Gratuitous evil is evil that God would not permit. In Alvin Plantinga’s terms, God would not actualize such evils either strongly (i.e. by directly creating them) or weakly (i.e. by allowing free creatures to commit them). A gratuitous evil is one that God would have no morally sufficient reason for actualizing (strongly or weakly). God has a morally sufficient reason for actualizing an evil e in a world W if and only if e is necessary in W for some good g, and W, containing both e and g, is better (by whatever criterion) than any world W*, actualizable by God, where W* contains neither e nor g. Put simply, where W is the real world, God permits evil e because e is necessary for the realization of good g, and g is realized, and the world with both e and g is better than it would be with neither e nor g.
The hackneyed example is that moral evil is permitted in the world because moral evil is necessary in our world for freely-chosen moral goodness, and there is no alternate world, actualizable by God, in which the overall balance of moral goodness to moral evil is better than in the real world. Perhaps there are possible worlds in which free creatures always choose to do good, and so, in those worlds, moral evil is not a necessary condition for moral goodness. However, perhaps no such world is actualizable by God. Perhaps, all possible free creatures suffer from what Plantinga calls “trans-world depravity,” that is, the “counterfactuals of freedom” (over which God has no control) are such that every free creature will freely choose to do some evil in any world in which it exists. In this case, not even God can create worlds with free creatures and no moral evil. Therefore, so far as human creatures can know (since we cannot know the counterfactuals of freedom) perhaps even the grossest moral evils are not gratuitous. I would add that, since we cannot know the counterfactuals of freedom, then neither can we know that moral evils are not gratuitous. Perhaps God could have created a better world after all.
I have to confess that, despite Alvin Plantinga’s gigantic reputation, when we start talking about “trans-world depravity” and “counterfactuals of freedom,” then I am afflicted with a palpable and undeniable sense that we have flown off into Cloud-Cuckoo-Land. Let me see if I can put the problem in less recherché terms: First, what would be a candidate for a gratuitous evil, that is, what kind of an event, if it did occur, would require a morally sufficient reason if it were preventable yet not prevented? I think that instances of unwanted and undeserved suffering would be our candidates. Some suffering may be desired. One who fancies himself as a Nietzschean übermensch, may welcome suffering as travails that the great must endure to prove their superiority. Other suffering may be deserved. Hitler’s suffering in his final days in the bunker need not excite our sympathy.
Unquestionably, however, there are and have been, among human and non-human animals, many cases of unwanted, undeserved suffering. How many? No one can say, but it seems reasonable to posit that, since organisms achieved the neurological complexity to suffer (far back in the Paleozoic, no doubt), there have been a trillion (1012) instances of unwanted, undeserved suffering. Note that if even a single one of these 1012 instances of suffering is gratuitous, that is, if even one is such that God would have no morally sufficient reason for permitting it, then God does not exist. If one Diplodocus suffered unnecessarily in the Jurassic, then there cannot be a perfectly good and all-powerful being who, by definition, would have prevented such suffering and could have prevented such suffering. In other words, if only one of the trillion possible cases of gratuitous evil truly was in fact gratuitous, then God does not exist.
“Skeptical theists” deny that we can know of any particular evil that it is likely to be gratuitous. Their argument is that we are not in the epistemological position to know the goods that omnipotence can achieve over the course of eternity. God is postulated to be omnipotent and to have all of eternity in which to fulfill his plans. Therefore, we cannot be sure, of any particular evil, and however pointless it appears to us, that God will not generate a redeeming good that at some future time (perhaps in the distant future) will serve to redeem that evil. That is, sub specie aeternitatis, the world will be better with that evil and its redeeming good than with neither the evil nor the good. However, the real problem that theists face here is not whether any particular evil can be known to be gratuitous, but whether any of the postulated 1012 evils is gratuitous. Even if each such evil has only a very small chance of being gratuitous, the disjunction of those small probabilities might sum up to virtual certainty. Put equivalently, even if it is highly probable (but not certain) that each instance of unwanted, undeserved suffering is not gratuitous, then the product of those probabilities, the probability that NO such instance is gratuitous still might be very small—close to zero.
It appears, then, that if they are to reasonably deny that any instance of unwanted, undeserved suffering was in fact gratuitous, theists must be confident that none of those 1012 instances of undeserved, unwanted suffering was gratuitous. Merely arguing that it cannot be known that no particular one of these was gratuitous will not do the job here. Not even close. I can give very good reasons for saying that I will not be involved in a fatal accident the next time that I drive, but I cannot say with much confidence that over a, say, sixty year driving career, I will not be involved in such an accident. What grounds, then, can a theist have for being confident that no instance of unwanted, undeserved suffering was gratuitous?