VNM expected utility theory: uses, abuses, and interpretation — LessWrong