Against the weirdness heuristic — LessWrong