1 Title: A Hill of Validity in Defense of Meaning
4 Tags: autogynephilia, bullet-biting, cathartic, Eliezer Yudkowsky, Scott Alexander, epistemic horror, my robot cult, personal, sex differences, two-type taxonomy
7 > If you are silent about your pain, they'll kill you and say you enjoyed it.
11 Recapping our story so far—in a previous post, ["Sexual Dimorphism in Yudkowsky's Sequences, in Relation to My Gender Problems"](/2021/May/sexual-dimorphism-in-the-sequences-in-relation-to-my-gender-problems/), I told the the part about how I've "always" (since puberty) had this obsessive sexual fantasy about being magically transformed into a woman and also thought it was immoral to believe in psychological sex differences, until I got set straight by these really great Sequences of blog posts by Eliezer Yudkowsky, which taught me (incidentally, among many other things) how absurdly unrealistic my obsessive sexual fantasy was given merely human-level technology, and that it's actually immoral _not_ to believe in psychological sex differences given that psychological sex differences are actually real. In a subsequent post, ["Blanchard's Dangerous Idea and the Plight of the Lucid Crossdreamer"](/2022/TODO/blanchards-dangerous-idea-and-the-plight-of-the-lucid-crossdreamer/), I told the part about how, in 2016, everyone in my systematically-correct-reasoning community up to and including Eliezer Yudkowsky suddenly starting claiming that guys like me might actually be women in some unspecified metaphysical sense, and insisted on playing dumb when confronted with alternative explanations of the relevant phenomena or even just asked what that means, until I eventually had a stress- and sleep-deprivation-induced delusional nervous breakdown, got sent to psychiatric prison once, and then went crazy again a couple months later.
13 That's not the really egregious part of the story. The thing is, psychology is a complicated empirical science: no matter how "obvious" I might think something is, I have to admit that I could be wrong—[not just as an obligatory profession of humility, but _actually_ wrong in the real world](https://www.lesswrong.com/posts/GrDqnMjhqoxiqpQPw/the-proper-use-of-humility). If my fellow rationalists merely weren't sold on the autogynephilia and transgender thing, I would certainly be disappointed, but it's definitely not grounds to denounce the entire community as a failure or a fraud. And indeed, I _did_ [end up moderating my views](/2022/Jul/the-two-type-taxonomy-is-a-useful-approximation-for-a-more-detailed-causal-model/) compared to the extent to which my thinking in 2016–7 took Blanchard–Bailey–Lawrence as received truth. At the same time, I don't particularly regret saying what I said in 2016–7, because Blanchard–Bailey–Lawrence is still very obviously _directionally_ correct compared to the nonsense everyone else was telling me.
15 But a striking pattern in my attempts to argue with people about the two-type taxonomy in late 2016 and early 2017 was the tendency for the conversation to get _derailed_ on some variation of, "Well, the word _woman_ doesn't necessarily mean that," often with a link to ["The Categories Were Made for Man, Not Man for the Categories"](https://slatestarcodex.com/2014/11/21/the-categories-were-made-for-man-not-man-for-the-categories/), a 2014 post by Scott Alexander arguing that because categories exist in our model of the world rather than the world itself, there's nothing wrong with simply _defining_ trans people to be their preferred gender, in order to alleviate their dysphoria.
17 After Yudkowsky had stepped away from full-time writing, Alexander had emerged as our subculture's preeminent writer. Most people in an intellectual scene "are writers" in some sense, but Alexander was the one "everyone" reads: you could reference a _Slate Star Codex_ post in conversation and expect people to familiar with the idea, either from having read it, or by osmosis via other people having referenced the idea in conversation. The frequency with which "... Not Man for the Categories" was cited at me, seemed to suggest it had become our subculture's "party line" on trans issues.
19 But the post is very wrong in very obvious ways. To be clear, it's true _that_ categories exist in our model of the world, rather than the world itself—categories are "map", not "territory"—and it's true that trans women might be women _with respect to_ some genuinely useful definition of the word "woman." However, Alexander goes much further, claiming that we can redefine gender categories _in order to make trans people feel better_:
21 > I ought to accept an unexpected man or two deep inside the conceptual boundaries of what would normally be considered female if it'll save someone's life. There's no rule of rationality saying that I shouldn't, and there are plenty of rules of human decency saying that I should.
