I'm more than happy to discuss Plantinga's Free Will Defense further with those interested (see previous post), but for now, here's my tentative summary and conclusion on the matter, prefaced with some contextual stage-setting:
A standard way to state the deductive argument from evil is the one we've inherited from Hume in his Dialogues Concerning Natural Religion (who in turn inherited it from Epicurus):
"Is he willing to prevent evil, but not able? then is he impotent. Is he able, but not willing? then is he malevolent. Is he both able and willing? whence then is evil?"
The reasoning here can be teased out as follows:
1. Evil exists. (Premise)
2. If God exists, then God is omnipotent, omniscient, and perfectly good. (Premise)
3. If God is omniscient and omnipotent, then He is able to prevent evil. (Premise)
4. If God is perfectly good, then He is willing to prevent evil. (Premise)
5. If God is willing and able to prevent evil, then evil does not exist. (Premise)
6. Therefore, God is not both willing and able to prevent evil. (From 1 and 5)
7. Therefore, God is not omnipotent, omniscient, and perfectly good. (From 3, 4 and 6)
8. Therefore, God does not exist. (From 2 and 7)
A standard response to the argument is the appeal to free will: God takes no delight in robots; he wants a community of persons who have the ability to freely enter into a relationship with himself and others. Furthermore, he wants a world that contains not just aesthetic beauty, but also moral goodness that's grounded in the free actions of his creatures. But the latter is a great good that can only be gotten by creating free creatures in the first place. But once you do that, you must allow for the possibility that they'll abuse that freedom and freely choose to do wrong. That is, it's logically impossible for God to *force* his creatures to *freely* do right. Thus, since it's no limit on omnipotence to be unable to do the logically impossible, premise (4) is false, and the argument is unsound.
But then along came J.L. Mackie. He pointed out that if a person is free to do wrong, then it follows that it's possible for such a person to freely do right. Thus, there are possible worlds at which people freely do right at least some of the time. And if it's possible to freely do right some of the time, then on the basis of the same sort of reasoning, it's possible for free creatures to always freely do right all of the time. Thus, there are possible worlds in which free creatures always freely do right. But if so, then contrary to the old version of the free will defense, it's not beyond God's omnipotence to create a world in which free creatures always do right, in which case premise (4) remains unscathed, leaving the deductive argument from evil unrefuted.
But then along came Alvin Plantinga. In God and Other Minds, in The Nature of Necessity, and in God, Freedom, and Evil, he took issue with Mackie. In particular, he argued that Mackie falsely assumes that God can do whatever is logically possible (Plantinga calls this assumption, "Leibniz Lapse"). But this claim is false -- there are possible worlds that not even an omnipotent God can actualize. So, for example, suppose that in some possible world W, some person -- call him "Steve" -- freely chooses to do something morally wrong. Then it's not possible for God to actualize **W** (that very world) and yet prevent Steve from doing the wrong action. For the counterfactuals of freedom are fixed by what people freely do in a given possible world, and are thus restraints on what God can do at that world. Plantinga took this insight about counterfactuals of freedom and used it to develop his notion of "transworld depravity" (or'TWD' for short). In a nutshell, and very roughly, a person suffers from transworld depravity if, among the worlds *that God can actualize*, that person would perform at least one wrong action *in those worlds* -- the God-actualizable worlds -- no matter what circumstances or sequence of circumstances God puts them in.
Now it turns out that there are two basic interpretations of TWD in the literature: what I'll label as 'the 80's interpretation' and 'the 90's interpretation'. According to the 80's interpretation, a person suffers from transworld depravity just in case they would freely do at least one wrong thing in *every* God-actualizable world. But according to the 90's interpretation, the definition is relativized to a given possible world: a person suffers from transworld depravity *in a given world W* just in case God can't actualize **W** (including any of the worlds *counterfactual* to W) without that person performing at least one immoral action. This is a much more convoluted account than the 80's interpretation. Perhaps the simplest way to put it is this: Suppose you exist in possible world W1, and you are transworld depraved at W1. Then no matter what God could have done *at W1* (and at worlds *counterfactual* to W1 ) in order to ensure that you always freely do right, it won't work -- you'll blow it at least once.
The two interpretations differ in that the 80's version makes free creatures out to be transworld depraved in all possible worlds in which they exist -- i.e., as an ordinary essential property. However, the 90's interpretation is more subtle. It takes transworld depravity to be a *contingent* property of free creatures: they're transworld depraved at some worlds, but not at others. Furthermore, according to the 90's interpretation, transworld depravity is defined in terms of *counterfactual worlds* -- i.e., just the worlds counterfactual to a given world W -- while the 80's interpration defines it in terms of ordinary possible worlds.
With these distinctions in mind, we can now see the main idea of Plantinga's Free Will Defense: What if *every* possible free creature suffers from transworld depravity? That is what if:
(<> TWD) Possibly, every creaturely essence suffers from transworld depravity.
According to the 80's interpretation, this means that for all we know, every world that it's in God's power to create will have at least one free creature that performs at least one morally wrong action. If so, then God's hands are tied as to what worlds he can create. In particular, he would be unable to create a world with free creatures (and thus moral good) without also creating a world that contains moral evil. If so, then even though there are possible worlds at which free creatures never do wrong, God can't create any of them. But if not, then Premise (4) is false, and the deductive argument is unsound -- or at the very least, we can't rule it out as false that it's possible that every free creature suffers from transworld depravity, in which case premise (4) is at least dubious.
According to the 90's interpretation, the implications are a bit different. It implies that for all we know, there is at least one possible world W at which all possible free creatures exist, such that no matter what God could've done at W, at least one free creature there performs at least one wrong action at W. If so, then God's hands are tied with respect to what he can do *at W*. In particular, he would be unable to create *W* with free creatures (and thus moral good) without W also containing moral evil.
Now in light of the above, and in light of my previous posts, it seems to me that the Free Will Defense is subject to the following dilemma:
Either the 80's interpretation of transworld depravity is correct or the 90's interpretation is correct. Now if the 80's interpretation is correct, then (<>TWD) is only weakly epistemically possible at best (for non-theologically conservative Christians) and flatly false at worst (for theologically conservative Christians) -- either way it's not a successful defense against the deductive argument from evil (see my "On the Force of "Possibly" in Plantinga's Free Will Defense" for the details of the argument for this horn of the dilemma).
On the other hand, if the 90's interpretation of transworld depravity is correct, then (<>TWD) is subject to the following to criticisms (See my last post for the sources in the literature for these criticisms):
(i) Plantinga hasn't succeeded in showing that such a world (i.e., a TWD world) is *metaphysically* possible. To see this, consider the following: for all we know, it's possible that, necessarily, some essence or other is blessed with *transworld sanctity* (TS). But if we can't rationally rule this out, then since it's incompatible with (<>TWD), then we're only justified in taking (<>TWD) to be *epistemically* possible at best.
(ii) But more importantly, even if we grant that (<>TWD) is metaphysically possible, the correct response would then be,"so what?" If TWD is true at W, then this only prevents God from actualizing free creatures who never blow it *if* he freely chooses to actualize *W*. But there are *other* possible worlds at which free creatures don't have these cruddy counterfactuals of freedom, and thus always freely do what is morally *right*. If so, then Mackie's objection remains: why didn't God actualize one of these *other* worlds?
Therefore, on either interpretation, apologists are being misleading when they claim that Plantinga has refuted the deductive argument from evil. At best, he's shown that we can't be confident that the deductive argument from evil is sound.