If it were _just_ a matter of different priors (where my stronger [inductive bias](https://en.wikipedia.org/wiki/Inductive_bias) lets me learn faster from less data, at the cost of [being wrong in universes that I think mostly don't exist](https://en.wikipedia.org/wiki/No_free_lunch_in_search_and_optimization)), I would expect you to express more uncertainty. I would _totally_ respect it if you were merely _uncertain_ about the AGP→gender-ID _vs._ gender-ID→AGP causality. [I _agree_ that causality is _much harder_ to pin down than mere correlation.](http://unremediatedgender.space/2021/Feb/you-are-right-and-i-was-wrong-reply-to-tailcalled-on-causality/)
But on Discord, you said "it just seemed totally wrong"!! If you're _not_ playing a "does the evidence permit me to believe" game, I just don't see how you think the _SSC_ survey data is powerful enough to answer that question one way or the other! If I had a prior belief that invisible dragons were plausible, I would remain _agnostic_ about the no-dragon _vs._ invisible-dragon hypotheses upon seeing an apparently empty garage. But to say that the no-dragon hypothesis "just seems totally wrong" ... ?!?!
+
+-----
+
+If we have a big causal graph with C at the root (E₂ ← E₁ ← C → E₃ ...) with real-valued variables, and someone proposes a theory about what happens to the E_i when C is between 2 and 3 or between 5 and 6 or above 12, that's very unparsimonious: why would such a discontinuous hodge-pause of values for the cause, have consistent effects?
+
+In my worldview, "gender" (as the thing trans women and cis women have in common) looks like a hodge-podge as far as biology is concerned. (It can be real socially to the extent that people believe it's real and act accordingly, which creates the relevant conditional indpendence structure in their social behavior—but sexuality looks more "biological" than "social".)
projects / Ultimately_Untrue_Thought.git / blobdiff