One of my favorite lessons from Bayesianism is that the task of calculating the probability of an event can be broken down into simpler calculations, so that even if you have no basis for assigning a number to P(H) you might still have success estimating the likelihood ratio.