How do we know that some instance of evil is gratuitous? I think that there is much to say in favor of the idea that we simply see that the evil is gratuitous. That is, in observing some bad event, I observe directly that this event is neither necessary for the occurrence of some compensating good nor for the prevention of some worse evil. I see, for example, a child fall while trying to climb a tree and then start crying because of a sliver stuck in the palm of her hand. I don’t think, “Well, maybe the pain she is experiencing is good for her. Or maybe it will bring about unknown further goods.” No, I know that the pain is bad, that it isn’t doing her (or anybody else) any good, and so I get out the tweezers and remove the sliver. And I do this automatically without consciously pondering over the circumstances or evening making any kind of explicit inference. So it seems to me perfectly reasonable to say that, in at least many instances, we know directly that instances of evil are gratuitous. Now, of course, as with all cases of perception, the fact that I am apparently seeing a case of gratuitous evil is not a perfect guarantee that I actually am seeing a case of gratuitous evil. There are undoubtedly instances of illusory experiences of gratuitous evil; instances in which what appears to be gratuitous is not actually gratuitous. Nonetheless, it seems to me that it is reasonable to treat an experience in which I seem to witness an instance of gratuitous evil as a defeasible veridical perception. We can be wrong in our assessments of whether some evil is gratuitous, but in the absence of compelling evidence, we ought to trust our initial judgements concerning whether some evil is gratuitous.

If all of this is right, then the suggestion that there are unknown (and perhaps unknowable) compensating goods whose promotion would justify a divine being’s failure to intervene to prevent some instance of evil entails that we are systematically deceived about the existence of gratuitous evil. There are two points that I want to make about such a claim. First, we should recognize the way in which such a suggestion parallels arguments for external world skepticism. These arguments claim that the judgements that we arrive at concerning the existence of objects external to our minds are undermined by the possibility that there are unknown phenomena (evil demons, for example) that ensure that our experiences of the external world are illusory (i.e., that ensure that when I seem to see a computer in front of me, I am deceived; there is no computer). And since we are not in a position to know whether such unknown phenomena exist, we are not in a position to say that our experiences of external objects are veridical. If this parallel is indicative of an underlying analogy, then the suggestion that there are unknown goods that justify God’s allowing the occurrence of evils is epistemically on a par with the claim that I am being deceived by an evil demon. This would be a devastating result since the kind of skepticism at work here threatens to undermine all claims to knowledge, including any claim about the nature and existence of God. The argumentative tactic (known as skeptical theism), taken to its logical conclusion, might end up defeating the argument from evil at the expense of losing all grounds for theism.

So, I would suggest that we are better off concluding that the natural and mostly automatic judgements we make about gratuitous evil are generally reliable; that we should trust them unless we are provided, in specific circumstances, with a compelling reason for thinking that their deliverances are wrong. This is especially compelling in the light of the importance that recognizing gratuitous evil plays in our moral assessments and in moral decision making (a point that Stephen Maitzen has effectively employed in an argument against skeptical theism). But, and this is the second point, even if we are not convinced that we know directly whether most instances of evil are gratuitous, we can nonetheless make very reasonable inferences to this effect. Those who suggest that because of our cognitive limitations we are not in a position to know whether any particular instance of evil is gratuitous are underestimating the power of such inferences.

The inference that I will describe is more narrowly focused than what I’ve talked about above. I will be examining the inference from the occurrence of apparent horrors to the existence of gratuitous evil. Focusing on horrors is useful because it is much more clear both that most horrors appear gratuitous and that we have good reason to think that they are gratuitous. I will be using ‘horror’ in the following, fairly common sense: instances of evil of an extreme duration, magnitude or intensity; the kind of evil that, as Marilyn McCord Adams has said, provides the sufferer with a reason to think that life is not worth living. Let’s start with some important observations:

Horrors are significant for us because they raise special problems concerning how, if at all, they can be compensated. Dostoevsky famously illustrated the problems in an important section of The Karamazov Brothers. In discussing whether God might have reasons for permitting suffering, Ivan Karamazov asks his brother Alyosha,

Imagine that you are creating a fabric of human destiny with the object of making men happy in the end, giving them peace and rest at last, but that it was essential and inevitable to torture to death only one tiny creature- that baby beating its breast with its fist, for instance- and to found that edifice on its unavenged tears, would you consent to be the architect on those conditions?

The problem that Ivan is raising is that for some evils it does not seem appropriate to consider compensation (in the normal sense) when considering what would permit us to justifiably allow the evil to occur. In normal cases of evil, it is appropriate to think in terms of compensating goods. In other words, for many instance of run-of-the-mill evil, all that is required by way of justifying compensation is some good that outweighs the evil, such that the evil is required in order for the good to be realized. In cases of horrors, this notion breaks down.

