Half of what people describe as tolerance is a change in the underlying condition, and the two need different answers.

Domain age is one of the few signals that is expensive to fake and free to check.

Where does the human data stop and the extrapolation begin? That is the whole thread.