Straggler writes:
Should not the default position, i.e. the position in the absence of any evidence always be disbelief rather than belief?
You criticize (some) religious people for the assumption that in the absence of any evidence regarding, for example, the existence of a soul, we should adopt a "flat prior" - all the competing hypotheses have equal probability of being correct.
There is actually a strong tradition in Bayesian inference of adopting a flat prior in some cases, and it is often used in evolutionary analysis! (each possible phylogenetic tree may be considered equally likely a priori, and we simply revise the prior probabilities assigned to each tree in the light of the molecules and the morphological characteristics of each taxon). So I must say that the religious people who propose a "flat prior" are not completely isolated from the scientific tradition in this regard.
The big difference lies in the fact that in science a flat prior is proposed only so that it can be modified under analysis of empirical evidence. If religious people propose a flat prior of the form "its 50/50 that the soul exists or doesn't exist" they should be aware that this has no bearing on the actual probability of the soul's existence - the prior probabilities are completely uninformative since we are unable to revise them in the light of empirical data. They are literally plucked from the air and represent nothing more than our ignorance. Since the number of possible religious beliefs is unbounded, the prior probability of any specific belief tends toward zero. Furthermore there is no consensus on how these priors ought to be chosen. Choosing a flat prior may be convenient but it is essentially a speculation, the choice cannot be justified with respect to alternative priors.
If the priors cannot be revised, then there is no point in having them. This is the sense in which religious views are "not even hypotheses". Religious people can not call upon Bayesian inference and the presumption of a flat prior if they at the same time claim that no empirical evidence can be brought to bear on the question.
While there is an intuitive appeal to the notion that we should treat competing unfalsified claims as having equal likelihood, I agree with you that an attitude of scepticism is usually the most rigorous and coherent one. This is after all why there exists the concept of the null hypothesis - the null hypothesis is often simply that some postulated pattern in the world is actually the result of a random process. For example if we want to test the hypothesis that Germans are taller than the French, we would calculate how likely the empirical data regarding German height is to result from a null distribution in which the heights of French and German people are drawn from a single random probability distribution. So the null hypothesis is simply "there is no intersting pattern here that needs to be explained" and somebody who disagrees must demonstrate it.
The null hypothesis in questions of religion is also "there is no interesting pattern here that needs to be explained". If religious people are arguing that no empirical evidence can be brought to bear upon some religious question, then they are essentially agreeing with the null hypothesis. There is no interesting empirical pattern to be explained with respect to the existence of the soul or Vishnu. If religious people really do make such arguments, then it is gratifying that they agree with the atheists! We can happily proceed to make ethical judgements based on the null hypothesis that there is no soul nor Vishnu.