There are different ways of thinking about what is required in order that some evil be compensated. Having healthy teeth and gums is more valuable than not experiencing the pain of a root canal, and so, given that we cannot achieve the good without the evil, we rightly choose to undergo the pain of surgery knowing that this evil is well compensated. This is an easy example, but it is not obvious what general conclusion we can draw about compensation for evils. I don’t think we actually need any kind of robust account of compensation however; we can make do with some general observations. To begin, let’s start with a fairly minimal conception:

An instance of evil, e, is compensated just in case there is some good state of affairs, g, that is at least as good as the state of affairs consisting of the absence of e.

This is a fairly weak notion of compensation (we could call it “weak compensation”), but it will suffice as a starting point. Note that the fact that some instance of evil is weakly compensated is not sufficient to justify allowing it to occur. If I steal your car today, then feel bad about it and so return it to you unharmed tomorrow, my stealing your car does not thereby become justified. In order for me to be justified for knowingly bringing about or causing some evil, I must, at least, be aware of good reasons to believe that this evil is required for the occurrence of some compensating good.

The point of Ivan Karamazov’s speech (well, one of the points, anyway) is that there are some evils that are of such magnitude that it does not seem to make sense to speak of compensation (at least in the sense of compensation described above). When we consider the horrors of the holocaust or of children suffering and dying from starvation or horrendous diseases, it seems rather inappropriate to consider the possibility of any old compensating good whose realization might justify permitting these horrors to occur. It is certainly not enough, in these cases, to point to states of affairs that are at least as good as the absence of the horror and are such that they cannot occur in the absence of the horror. More carefully, for at least some horror, h, even if there is some state of affairs, s, that is at least as good as the absence of h, and s cannot occur in the absence of h, s still does not compensate for h.

Ivan’s claim is very strong; he says that even the good that consists of all of humanity being happy and achieving peace and rest would not compensate for the evil of torturing one child to death. The good that Ivan imagines has incalculable value. Billions of people happy and at peace has more positive value than a single child being tortured to death has negative value. Nonetheless, Ivan’s brother Alyosha does not hesitate when he responds to Ivan’s question by indicating that he would not consent to have the one child tortured if it was necessary to bring about this incalculable good. And I think that many of us (most of us?) would agree with Alyosha.

So my observation is this: when it comes to horrors, it is not sufficient for them to be compensated with goods that merely have equal value to the absence of the horror. Either something more is required or else some horrors cannot be compensated. If they can be compensated, then I think we should admit that we we don’t know what the something more is that would constitute compensation. But this is not particularly relevant for my argument. What matters is that we understand that compensation in the sense exemplified by the tooth example is not enough.

Now we can describe scenarios in which, for example, because of the whims of a powerful madman, in order to prevent the torture and death of five children it was necessary to torture to death one single child. In this case the good consisting of the absence of the suffering and deaths of five would outweigh the torture of the one child. Before we consider whether this would be an example of a compensated horror, we should recognize that in order for such imagined scenarios to be at all relevant to the problem of evil, the example must be such that the good cannot be realized in the absence of the evil. And the relevant problem in the current example is that, while we can easily describe such scenarios, it is very difficult to imagine actual circumstances in which we could not prevent the torturing-to-death of five children except by torturing another child to death. This problem is magnified in in the case of an omnipotent being. In this case the relevant modality is logical necessity; God is only justified in permitting the torturing of the one because of the five others who face torture and death if it is logically impossible to prevent the torture and death of the five without permitting it. And this strains credulity. It is very difficult to imagine a scenario in which torturing a child to death would in any way serve to prevent five other children not being tortured to death, let alone one in which torturing a child to death is logically necessary for saving five. In any scenario in which five people are in danger of being tortured to death, even we weak humans could presumably just endeavor to stop that torture; there is no reason to think we would need to involve any other child. Again, God’s omnipotence just amplifies the problem; surely God, being omnipotent, could just prevent the torture without needing to involve anyone else. There is nothing logically impossible about that.

