Integral versus differential ethics — LessWrong