Bayeswatch 11: Parabellum — LessWrong