Superintelligence 14: Motivation selection methods — LessWrong