A technical note: Bayesianism is not logic, statistics is not rationality

Per­haps I am beat­ing a dead horse for this com­mu­nity, but this is a very lu­cid ex­pla­na­tion of what prob­a­bil­is­tic/​statis­ti­cal rea­son­ing can­not do. Namely: first or­der logic. There’s re­ally no way of en­cod­ing re­la­tions or quan­tifiers into purely Bayesian in­fer­ence, which ac­tu­ally makes it quite weak in terms of model build­ing.

Fur­ther, in­te­grat­ing prob­a­bil­ity and logic is a huge un­solved prob­lem! We ac­tu­ally have very lit­tle idea how to com­bine our two great­est suc­cesses in for­mal­iz­ing ra­tio­nal­ity.

I found this tremen­dously clar­ify­ing, though not im­me­di­ately use­ful. But it has definitely broad­ened my think­ing.