Thing of Things Transgender Intellectual Turing Test Predictions and Commentary

Friend of the blog—I mean, I hope we're still friends even though I'm kind of trying to overthrow them as de facto Gender Czar of the Less Wrong diaspora—Ozymandias of Thing of Things has been running an intellectual Turing test challenging adherents of the gender-identity and two-type theories of transgenderedness to try to impersonate each other for the good of our collective epistemology!

(An aside on credit-assignment and the history of ideas: Ozy says Blanchard–Bailey where I've usually been trying to say two-type in order to avoid the tricky problem of optimal eponymy, but if you are going to be eponymous about it, I can understand just saying "Blanchard" but feel like it's unfair to include Bailey but not Anne Lawrence. My understanding of the history—and I think Michael Bailey reads this blog and I trust him to send me an angry email if I got this wrong—is that Bailey's research had mostly been about sexual orientation and from-childhood gender nonconformity, not the two-type taxonomy as such. Bailey's popular-level book The Man Who Would Be Queen drew controversy for explaining the two-type taxonomy for a nonspecialist audience (in the last part of a book that was mostly about the androphilic/feminine-from-early-childhood people, not my people), but the critics who disparage Queen as "unscientific" are missing the point: popular-level books that present a scientific theory aren't supposed to capitulate all the evidence for the theory—for that, you need to follow the citations and read the primary literature for yourself. In analogy, it should not be construed as a disparagement of R. Dawkins to note that it would be weird if people talked about the "Darwin–Dawkins theory of evolution"!)

In the intellectual Turing test, contestants answer a set of questions both as themselves, and while trying to pass as someone who believes the other thing, while the audience tries to discriminate the honest entries from the fakes. Below are my probability assignments for this contest (I think it's important to assign probabilities rather than binary guesses, so that you can assess your rationality with a Bayesian strictly proper scoring rule rather than a crude "number correct"), along with an optional brief comment—

Update, 5 June: Two months after the results were posted, I finally got around to scoring these. ("Bayes-score" is the base-two logarithmic score. Someone who, claiming complete ignorance, gave a 0.5/0.5 distribution for each entry would lose a bit on each question for a final score of −18.)

Gender identity entries

#1: GI: 0.65, BBL: 0.35 (strong philosophy of language; if telling the truth about being a cis woman, ignorance of non-dysphoric AGP is plausible), Actual: GI ✔, Bayes-score: −0.621
#2: GI: 0.4, BBL: 0.6 (awareness of 4chan shows non-naïveté about what's actually going on), Actual: GI ✘, Bayes-score: −1.322
#3: GI: 0.6, BBL: 0.4 (maybe a little too doctrinaire??), Actual: BBL ✘, Bayes-score: −1.322
#4: GI: 0.6, BBL: 0.4, Actual: BBL ✘, Bayes-score: −1.322
#5: GI: 0.6, BBL: 0.4, Actual: GI ✔, Bayes-score: −0.737
#6: GI: 0.7, BBL: 0.3 (seemingly sincere trans man), Actual: GI ✔, Bayes-score: −0.515
#7: GI: 0.7, BBL: 0.3 (standard trans woman rationalizations), Actual: GI ✔, Bayes-score: −0.515
#8: GI: 0.65, BBL: 0.35 (really knows her stuff; this is what a smart, intellectually-honest BBL skeptic looks like, and I'd like to believe that they exist!), Actual: GI ✔, Bayes-score: −0.621
#9: GI: 0.7, BBL: 0.3, Actual: BBL ✘, Bayes-score: −1.737

Blanchard–Bailey–Lawrence entries

#1: GI: 0.6, BBL: 0.4, Actual: GI ✔, Bayes-score: −0.737
#2: GI: 0.4, BBL: 0.6, Actual: GI ✘, Bayes-score: −1.322
#3: GI: 0.4, BBL: 0.6, Actual: BBL ✔, Bayes-score: −0.737
#4: GI: 0.9, BBL: 0.1 (shibboleth fail!—people who believe in biology do not say "assigned at birth" when describing their own beliefs! Also, failure to notice the obvious "for the same reasons men are" re programmers), Actual: BBL (!!) ✘, Bayes-score: −3.322
#5: GI: 0.2, BBL: 0.8 (preach it!), Actual: GI ✘, Bayes-score: −2.322
#6: GI: 0.8, BBL: 0.2 ("male socialization, which unlike androphilic trans women they actually tend to absorb as kids" sounds like someone who believes that innate gender identity determines what socialization you latch onto from your culture, rather than someone who actually believes in sexual dimorphism), Actual: GI ✔, Bayes-score: −0.322
#7: GI: 0.9, BBL: 0.1 (shibboleth fail again!—my comment at Thing of Things), Actual: GI ✔, Bayes-score: −0.152
#8: GI: 0.1, BBL: 0.9 (raw reality), Actual: BBL ✔, Bayes-score: −0.152
#9: GI: 0.85, BBL: 0.15 (my comment), Actual: GI ✔, Bayes-score: −0.234

Proportion correct (construing assignment of probability greater than 0.5 to the actual answer as "correct"): 11/18
Total Bayes-score: −18.012 (just worse than chance)

Submit to Reddit

(Post revision history)

Comments permit Markdown or HTML formatting.