Skirting the mere addition paradox — LessWrong