To put this another way: if you made a model combining some measure of “how hard is this chemical to obtain / how hard is this lifestyle intervention to practice?” and “how novel and high-tech does it feel?”, plus one or two other things like “is this a stimulant?”, it feels like this would predic the results almost perfectly. Does anything stand out as doing substantially worse than the simple model would predict?

Scott Alexander
“Troof On Nootropics”
Astral Codex Ten