This is an Scientific American article on the usefulness of Bayes' theorem in assessing such things as theism.
Well, well, of course, of course. Bayes' theorem doesn't give you objective antecedent probabilities. So we have to go with the ones we've got, and conditionalize our beliefs based on the evidence we received? Why go with the ones we've got? Can you suggest any others, without introducing an incoherence in our beliefs?
That's why I'm an annoying Bayesian subjectivist when it comes to prior probabilities. Let people have the antecedent probabilities that they have, and let us see what evidence moves the scale up or down.
This raises an issue, I realize, for advocates of the "outsider test." One one level, we can't get outside of our belief system and shouldn't try. Otherwise, we end up in a general skepticism, which is where Descartes should have landed if he had been consistent. But if the outsider test argument works, isn't it an argument attempting to show that Christian theism is less probable than you originally thought it was?