23 This is wrong because categories exist in our model of the world _in order to_ capture empirical regularities in the world itself: the map is supposed to _reflect_ the territory, and there _are_ "rules of rationality" governing what kinds of word and category usages correspond to correct probabilistic inferences. [Yudkowsky wrote a whole Sequence about this](https://www.lesswrong.com/s/SGB7Y5WERh4skwtnb) back in 'aught-eight, as part of the original Sequences. Alexander cites [a post](https://www.lesswrong.com/posts/yA4gF5KrboK2m2Xu7/how-an-algorithm-feels-from-inside) from that Sequence in support of the (true) point about how categories are "in the map" ... but if you actually read the Sequence, another point that Yudkowsky pounds home _over and over and over again_, is that word and category definitions are nevertheless _not_ arbitrary: you can't define a word any way you want, because there are [at least 37 ways that words can be wrong](https://www.lesswrong.com/posts/FaJaCgqBKphrDzDSj/37-ways-that-words-can-be-wrong)—principles that make some definitions _perform better_ than others as "cognitive technology."
25 In the case of Alexander's bogus argument about gender categories, the relevant principle ([#30](https://www.lesswrong.com/posts/d5NyJ2Lf6N22AD9PB/where-to-draw-the-boundary) on [the list of 37](https://www.lesswrong.com/posts/FaJaCgqBKphrDzDSj/37-ways-that-words-can-be-wrong)) is that if group things together in your map that aren't actually similar in the territory, you're going to be misled into making bad predictions.
27 Importantly, this is a very general point about how language itself works _that has nothing to do with gender_. No matter what you believe about politically-controversial empirical questions, intellectually honest people should be able to agree that "I ought to accept an unexpected [X] or two deep inside the conceptual boundaries of what would normally be considered [Y] if [positive consequence]" is not the correct philosophy of language, _independently of the particular values of X and Y_.
29 Also, this ... really wasn't what I was trying to talk about. _I_ thought I was trying to talk about autogynephilia as an _empirical_ theory of psychology, the truth or falsity of which obviously cannot be altered by changing the meanings of words. But at this point I still trusted people in my robot cult to be basically intellectually honest, rather than fucking with me because of their political incentives, so I endeavored to respond to the category-boundary argument as if it were a serious argument: when I quit my dayjob in March 2017 in order to have more time to study and work on this blog, the capstone of my sabbatical was an exhaustive response to Alexander, ["The Categories Were Made for Man to Make Predictions"](/2018/Feb/the-categories-were-made-for-man-to-make-predictions/) (which Alexander [graciously included in his next links post](https://archive.ph/irpfd#selection-1625.53-1629.55)). A few months later, I followed it up with ["Reply to _The Unit of Caring_ on Adult Human Females"](/2018/Apr/reply-to-the-unit-of-caring-on-adult-human-females/), responding to a similar argument. I'm proud of those posts: I think Alexander's and _Unit of Caring_'s arguments were incredibly dumb, and with a lot of effort, I think I did a pretty good job of explaining exactly why to anyone who was interested and didn't, at some level, prefer not to understand.
31 Of course, a pretty good job of explaining by one niche blogger wasn't going to put much of a dent in the culture, which is the sum of everyone else's blogposts; despite the mild boost from the _Slate Star Codex_ links post, my megaphone just wasn't very big. At this point, I was _disappointed_ with the limited impact of my work, but not to the point of bearing much hostility to "the community". People had made their arguments, and I had made mine; I didn't think I was _entitled_ to anything more than that.
33 ... and, really, that _should_ have been the end of the story. Not much of a story at all. If I hadn't been further provoked, I would have still kept up this blog, and I still would have ended up arguing about gender with people occasionally, but this personal obsession of mine wouldn't have been the occasion of a full-on robot-cult religious civil war.
