There is an important but little-known critique of Plantinga's Reformed Epistemology by "star" epistemologist and Christian Keith DeRose (Yale). You can find it here, at DeRose's Yale department webpage.
See also his newer paper on Direct Warrant Realism in the recent phil. of religion collection, God and the Ethics of Belief, edited by Dole and Chignell. It poses problems for Alston's religious epistemology as well. You can read an earlier draft at his department homepage as well, here.
The way the cosmological argument from contingency is often put, it sets things up in the mind of the reader that there are logically only two possible sorts of beings
(i) contingent and dependent beings. (ii) necessary and independent beings.
So that anything that is contingent is dependent on something else for its existence, and anything that is independent exists of logical or metaphysical necessity. But one wonders why this must be so. Why can't there be a third possible sort of being:
(iii) contingent but independent beings.
Beings of type-(iii) are like beings of type-(i), in that they exist in some worlds but not in others. But they are also like beings of type (ii), in that they depend on nothing else for their existence: they're free-standing beings (incidentally, this is how Swinburne thinks God is. If it's ok and coherent for there to be a god like this, why not for some non-divine beings?).
Thus, here's an epistemic possibility: there are just two sorts of be…
I haven't seen this argument made explicit by Moreland or Craig, but I believe the materials for the following dilemma can be gleaned from their writings:
The Dilemma: If a traversal of an infinite past is possible, then either the traversal requires a starting point or it doesn't. If it does, then the traversal could never get going, as one could never get a foothold in the beginningless series to begin the traversal (for there are an infinite number of moments before every day in such a past, and so to start at any point, you would have already had to have traversed an infinite number of moments to get to it). But if it doesn't, then the traversal should always be finished, for an infinite number of moments exist before every day in such a past, and thus all the time needed to complete the traversal exists before every day. Therefore, either the beginningless traversal can never get going or it's always completed. But both implications are absurd or otherwise unaccepta…