My current best guess is that singular learning theory has the right conceptual picture for the “selection theorems” prong, and the folks working on it are headed in the right direction.
News to me
News to me