Beyond Kolmogorov and Shannon — LessWrong