A key claim in Plantinga’s free will defense is that:
(TWD) Possibly, every creaturely essence suffers from transworld depravity.
To elaborate a bit: Say that a world is feasible if and only if it’s a possible world that God can at least weakly actualize. Then TWD asserts that there is a possible world W such that, from W, there is no feasible world W’ at which libertarianly free creatures freely do right on every occasion.
For our purposes, it’s important to note that TWD crucially involves the metaphysical possibility of libertarian agency. But suppose that knowledge of what’s possible is grounded in basic and inferential knowledge of what’s actual, and that knowledge of the possibility of libertarian agency isn’t grounded in basic or inferential knowledge of what’s actual. If so, then we don’t know whether libertarian agency is possible. And if not, then since Plantinga’s free will defense requires the metaphysical possibility of libertarian agency, it follows that we don’t know whether TWD is possibly true, in which case Plantinga’s free will defense fails to justify the claim that God and evil are compatible.
But perhaps one will reply that all Plantinga needs for the success of the free will defense is the mere epistemic possibility of TWD. Here we must be careful, though. For there are at least two construals of the notion of epistemic possibility:
Weak Epistemic Possibility: P is weakly epistemically possible if and only if P is logically consistent with what we know.
Strong Epistemic Possibility: P is strongly epistemically possible if and only if there's a decent chance that P is is the case, given what we know.
But clearly Plantinga’s free will defense cannot undercut the logical problem of evil if TWD is merely weakly epistemically possible. For then TWD is on a par with the skeptical scenarios associated with radical skepticism about perceptual knowledge, such as the possibility that we’re brains in vats, or the possibility that we’re in the Matrix. But just as the mere weak epistemic possibility of such skeptical scenarios is insufficient to undercut the prima facie justification enjoyed by our perceptual knowledge, so the mere weak epistemic possibility of TWD is insufficient to undercut the prima facie justification enjoyed by the key premises of the logical problem of evil.
But what if we construe it as a strong epistemic possibility? Clearly, Plantinga's Free Will Defense would be a success if TWD were strongly epistemically possible. Unfortunately, though, only a limited audience will find TWD to be strongly epistemically possible for them. Thus, consider John. John is convinced by careful evaluation of the relevant arguments and evidence that some non-libertarian theory of agency is true or plausibly true. Because of this, TWD isn’t strongly epistemically possible for John, in which case Plantinga’s FWD fails to deflect the logical or deductive argument from evil for him. Now here’s the rub: unless you fall within a certain subset of theists, or a much smaller subset of non-theists, you’re probably like John in your views about the nature of human freedom.
The upshot, then, is that if our knowledge of what's metaphysically possible is grounded in our knowledge of what's actual, then Plantinga’s free will defense is of very little help in responding to the logical problem of evil.