Sebastian Gäb has an interesting paper, Why you should be a religious skeptic (PDF)
In passing, I think it raises interesting points about a very common error, namely, assuming that the English words 'religion' and 'religious' actually identify a natural and coherent category of some kind. In particular, he is arguing that philosophers of religion should not be both realists and 'epistemic optimists' about 'religious statements'. Obviously the significant issue here is what a 'religious statement' is; and the fundamental problem is that whether we count something as a 'religious statement' depends entirely on the context.
Here is a 'religious statement': There exists some kind of world outside myself that continues to exist when I do not perceive it. Now, you may not classify this as a part of 'your religion'; it very definitely is part of mine, as Nicolas Malebranche famously noted, since I hold to a doctrine of creation, a doctrine of the Incarnation, and a religious ethics all of which require me to hold this. You may not have a commitment to this position that you would yourself classify as 'religious commitment'; I very much do. And it's not difficult to find accounts of this position that very clearly and specifically tie it to other statements that are often considered 'religious statements'; Malebranche is one, since he thinks religious faith is the only way to be certain about the existence of the external world, but also Berkeley and Shankara and many others, in a thousand different ways. You may well have particular arguments for the position that you yourself would not classify as 'religious'; but perhaps others would, and, even if not, nothing whatsoever requires a principle of conservation of religiousness -- nothing about a statement being classified as 'religious' requires that it can only be supported by statements that are also so classified. You might not attend the church picnic as a religious event, but nothing prevents it from being one for everyone else; you might not classify your experience of Bach as a religious experience, but others certainly could classify their own experiences as such.
In any case, as is clear from this, many of the things Gäb says about 'religious statements' are really things that would just apply to statements regardless of how one classifies them. (He himself recognizes this.) Gäb characterizes 'realism' about a certain class of statements as having three elements:
(1) Statements have propositional content.
(2) Statements have definite truth-values even if we do not know them.
(3) Statements may be irremediably undecidable by us.
I think a problem for Gäb's argumetn is that while a realist might accept (3), nothing about realism itself actually requires such an acceptance. For instance, if you're a realist about a small class of statements, you might well hold that all of those statements are not only decidable but actually decided; if you're a realist about a very large class of statements, you might hold that all of them are in principle decidable even though we could never actually consider them all. Nobody is a realist about everything, and everybody is a realist about something; because of this, realists are not technically committed by realism to a gap between what is known and what can be known, it's just that human limitations are going to split the two apart for many subject domains. But the gap between what is known and what can be known also does not actually give us (3), either, as is clear from the person who thinks that every particular thing we're talking about is in principle knowable by us but that in practice we just can't get to them all with the attention they would need.
Epistemic optimism Gäb characterizes as the view that, for the class of statements in question, it is rational for us to accept that at least some of our beliefs are true. Gäb wants to argue that realists can't be epistemic optimists, despite the fact that many are.
This is an initially odd claim, since a common reason for being a realist about a class of statements is that you are an epistemic optimist about some of them. Gäb's argument depends on a premise of what he calls 'epistemic Copernicanism', which he characterizes as the view that it is irrational to hold that we have an epistemically privileged position unless we have reason to think otherwise. I find it baffling why anyone would think this is true in general, since it would seem that we would eventually have to get to reasons that we just take ourselves to have because we are in a good position to have them. I take myself to have good reasons to think that there is a coaster on the table beside me, because I take myself to be in a good position to know it -- I am awake, I am sitting right here, I am wearing my glasses, I am healthy and in my right mind, and I see it. The reasons in this case just are the epistemically privileged position. What is more, I think everyone is a natural 'anti-Copernican' on this point -- we just assume that we are in an epistemically privileged position unless we have reason to think otherwise. That's certainly how I am with the coaster, and an infinity of other things.
Gäb's argument for epistemic Copernicanism is the principle of mediocrity: if we draw at random from a large set, we will be more likely to get results from the largest class of possible results. A problem with this, which besets a lot of use of probability in epistemology (Bayesianism, for instance), is that belief is not actually anything like drawing marbles from a bag, even at a very abstract level. We see immediately how this affects the argument, when it becomes clear that Gäb's argument is that realism requires that there is a gap between what is known and what can be known (as we've seen, it strictly speaking does not, but such gaps often exist) which is then combined with epistemic Copernicanism, taken as implying that it is probable that we are not actually capable of knowing any of these particular statements. But in reality, we do not form beliefs at random, like drawing from a large set of objects, and his only argument for epistemic Copernicanism depends crucially on the assumption that we do.
There is another argument that Gäb uses for his ultimate conclusion, the deep time argument, but it seems to me to that it assumes falsely that increasing the things that could be known makes it hard to know anything.
