Bayesian problems of old evidence