35 The _causis belli_ for the religious civil war happened on 28 November 2018. I was at my new dayjob's company offsite event in Austin. Coincidentally, I had already spent much of the afternoon arguing trans issues with other "rationalists" on Discord. [TODO: review Discord logs; email to Dad suggests that offsite began on the 26th, contrasted to first shots on the 28th]
37 Just that month, I had started a Twitter account in my own name, inspired in an odd way by the suffocating [wokeness of the open-source software scene](/2018/Oct/sticker-prices/) where I [occasionally contributed diagnostics patches to the compiler](https://github.com/rust-lang/rust/commits?author=zackmdavis). My secret plan/fantasy was to get more famous/established in the that world (one of compiler team membership, or conference talk accepted, preferably both), get some corresponding Twitter followers, and _then_ bust out the Blanchard retweets and links to this blog. In the median case, absolutely nothing would happen (probably because I failed at being famous), but I saw an interesting tail of scenarios in which I'd get to be a test case in the Code of Conduct wars.
39 [TODO: SECTION SUMMARIZING AND RESPONDING TO "Hill of Validity" Tweet thread]
41 I was physically shaking. I remember going downstairs to confide in a senior engineer about the situation. But if Yudkowsky was _already_ stonewalling his Twitter followers, entering the thread myself didn't seem likely to help. (And I hadn't intended to talk about gender on that account yet, although that seemed less important given the present crisis.)
43 It seemed better to try to clear this up in private. I still had Yudkowsky's email address. I felt bad bidding for his attention over my gender thing _again_—but I had to do _something_. Hands trembling, I sent him an email asking him to read my ["The Categories Were Made for Man to Make Predictions"](/2018/Feb/the-categories-were-made-for-man-to-make-predictions/), suggesting that it may qualify as an answer to his question about ["a page [he] could read to find a non-confused exclamation of how there's scientific truth at stake"](https://twitter.com/ESYudkowsky/status/1067482047126495232), and that, because I cared very much about correcting what I claim are confusions in my rationalist subculture, that I would be happy to pay up to $1000 for his time, and that, if he liked the post, he might consider Tweeting a link.
45 [TODO (Subject: "another offer, $1000 to read a ~6500 word blog post about (was: Re: Happy Price offer for a 2 hour conversation)")]
47 The monetary offer, admittedly, was awkward: I included another paragraph clarifying that any payment was only to get his attention, and not _quid quo pro_ advertising, and that if he didn't trust his brain circuitry not to be corrupted by money, then he might want to reject the offer on those grounds and only read the post if he expected it to be genuinely interesting.
49 Again, I realize this must seem weird and cultish to any normal people reading this. (Paying some blogger you follow one grand just to _read_ on of your posts? Who _does_ that?) To this, I again refer to [the reasons justifying my 2016 cheerful price offer](/2022/TODO/blanchards-dangerous-idea-and-the-plight-of-the-lucid-crossdreamer/#cheerful-price-reasons)—and that it was a way to signal that I _really really didn't want to be ignored_.
51 [TODO: and cc'd Michael and Anna as character references]
53 [TODO: I did successfully interest Michael, who chimed in—Michael called me up and we talked about how the "rationalists" were over]
55 [TODO: as with earlier cheerful price offer, I can't confirm or deny whether Yudkowsky replied or whether he accepted the cheerful price offer]
57 [TODO: "not ontologically confused" concession. You might think that should be the end of the matter—but this little "not ontologically confused" at the bottom of the thread was much less visible and loud than the bold, arrogant top-level pronouncement insinuating that GCs are philosophically confused. Was I greedy to want something louder? https://twitter.com/ESYudkowsky/status/1068071036732694529 ]
59 Greedy or not, I wasn't satisfied. On 1 December, I wrote to Scott Alexander, asking if there was any chance of an _explicit_ and _loud_ clarification or partial-retraction of ["... Not Man for the Categories"](https://slatestarcodex.com/2014/11/21/the-categories-were-made-for-man-not-man-for-the-categories/) (Subject: "super-presumptuous mail about categorization and the influence graph"). _Forget_ my boring whining about the autogynephilia/two-types thing, I said—that's a complicated empirical claim, and _not_ the key issue.
61 The _issue_ is that category boundaries are not arbitrary (if you care about intelligence being useful): you want to [draw your category boundaries such that](https://www.lesswrong.com/posts/d5NyJ2Lf6N22AD9PB/where-to-draw-the-boundary) things in the same category are similar in the respects that you care about predicting/controlling, and you want to spend your [information-theoretically limited budget](https://www.lesswrong.com/posts/soQX8yXLbKy7cFvy8/entropy-and-short-codes) of short words on the simplest and most wide-rangingly useful categories.
