Here's a thought I'm toying with: Suppose you agree with a number of epistemologists that the way things seem or appear to one is defeasible, prima facie evidence for the way things are. To use Plantinga's terminology: beliefs that appear to one to be true are "properly basic" for one. Now suppose further that it appears to some person -- let's call her 'Sue' -- that nothing could morally justify some of the evils that occur in the world (the Holocaust, for example). Then it follows that Sue has defeasible, prima facie evidence that nothing could justify some of the evils that occur in the world. Finally, assume that Sue accepts the fairly standard view that an all-powerful, all-knowing, perfectly good god wouldn't allow an evil to occur without sufficient moral justification. Then Sue would have a good reason to think there is no all-knowing, all-powerful, perfectly good god.
How could one defeat Sue's basis for her atheism? If we assume the truth of the premises that (i) the way things appear is defeasible, prima facie evidence for the way things are, and that (ii) no all-knowing, all-powerful, perfectly good god would allow evil without sufficient moral justification, this will amount to offering Sue, at a minimum, an undercutting defeater for her belief that nothing could justify certain evils in the world. But that belief is based on a seeming, and to undercut such a seeming-backed belief, it's not enough to show her that it's epistemically possible that God has a good reason for allowing these evils to occur. Rather, one would have to provide her with a good reason for thinking that it's metaphysically possible that God has a good reason for allowing these evils to occur.
What, then, could one offer Sue to undercut the basis of her atheism? I imagine some would be inclined to tell her all about Plantinga's free will defense. Here's the rub, though: Plantinga's free will defense seems to be a mere epistemic possibility. But if that's right, then how could Plantinga's free will defense undercut Sue's unbelief?
But surely, you might say, it's odd to think that anyone is in Sue's predicament. But is that really true? Richard Swinburne, for example, thinks the proposition that nothing could justify some of the evils in the world is (absent defeaters) properly basic for just about any rational, mature adult who reflects on the matter; I'm inclined to think he's right about that.
Anyway, as I said, it's just a thought I've been toying with. I should also say that it's not original with me. As I mentioned above, Swinburne has raised it. Furthermore, Paul Draper raises the issue in his review of Swinburne's Providence and the Problem of Evil.
 Some recent proponents of this sort of principle (e.g., Michael Huemer) offer a weakened version of it, according to which seemings provide at least prima facie pro tanto justification for the way things are. To avoid this debate, let's stipulate that Sue's seeming is sufficiently forceful to furnish her with very strong justification for her belief.
 I'm using the notion of epistemic possibility here in a way that's more in keeping with its occasional usage in the contemporary epistemology of modality literature than with epistemology proper. Thus, say that something P is metaphysically possible just in case there is a genuinely viable or possible world W at which P obtains, and say that P is epistemically possible just in case our evidence can't rule out that there is a metaphysically possible world W at which P obtains.
On this way of making the distinction, there is room for some P to be epistemically possible -- i.e., our evidence doesn't rule out that there is a metaphysically possible world W at which P obtains --, and yet P is nonetheless metaphysically impossible.
Example: consider Alvin Plantinga’s modal ontological argument. The key premise is that, roughly, there is a possible world at which a necessarily existent individual exists. But given Axiom S5 – the one that says that what’s possibly necessary is necessary simpliciter -- the truth of the key premise entails that Plantinga’s god exists. Now suppose my evidence can't rule out the God described by Plantinga as metaphysically impossible. Then Plantinga's god is epistemically possible. But if epistemic possibility entails metaphysical possibility, then it follows that Plantinga's god exists.
Perhaps some will find no fault with the conclusion above. But now consider Peter van Inwagen’s “knownos” . A knowno is a being that knows there are no necessary beings. Now my evidence doesn't rule out the metaphysical impossibility of knownos. So knownos are also epistemically possible, in. So if epistemic possibility entails metaphysical possibility, then necessary beings are metaphysically impossible.
Something has gone wrong in our reasoning. For necessarily, only one of the two claims above can be true. For an Anselmian being is possible just in case a knowno is impossible. Diagnosis: what has gone wrong is that we have assumed that epistemic possibility entails metaphysical possibility.
Here's the payoff (and the punchline): Plantinga has only shown that the heart of his free will defense -- his thesis of transworld depravity -- is epistemically possible. and the problem is that that's not enough to undercut Sue's justification for believing that there is no metaphysically possible world W such that (i) God exists at W and (ii)some tokens of the types of evils that obtain at the actual world obtain at W.
Review of Draper and Schellenberg (eds.), <I>Renewing Philosophy of Religion: Exploratory Essays</I>
Adam Green reviews the book for NDPR.
0. Introduction 0.1 Mackie argues that the problem of evil proves that either no god exists, or at least that the god of Orthodox Judaism, ...
Notes on Swinburne’s “Why God Allows Evil” 1. The kinds of goods a theistic god would provide: deeper goods than just “thrills of pleasure ...
"...[O]ne can have a system of beliefs that is similar to those which Plantinga describes, involving massive misconceptions which are p...