Introduction. It is commonly thought that the evil and suffering in our world constitutes strong evidence for atheism, the thesis that no omnipotent, omniscient and wholly good God exists. But exactly how is it strong evidence? And is it strong enough to make it rational to believe there is no God?

Suppose God and evil are incompatible; then, since there clearly is evil, we have enormously strong evidence for atheism. Very few philosophers today who study our topic would endorse this argument, however. Why? Because it seems that God and evil are, strictly speaking, compatible. We can think of various reasons God might have to permit a fair bit of evil; and to the extent that we cannot think of any reason for God to permit so much, we have no good grounds to think that there could not be a justifying reason we do not know of.

Nevertheless, even if God and evil are compatible, and even if we can see how God might be justified in permitting a good deal of evil and suffering, certain facts about evil and suffering may constitute strong evidence for atheism, so strong that it is rational to be an atheist--provided one has no equally good grounds to think there is a God. This is William Rowe’s thesis in his justly famous essay "The Problem of Evil and Some Varieties of Atheism".

The argument stated. In what follows, let’s suppose--just to see what follows, or perhaps because we really believe it’s true--that there are no equally good grounds for theism. With this supposition in place, let’s turn to our title question: does evil make it rational to be an atheist? Rowe answers ‘yes’, and he gives a powerful argument for it. In its official form, it can be put as follows:

For the purposes of this paper, let’s grant the second premise, although I don’t mean to give the impression that all is clear sailing here. [1] What about the first? It is equivalent to a claim that something of a certain sort does not exist, specifically this claim:

Rowe’s argument for premise 1. Well, what should we make of premise 1? Rowe offers one particular instance of intense suffering that God could have prevented without thereby losing some greater good or permitting some equally bad or worse evil—the case of the fawn who is horribly burned in a forest fire, lying in agony for days before she dies, doubtlessly one of millions of such cases throughout history. In other essays, he points to other occurrences of especially horrible suffering. So Rowe initially supports premise 1 by pointing to certain cases of suffering, saying of each one that there is no greater good that would have been lost nor any equally bad or worse evil that would have been permitted if God had prevented it, that very occurrence of suffering. In effect, then, he draws up a list that begins like this:

An important question arises: are any of them true? Rowe briefly sketches his answer to this question:

Here we have only considered the cases of Ashley Jones and the fawn. The other cases on our list must likewise be examined. Suppose we are able to say of each one what Rowe said of the fawn and what I suggested we should say about Ashley Jones: "So far as we can see, there is no greater good that would have been lost nor any other equally bad or worse evil that would have been permitted if God had prevented it." In that case, we can draw up a second list that exactly parallels our first list. It would start like this:

Now we’re in a position to sum up our characterization of Rowe’s initial defense of premise 1. It goes like this: none of the propositions on Our Noseeum List are false. [3] At least some of them make it highly likely that the corresponding propositions on Our List of Pointless Evils are true. If this last claim is correct, then premise 1 is highly likely to be true, and thus it is "altogether reasonable to believe" it. But do any of the items on Our Noseeum List make it highly likely that any of the corresponding items on Our List of Pointless Evils are true? More generally, is it really reasonable to infer that there is no greater good that would be lost on the grounds that, so far as we can tell, there is none? These are the questions that we must now answer. To get at them, let’s pause to reflect on what kind of inference is involved here.

Rowe-style noseeum inferences. Suppose that, after rummaging around carefully in my fridge, I can’t find a carton of milk. Naturally enough, I infer that there isn’t one there. Or suppose that, on viewing a chess match between two novices, Kasparov says to himself, "So far as I can tell, there is no way for John to get out of check," and then infers that there is no way. These are what we might call no-see-um inferences: we don’t see ‘um, so they ain’t there! [4]

Notice four things about noseeum inferences. First, they have this basic shape: "So far as we can tell, there is no x; so, there is no x." Second, note that in each of the cases just mentioned, it is possible for the conclusion to be false even if the premise is true. Even though I rummaged through the fridge carefully and my vision is in tip-top shape, I could just simply miss the carton of milk. And even Kasparov can have an off day. Nevertheless—and this is the third point—in each case the argument is a strong one. Under certain conditions (about which I will say more shortly), our inability to see something makes it highly likely that there isn’t anything of the sort we failed to see. Finally, it won’t do to object to any particular noseeum inference that even if the premise is true the conclusion might be false. For every noseeum inference—even the strongest of them—is like that. When evaluating a particular noseeum inference, we can’t just write it off with a casual, "Ah! But there might be an x we don’t know of even if so far as we can tell there isn’t one." That’s true, but it’s irrelevant to the strength of the inference.