63 It's true that the reason _I_ was continuing to freak out about this to the extent of sending you him this obnoxious email telling him what to write (seriously, what kind of asshole does that?!) has to with transgender stuff, but that's not the reason _Scott_ should care. Rather, it's like [his parable about whether thunder or lightning comes first](http://slatestarcodex.com/2017/10/23/kolmogorov-complicity-and-the-parable-of-lightning/): there aren't many direct non-rationalizable-around consequences of the sacred dogmas that thunder comes before lightning or that biological sex somehow isn't real; the problem is that the need to _defend_ the sacred dogma _destroys everyone's ability to think_. If our vaunted rationality techniques result in me having to spend dozens of hours patiently explaining why I don't think that I'm a woman and that [the person in this photograph](https://daniellemuscato.startlogic.com/uploads/3/4/9/3/34938114/2249042_orig.jpg) isn't a woman, either (where "isn't a woman" is a convenient rhetorical shorthand for a much longer statement about [naïve Bayes models](https://www.lesswrong.com/posts/gDWvLicHhcMfGmwaK/conditional-independence-and-naive-bayes) and [high-dimensional configuration spaces](https://www.lesswrong.com/posts/WBw8dDkAWohFjWQSk/the-cluster-structure-of-thingspace) and [defensible Schelling points for social norms](https://www.lesswrong.com/posts/Kbm6QnJv9dgWsPHQP/schelling-fences-on-slippery-slopes)), then our techniques are _worse than useless_. If Galileo ever muttered "And yet it moves", there's a long and nuanced conversation you could have about the consequences of using the word "moves" in Galileo's preferred sense or some other sense that happens to result in the theory needing more epicycles. It may not have been obvious in 2014, but in retrospect, _maybe_ it was a _bad_ idea to build a [memetic superweapon](https://archive.is/VEeqX) that says the number of epicycles _doesn't matter_.
65 And the reason to write this is a desperate email plea to Scott Alexander when I could be working on my own blog, was that I was afraid that marketing is a more powerful force than argument. Rather than good arguments propagating through the population of so-called "rationalists" no matter where they arise, what actually happens is that people like Eliezer and you rise to power on the strength of good arguments and entertaining writing (but mostly the latter), and then everyone else sort-of absorbs most of their worldview (plus noise and [conformity with the local environment](https://thezvi.wordpress.com/2017/08/12/what-is-rationalist-berkleys-community-culture/)). So for people who didn't [win the talent lottery](http://slatestarcodex.com/2015/01/31/the-parable-of-the-talents/) but think they see a flaw in the Zeitgeist, the winning move is "persuade Scott Alexander".
67 So, what do you say, Scott? Back in 2010, the rationalist community had a shared understanding that the function of language is to describe reality. Now, we don't. If you don't want to cite my creepy blog about my creepy fetish, that's fine. I like getting credit, but the important thing is that this "No, the Emperor isn't naked—oh, well, we're not claiming that he's wearing any garments—it would be pretty weird if we were claiming that!—it's just that utilitarianism implies that the social property of clothedness should be defined this way because to do otherwise would be really mean to people who don't have anything to wear" gaslighting maneuver needs to _die_. You alone can kill it.
69 ... Scott didn't get it.
71 [TODO: paraphrase remaining interaction with Scott, or not worth the space?
73 > I don't have a simple, mistake-theoretic characterization of the language and social conventions that everyone should use such that anyone who defected from the compromise would be wrong. The best I can do is try to objectively predict the consequences of different possible conventions—and of conflicts over possible conventions.
75 helping Norton live in the real world
77 If any concrete negative consequence of taking gender self-identity categories (e.g., more sexual assaults) is going to be waved away with "Oh, that's a mere policy decision that can be dealt with on some basis other than gender, and therefore doesn't count as an objection to the new definition of gender categories", then I guess I can't argue with that. But I wish I knew what happened to my wallet.
79 Scott says, "It seems to me pretty obvious that the mental health benefits to trans people are enough to tip the object-level first-pass uilitarian calculus."; I don't think _anything_ about "mental health benefits to trans people" is obvious.
