Approaching Logical Probability — LessWrong