Now, in effect, Rowe bids us to use at least some of the items on Our Noseeum List as a basis for believing some of the items on Our List of Pointless Evils, and this amounts to deploying some, perhaps several, noseeum inferences, like the inference from 1a1 to 1a:

Obviously enough, many noseeum inferences are reasonable, like the ones mentioned at the beginning of this section. And just as obviously, many are not. For example, looking at my distant garden from my kitchen window, the fact that so far as I can tell, there are no slugs there hardly makes it likely that there are none. Likewise, a beginner viewing a chess match between Kasparov and Deep Blue would be ill-advised to reason: "I can’t see any way for Deep Blue to get out of check; so, there is none." Or imagine us listening to the best physicists in the world discussing the mathematics used to describe quantum phenomena or the theory of general relativity. Presumably it would be unreasonable for us to infer that, since we can’t comprehend or grasp what they are saying, there is nothing there to be grasped. The crucial question, then, is this: what distinguishes the reasonable noseeum inferences from the lousy ones?

Consider the cases already sketched. Notice that it is quite likely that I would see a milk jug in the fridge if one were there, and it is very likely that Kasparov would see a way out of check if there were one. That’s because Kasparov and I have what it takes to discern the sorts of things in question. On the other hand, it is not very likely that I would see a slug in my garden even if there were one there, at least not from my kitchen window. Nor is it very likely that a beginner would be able to see a way out of check for Deep Blue even if there were one since strategy at the grandmaster level can be very complex. And the same goes for our comprehending exceedingly complex mathematics: even if what the physicists were talking about did make sense, it isn’t very likely that we would be able to understand it.

We can distill these reflections in the following principle, which marks an important difference between reasonable and unreasonable noseeum inferences:

• A noseeum inference is reasonable only if it is reasonable to believe that we would very likely see (grasp, comprehend, understand) the item in question if it existed.

• The move from "So far as we can tell, there is no greater good..." to it is very likely that "There is no such greater good" is reasonable only if it is reasonable to believe that we would very likely see or comprehend a greater good, if there were one.

Now we are in a position to raise a very important question: Is it reasonable to believe Rowe’s Noseeum Assumption? Several arguments for Rowe’s Noseeum Assumption have been given in the literature, and I have explained elsewhere why I think that those arguments fail. [6] Here I will simply offer reasons to think that it is not reasonable to believe Rowe’s Noseeum Assumption.

Two strategies for assessing Rowe’s noseeum assumption. I begin by distinguishing two strategies for assessing whether it is reasonable to believe Rowe’s Noseeum Assmption. We can get at them by way of analogy, comparing two questions.

First: is it highly likely that I would see a slug in my garden from the kitchen window if one were there? Not at all. I know that slugs are relatively small and I know that the unaided human eye is not suited to see such small things at a hundred feet; moreover, my garden is over an acre large and, per usual, it’s overgrown. So we have superb reason to think that it is false that I would very likely see a slug in my garden even if one were there. Now, another question: is it highly likely that extra-terrestrial life forms would contact us if they existed? The only answer suitable here is "How should I know?" If there were extra-terrestrial life forms, how likely is it that some of them would be intelligent enough to consider contact? And of those intelligent enough, how many would care about it? And of those with the smarts and the desire, how likely is it that they would have the means at their disposal to try? And of those with the means, how likely is it that they would succeed? I haven’t the foggiest idea how to answer any of these questions. I can’t even begin to say with even the most minimal degree of confidence that the likelihood is low or middling or high. I just don’t have enough to go on. In that case, I should be in doubt about how likely it is that extra-terrestrial life forms would contact us if they existed. I should be of two minds, neither for it, nor against it. I should just shrug my shoulders and say "I don’t know. I’m in the dark on that score."

