"incomparable" outcomes--multiple utility functions? — LessWrong