How should I optimize my decision making model for 'ideas'? — LessWrong