The novelty quotient — LessWrong