diagnosed with gender dysphoria [ended up growing out of it by puberty](/papers/steensma_et_al-factors_associated_with_desistence_and_persistence.pdf). In the culture of the current year, it seemed likely that a lot of those kids would get affirmed into a cross-sex identity (and being a lifelong medical patient) much earlier, even though most of them would have otherwise (under a "watchful waiting" protocol) grown up to be ordinary gay men and lesbians. -What made this crazy, in my view, was not just that it was a dubious treatment decision, but that it was a dubious treatment decision made on the basis of the obvious falsehood that "trans" was one thing: the cultural phenomenon of "trans kids" was being used to legitimize trans _adults_, even though the vast supermajority of trans adults were in the AGP taxon and therefore _had never resembled_ these HSTS-taxon kids. That is: pre-gay kids are being sterilized in order to affirm the narcissistic delusions of _guys like me_. +What made this crazy, in my view, was not just that child transition is a dubious treatment decision, but that it's a dubious treatment decision made on the basis of the obvious falsehood that "trans" was one thing: the cultural phenomenon of "trans kids" was being used to legitimize trans _adults_, even though the vast supermajority of trans adults were in the AGP taxon and therefore _had never resembled_ these HSTS-taxon kids. That is: pre-gay kids are being sterilized in order to affirm the narcissistic delusions of _guys like me_. + +That much was obvious to anyone who's had their Blanchardian enlightenment, and wouldn't have been worth the effort of writing a special private Document about. The disturbing hypothesis that occured to me in early 2020 was that, in the culture of the current year, affirmation of a cross-sex identity might happen to kids _who weren't even HSTS-taxon at all_. + +Very small children who are just learning what words mean say a lot of things that aren't true (I'm a grown-up; I'm a cat; I'm a dragon), and grownups tend to play along _in the moment_ as a fantasy game, but they don't _coordinate to make that the permanent new social reality_. Ten years ago, if an otherwise ordinary three-year-old boy had occasionally claimed to be a girl, I think his nice smart liberal unschooling grownups would treat it about the same way as when the kid claims to be a cat. (I'm going with the MtF case with only a little loss of generality; I don't think the egregore is quite as eager to "trans" females at this age.) + +But if the grown-ups have been trained to believe that "trans kids know who they are"—if they're emotionally eager at the prospect of having a transgender child, or fearful of the damage they might do by not affirming—they might selectively attend to confirming evidence that the child "is trans", selectively ignore evidence that the kid "is cis", and end up reinforcing a cross-sex identity that _would not have existed_ if not for their belief in it. + +Crucially, if innate gender identity _isn't_ a feature of toddler psychology, _the child has no way to know anything is "wrong."_ If none of the grown-ups can say, "You're a boy because boys are the ones with penises" (because that's not what people are supposed to believe in the current year), how is the child supposed to figure that out independently? [_Toddlers_ are not very sexually dimorphic](/2019/Jan/the-dialectic/), but sex differences in play styles tend to emerge within a few years. (Did you know the [sex difference in preference for toy cars is _d_ ≈ 2.44?!](/papers/davis-hines-how_large_are_gender_differences_in_toy_preferences.pdf)) What happens when the kid develops a self-identity as "a girl", only to find out, potentially years later, that she noticeably doesn't fit in with the (cis) girls on the [many occasions that no one has explicitly spelled out in advance](/2019/Dec/more-schelling/) where people are using "gender" (percieved sex) to make a prediction or decision? + +Some might protest, "But what's the harm? She can always change her mind later if she decides she's actually a boy." I don't doubt that if the child were to clearly and distinctly insist, "I'm definitely a boy," the nice smart liberal grown-ups would unhesitatingly accept that. + +But the harm I'm theorizing is _not_ that the child has an intrinsic male identity that hurts to not be respected. (What _is_ an "identity", apart from the ordinary factual belief that one is of a particular sex?) Rather, the concern is that social transition prompts everyone, _including the child themself_, to use their mental models of girls (juvenile female humans) to make (mostly subconscious rather than deliberative) predictions and decisions about the child, which will be a systematically worse statistical fit than their models of boys (juvenile male humans), because the child is, in fact, a boy (juvenile male human), and those miscalibrated predictions and decisions will make the child's life worse in a complicated, illegible way that doesn't necessarily result in the child spontaneously verbally asserting, "I prefer that you call me a boy." +Scott Alexander has written about how [concept-shaped holes can be impossible to notice](https://slatestarcodex.com/2017/11/07/concept-shaped-holes-can-be-impossible-to-notice/). A culture whose [civic religion](https://slatestarcodex.com/2019/07/08/gay-rites-are-civil-rites/) celebrates being trans, and denies that gender has truth conditions other than the individual's say-so, has concept-shaped holes that make it hard to notice the hypothesis "I'm having a systematically worse childhood than I otherwise would have because all the grown-ups in my life have agreed I was a girl since I was three years old, even though all of my actual traits are sampled from the joint distribution of juvenile male humans, not juvenile female humans", even if it's true. + +... anyway, that's just a hypothesis that occured to me in early 2020, about something that _could_ happen in the culture of the current year, hypothetically, as far as I know. I'm not a parent and not a professional scientist. (And even if the "Clever Hans" etiological pathway I conjectured is real, the extent to which it might apply to any particular case is complex; you could imagine a kid who was "actually trans", whose social transition merely happened earlier than it otherwise would have due to these dynamics.) + +For some reason, it seemed really important that I draft a Document about it with lots of citations to send to a few friends. If I get around to it, I might clean it up and publish it as a blog post (working title: "Trans Kids on the Margin; and, Harms from Misleading Training Data"). + +Given that I spent so many hours on this little research/writing project in early 2020, I think it makes sense for me to mention at this point in the memoir, where it fits in chronologically. I have an inalienable right to talk about my own research interests, and talking about my research interests obviously doesn't violate any norm against leaking private information about someone else's family, or criticizing someone's parenting decisions. + +(Only—you two have such beautiful children!) + +----- +[TODO: "Human Diversity" review; all swans are white!] [TODO: pandemic starts] diff --git a/notes/a-hill-of-validity-sections.md b/notes/a-hill-of-validity-sections.md index 35d0c91..435789f 100644 --- a/notes/a-hill-of-validity-sections.md +++ b/notes/a-hill-of-validity-sections.md @@ -13,6 +13,7 @@ _ excerpt 2nd "out of patience" email with internet available— _ "watchful waiting" +_ Atlantic article on "My Son Wears Dresses" https://archive.is/FJNII _ in "especially galling" §: from "Changing Emotions"—"somehow it's always about sex when men are involved"—he even correctly pinpointing AGP in ordinary men (as was obvious back then), just without the part that AGP _is_ "trans" _ "look at what ended up happening"—look up whether that exact quote from from http://www.hpmor.com/chapter/47 or https://www.hpmor.com/chapter/97 _ Discord history with Scott (leading up to 2019 Christmas party, and deferring to Tailcalled on SSC survey question wording) @@ -57,6 +58,7 @@ _ explain why I'm not being charitable in 2018 thread analysis, that at the time _ January 2019 meeting with Ziz and Gwen _ better summary of Littman _ explain Rob +_ edit the child transition section in a way that Kay Brown would be OK with, have a few sentences about Clever Hans before the wrap-up terms to explain on first mention— -- 2.17.1