But there is another problem here that is more relevant to the question of how, if at all, horrors can be compensated. It is this: even though saving the five outweighs the torture of the one, it is not clear that the torture of the one really is compensated since the tortured child herself is not the beneficiary of the compensation. For some evils, it is arguably sufficient that they be compensated in a general sense; there is no requirement that the individuals who suffer the evil be compensated. Suppose we could, by raising taxes on 100 billionaires by .5%, raise enough money to purchase and distribute enough rotavirus vaccine to children who would otherwise not get it to save 5000 children from death due to diarrhea. It is fair to say that the evil consisting of the loss of income for the billionaires would be more than compensated. And it does not matter that the billionaires are not the beneficiaries of the compensation. In the case of horrors, however, this it is less clear that such overall compensation will suffice. It is not unreasonable to think that if we are going to speak of compensation for horrors, then the sufferer of the horror should be benefited by the compensation.

The upshot of this discussion is that speaking thoughtfully about compensation for horrors, and thus speaking thoughtfully about circumstances in which an omnipotent being can justly permit the occurrence of horrors, is far more difficult a thing than it is for lesser evils. I am not trying to give any kind of general account of what would be required, even at a minimum, for horrors to be compensated (I don’t think such an account is possible). I am only drawing our attention to the very great difficulty involved in supposing that horrors can be compensated. With all of these observations in the background, I will now offer the following argument that demonstrates that the inference from the occurrence of apparent horrors to the existence of gratuitous evils is perfectly reasonable.

(1) There are instances of evil that appear to be horrors.

(2) By definition, an omnipotent being can do whatever it is logically possible to do.

(3) Thus, in order for some good, g, to provide a justification for God’s failure to prevent some horror, h, g must not only compensate for h (however that might be achieved), g must also be such that it is logically impossible that it occur without the occurrence of h.

 (4) All of the goods that are known to exist and are (a) such that their value is sufficiently high that pursuing them could justify a person, who is in a position to prevent an instance of evil (horror or otherwise), to fail to prevent that evil are also such that (b) it is logically possible that they be realized without the occurrence of horrors.

Note that this is an empirical claim that is defeasible. I have never encountered any good that could conceivably satisfy (a) but not (b). Nor have I ever come across a purported imaginary example of such a good. But any good that I can imagine that satisfies (a) also satisfies (b).

(5) Thus, there are no known goods that would justify an omnipotent being’s failure to prevent any instance of horror.

(6) No human has ever been able to describe any possible good that would satisfy (a) but not (b).

 (6) does not just claim that we have been unable to discover such goods but that we cannot even describe what they would be like. The problem is that when we imagine goods of significant value such that they might outweigh some instance of horror, it immediately becomes obvious that the realization of such a good does not logically depend on the relevant horror (or any horror, for that matter). So, we have no idea what such a horror-compensating good would be like.

(7) The best explanation for (6) is that no such goods exist.

 (7) does not rule out the possibility of other explanations for (6), but claims only that the best explanation for (6) is that there are no unknown horror-compensating goods. It is possible that our cognitive limitations prevent us from imagining such goods or consistently describing what they might be like. But the problem with this alternative is that we don’t have any reason to believe that our cognitive capacities would be so limited. Hence,

(8) There is no compelling reason to suppose that there are goods that are beyond our ken and satisfy (a) but not (b).

What I mean here is that there is nothing about human beings and our cognitive capacities which would lead us to believe that there are goods (satisfying (a) but not (b)) that are impossible for us to comprehend, experience, or imagine. Further, nothing about the existence of a perfect being suggests that this should be so. The only reason people suppose that there are goods that are beyond our ken at all is that it is a way of responding to the problem of evil. But there is nothing intrinsic to human beings or in the nature of God that suggests that there are such goods. So, in the absence of any good reason to think that our limited cognitive capacities accounts for (6), the best remaining explanation is that no such goods exist. Hence (7) is true, and so,

(9) It is reasonable to extrapolate (4) to all goods; that is, it is most likely that there are no goods (known or unknown) that satisfy (a) but not (b)

Therefore, (10) At least some of the evil that appears gratuitous really is gratuitous.

This argument is clearly inductive, and I want to emphasize a consequence of this that may be unappreciated. If we had compelling reasons to think that God exists, this would give us compelling reason to think that there are no gratuitous evils. And so if a person thinks that she has compelling reason to believe that God exists, this argument will not convince her. But that the argument does not convince a committed theist is not very significant. It is no more significant than is the fact that further investigation could conceivably show that one or more of my premises (5, say) is false. My point has been to show that at least some inferences from the fact that some evil appears to be gratuitous to the conclusion that the evil is gratuitous is not at all problematic. Any reasonably probabilistic inference can be rendered unreasonable in the light of new information. So, that the inferences and claims in my argument are defeasible is hardly interesting. What is important is that, in the absence of further evidence, they render the relevant inference, from the appearance of gratuitousness to the reality, reasonable.