x
Locally optimal strategies — LessWrong