A terrifying variant of Boltzmann's brains problem — LessWrong