-The reason it makes any sense at all for you to be skeptical, is because our beliefs about the existence and etiology of "bipolar disorder", don't completely stand or fall on this particular test. People _already_ had many observations pointing to the idea of "bipolar disorder" as a common cluster of symptoms. From your years of clinical experience, you think you know with your eyes what the cluster looks like. So when people whose favorite team lost the Super Bowl happen to answer "Yes" to the some of the same survey questions as people who you've _seen_ in the frenzy of mania and depths of depression, you generate the hypothesis: "Gee, maybe different populations are interpreting the question differently." Not as a _certainty_—maybe further research will provide more solid evidence that "bipolar disorder" isn't what you thought—but there's nothing un-Bayesian about thinking that your brain's pattern-matching capabilities are on to something important that this particular survey instrument isn't catching. You're not scientifically obligated to _immediately_ jump to "Bipolar Is Common and Not Especially Related to Mania or Depression."
+The reason it makes sense for Alexander to be skeptical of the screening test is because our beliefs about the existence and etiology of "bipolar disorder" don't completely stand or fall on this particular test. People already had many observations pointing to the idea of "bipolar disorder" [as a taxon](https://lorienpsych.com/2020/10/30/ontology-of-psychiatric-conditions-taxometrics/). As an experienced clinician, when people whose favorite team lost the Super Bowl happen to answer "Yes" to the some of the same survey questions as people who you've _seen_ in the frenzy of mania and depths of depression, you generate the hypothesis: "Gee, maybe different populations are interpreting the question differently." Not as a certainty—maybe further research will provide more solid evidence that "bipolar disorder" isn't what you thought—but there's nothing un-Bayesian about thinking that your brain's pattern-matching capabilities are on to something important that this particular survey instrument isn't catching. You're not scientifically obligated to immediately jump to "Bipolar Is Common and Not Especially Related to Mania or Depression."
+
+The failure of surveys to generalize between populations shouldn't even be surprising when you consider the ambiguity and fuzziness of natural language: faced with a question, and prompted to give a forced-choice Yes/No or 1–5 response, people will assume the question was "meant for them" and try to map the words into some reference point in their experience. If the question _wasn't_ "meant for them"—if people who have never had a manic episode are given a set of questions formulated for a population of bipolar people—or if actual women are given a set of questions formulated for a population of males with a sex fantasy about being female—I think you _do_ get a lot of "Am I happy then sad sometimes? Sure, I guess so" out-of-distribution response behavior that doesn't capture what's really going on. In slogan form, [you are not measuring what you think you are measuring](https://www.lesswrong.com/posts/9kNxhKWvixtKW5anS/you-are-not-measuring-what-you-think-you-are-measuring).
+
+If Alexander is wary that a survey about moods done on a totally different population might not generalize to hospital inpatients, I think he should be still more wary that that a survey _about sexuality_ might not generalize to people _of different sexes_. Even if you're skeptical of most evopsych accounts of psychological sex differences (for there were no trucks or makeup in the environment of evolutionary adaptedness), sexuality is the one domain where I think we have very strong prior reasons to expect cross-sex [empahtic inference](https://www.lesswrong.com/posts/9fpWoXpNv83BAHJdc/the-comedy-of-behaviorism) to fail.
+
+This is why I expect the [standard cope](https://sillyolme.wordpress.com/2010/01/05/the-gostak-distims-the-doshes/) of "But cis women are autogynephilic too!!" to fall apart on further examination. I'm not denying the survey data itself (neither Alexander's nor [Moser 2009](http://unremediatedgender.space/papers/moser-agp_in_women.pdf)'s); I'm saying we have enough prior knowledge about what females and males are like, to strongly suspect that women who answer Yes to the same survey questions as AGP males are mostly in the position of saying that they got really happy and then really sad when their team lost the Super Bowl.
+
+The reason this isn't special pleading that makes my theory unfalsifiable, is because my skepticism is specifically about these mass survey questions where we haven't done the extra work to try to figure out whether the 1–5 question means the same thing to everyone; I'm happy to talk about qualitative predictions about what we see when we have a higher-bandwidth channel into someone's mind than a 1–5 survey response.