82 [TODO: connecting with Aurora 8 December, maybe not important]
84 Anna told me that my "You have to pass my litmus test or I lose all respect for you as a rationalist" attitude was psychologically coercive. I agreed—I was even willing to go up to "violent"—in the sense that it's [trying to apply social incentives towards an outcome rather than merely exchanging information](http://zackmdavis.net/blog/2017/03/an-intuition-on-the-bayes-structural-justification-for-free-speech-norms/). But sometimes you need to use violence in defense of self or property, even if violence is generally bad. If we think of the "rationalist" label as intellectual property, maybe it's property worth defending, and if so, then "I can define a word any way I want" isn't obviously a terrible time to start shooting at the bandits? What makes my "... or I lose all respect for you as a rationalist" moves worthy of your mild reproach, but "You're not allowed to call this obviously biologically-female person a woman, or I lose all respect for you as not-an-asshole" merely a puzzling sociological phenomenon that might be adaptive in some not-yet-understood way? Isn't the violence-structure basically the same? Is there any room in civilization for self-defense?
86 When I told Michael about this, he said that I was ethically or 'legally' in the right here, and the rationalist equivalent of a lawyer matters more for your claims than the equivalent of a scientist, and that Ben Hoffman (who I had already shared the thread with Scott with) would be helpful in solidifying my claims to IP defense. I said that I didn't _feel_ like I'm in the right, even if I can't point to a superior counterargument that I want to yield to, just because I'm getting fatigued from all the social-aggression I've been doing. (If someone tries to take your property and you shoot at them, you could be said to be the "aggressor" in the sense that you fired the first shot, even if you hope that the courts will uphold your property claim later.)
88 [TODO: re Ben's involvement—I shared Scott thread with Ben and Katie; Michael said "could you share this with Ben? I think he is ready to try & help." on 17 December
90 > talk more with Ben Hoffman, he will be more helpful in solidifying your claims about IP defense etc. ]
91 [TOOD: Sarah's involvement: I cc'd her on the 1 December Scott thread, and Sarah chimed in support in that thread, and 16 December draft to Michael]
94 > If we have this entire posse, I feel bad/guilty/ashamed about focusing too much on my special interest except insofar as it's actually a proxy for "has Eliezer and/or everyone else [lost the plot](https://thezvi.wordpress.com/2017/08/12/what-is-rationalist-berkleys-community-culture/), and if so, how do we get it back?"
97 > Your special interest is not just a coincidence here.
99 Sarah shying away, my rallying cry—
101 > There have been times during the last few weeks where it's felt like my mind has shut down with the only thought, "What the hell am I _doing_? This is _absurd_. Why am I running around picking fights about the philosophy of language—and worse, with me arguing for the _Bad_ Guys' position? I don't want to be one of the Bad Guys. Maybe I'm wrong and should stop making a fool out of myself. After all, using Aumann-like reasoning, in a dispute of 'Zack M. Davis and Michael Vassar vs. _everyone fucking else_', wouldn't I want to bet on 'everyone else'? Obviously."
103 > Except that I was raised to believe that you're supposed to concede arguments on the basis of encountering a superior counterargument that makes you change your mind. And I can't actually point to such a counterargument. "Maybe I'm making a fool out of myself by picking fights with all these high-status people" is not a counterargument. When I look at the world, it looks like [Scott](http://slatestarcodex.com/2014/11/21/the-categories-were-made-for-man-not-man-for-the-categories/) and [Eliezer](https://twitter.com/ESYudkowsky/status/1067183500216811521) and [Kelsey](https://theunitofcaring.tumblr.com/post/171986501376/your-post-on-definition-of-gender-and-woman-and) and [Robby Bensinger](https://www.facebook.com/robbensinger/posts/10158073223040447?comment_id=10158073685825447&reply_comment_id=10158074093570447&comment_tracking=%7B%22tn%22%3A%22R2%22%7D) seem to think that some variation on ["I can define a word any way I want"]() is sufficient to end debates on transgender identity.
105 > And ... I want to be nice to my trans friends, too, but that can't possibly be the definitive end-of-conversation correct argument. Not _really_. Not if you're being serious.
107 > But "whatever" is _not a counterargument!_
109 > But I want to live in a rationalist community that can actually _update_—where a bad argument by a high-status person will _eventually_ be publicly known to lose to a superior counterargument, even if the person originating the counterargument is lower status and the counterargument happens to have genuinely unfavorable political implications.
111 > But ultimately, I've got _something to protect_. This is even more humiliating, really. It's not even a good or wholesome something to protect that you'd feel comfortable blogging about under your own name. It's actually kind of obscene. Good or not, wholesome or not, obscene or not, it's for all the boys like me who had a dream whose proper realization happens to have [_more complicated implementation details_](https://www.lesswrong.com/posts/QZs4vkC7cbyjL9XA9/changing-emotions) than applying social pressure to make everyone pretend that they've coordinated to redefine a top-20 word. Those boys deserve to not be lied to.
113 > I've tried to be pretty explicit about only expecting buy-in for a minimal flag that says, "'I Can Define a Word Any Way I Want' can't be the end of the debate, because choosing to call things different names doesn't change the empirical cluster-structure of bodies and minds in the world; while the same word might legitimately be used with different definitions/extensions in different contexts
115 > I do want to distinguish between direct ideological conformity enforcement attempts, and "people not living up to their usual epistemic standards in response to ideological conformity enforcement in the general culture." So-called "rationalists" are actually very good at not-being-persecutory! (I don't think I'd be saying this in the nearby possible world where Scott Siskind didn't have a traumatizing social-justice-shaming experience in college, but it's true here.)
117 > I want to emphasize this, because I don't want to fall into the [bravery-debate](http://slatestarcodex.com/2013/05/18/against-bravery-debates/) trap of, "Look at me, I'm so heroically persecuted, therefore I'm right (therefore you should have sex with me)."
119 > My hope has been that it's possible to apply just enough "What the fuck kind of rationalist are you?!" social pressure to cancel out the "You don't want to be a Bad (Red) person, do you??" social pressure and thereby let people look at the arguments. I don't know if that actually works.
121 > I keep thinking of the part where Dagny entreats the great Dr. Robert Stadler to denounce a not-technically-lying statement by the State Science Institute
123 > Imagine your friends got a new front porch installed on your house, and you noticed that the boards were smoother and shinier than you expected. A question you might then ask your friend is, "Oh, is this real wood, or one of those wood-plastic composites?"
125 > In phrasing it that way, you're not saying that composites are bad; it's just that it makes sense to use language to asymmetrically distinguish between the natural thing that already existed, and the synthetic thing that has been deliberately engineered to resemble the original thing as much as possible.
127 [TODO: 4 January plea to Yudkowsky again—this is the part that I should base my thread analysis off of]
129 [TODO: ... email analysis goes up to 4 January]
131 [TODO: proton concession]
133 [TODO: minor psych episode in March 2019]
135 [TODO: "Blegg Mode", "Where to Draw the Boundaries?", and failure /2019/May/hiatus/ ]
137 [TODO: more blogging (https://www.lesswrong.com/posts/5aqumaym7Jd2qhDcy/containment-thread-on-the-motivation-and-political-context), 2019 Christmas party, disclaimer on "Categories Were Made"]
138 ["Univariate fallacy" also a concession]
140 [TODO: categories clarification from EY—victory?!]
142 [TODO: I didn't put this together until looking at my email just now, but based on the timing, the Feb. 2021 pronouns post was likely causally downstream of me being temporarily more salient to EY because of my highly-Liked response to his "anyone at this point that anybody who openly hates on this community generally or me personally is probably also a bad person inside" from 17 February; it wasn't gratuitously out of the blue]
144 [TODO: "simplest and best" pronoun proposal, sometimes personally prudent; support from Oli]
146 [TODO: the dolphin war, our thoughts about dolphins are literally downstream from Scott's political incentives in 2014; this is a sign that we're a cult]
148 [TODO: why you should care; no one should like Scott and Eliezer's proposals; knowledge should go forward, not back — what I would have hoped for, what you can do; hating that my religion is bottlenecked on one guy; the Church is _still there_ sucking up adherents; this is unambiguously a betrayal rather than a mistake]
149 [TODO: the Death With Dignity era]
projects / Ultimately_Untrue_Thought.git / blob