There are two points to see here. First, in each case it is not reasonable to believe that the proposition in question is highly likely to be true, although for different reasons. In the first case, it is not reasonable to believe it is highly likely that I would see a slug in the garden even if there were one there because it is reasonable to believe that the proposition is positively false—indeed,because the garden is large and overgrown and I am viewing it from quite a distance, I have good reason to believe it is very, very likely that I would not see a slug. In the second case, however, it is not reasonable to believe that the proposition in question is false. Rather, for the reasons mentioned, we have good reason to be in doubt about how likely it is that we would have been contacted by extra-terrestrials if there were any. Indeed, we don’t even have enough to go on to make a rough guess. As a consequence—and here is the second, absolutely crucial point—having good reason to be in doubt about the matter is good enough reason all by itself to think that it is not reasonable to believe that we would probably have been contacted. For how could it be reasonable for us to believe something about which we have good reason to think we are utterly in the dark?

Now let’s apply these points to Rowe’s Noseeum Assumption. To assess the reasonableness of believing that we would very likely see a greater good if there were one, we might consider whether Rowe’s Noseeum Assumption is false. In that case, we might try to think of reasons to believe that it is very likely that we would not see a greater good. We would then be treating Rowe’s Noseeum Assumption as I treated the proposition that it is very likely that I would see a slug in my garden from the kitchen window. On the other hand, we might consider whether we should be in doubt about whether we would very likely see a greater good. This would be to treat Rowe’s Noseeum Assumption as I treated the proposition that it is highly likely that extra-terrestrials would contact us if there were any. The crucial point to understand here is that even if all we have is good reason to be in doubt about whether it is highly likely that we would see a greater good, that is enough reason to deny that it is reasonable to believe Rowe’s Noseeum Assumption.

Reasons to doubt Rowe’s noseeum assumption. I suspect that it is not reasonable to believe Rowe’s Noseeum Assumption. In what follows, I focus on reasons to be in doubt about it rather than reasons to think it is false. Why? Because the only reasons I can think of for believing that it is false presuppose that God has informed us that we should expect to be unable to discern His purposes in permitting particular horrors, which, of course, presupposes that there is a God. I don’t want to presuppose that here. I want to show that even a nonbeliever can have good reason to refrain from believing Rowe’s Noseem Assumption. So, what considerations might put us in doubt about whether it is reasonable to believe Rowe’s Noseeum Assumption? Several have emerged in the literature. I have space for three.

Alston’s analogies. The first consideration targets Rowe-style Noseeum Inferences directly. It says that such inferences involve two aspects which should make us wary of our ability to tell whether we would very likely see a greater good or a God-justifying reason if there were any.

First, a Rowe-style Noseeum Inference takes "the insights attainable by finite, fallible human beings as an adequate indication of what is available in the way of reasons to an omniscient, omnipotent being." But this is like supposing that when I am confronted with the activity or productions of a master in a field in which I have little expertise, it is reasonable for me to draw inferences about the quality of her work just because I "don’t get it". I’ve taken a year of university physics. I’m faced with some theory about quantum phenomena and I can’t make heads or tails of it. Certainly it is unreasonable for me to suppose it’s likely that I’d be able to make sense of it. Similarly for other areas of expertise: painting, architectural design, chess, music, and so on.

Second, a Rowe-style Noseeum Inference "involves trying to determine whether there is a so-and-so in a territory the extent and composition of which is largely unknown to us." It is like someone who is culturally and geographically isolated supposing that if there were something on earth beyond their forest, they’d likely discern it. It is like a physicist supposing that if there were something beyond the temporal bounds of the universe, we’d probably know about it (where those bounds are the big bang and the final crunch).

All these analogies point in the same direction: we should be in doubt about whether we would very likely discern greater goods that would justify God’s permission of particular horrors even if one were there. [7]

The progress argument. Knowledge has progressed in a variety of fields of enquiry, especially the physical sciences. The periodic discovery of previously unknown aspects of reality strongly suggests that there will be further progress of a similar sort. Since, future progress implies present ignorance, it is very likely that there is much we are now ignorant of. Now, what we have to go on in charting the progress of the discovery of intrinsic goods by our ancestors is meager to say the least. Indeed, given the scant archeological evidence we have, and given paleontological evidence regarding the evolutionary development of the brain in homo sapiens, it would not be surprising at all that humans discovered various intrinsic goods over tens of thousands of years dotted by several millenia-long gaps in which nothing was discovered. Hence, given what we have to go on, it would not be surprising if there has been the sort of periodic progress that strongly suggests that there remain goods to be discovered. Thus it would not be surprising if there are goods of which we are ignorant, goods of which God—in His omniscience—would not be ignorant.

The argument from complexity. One thing Mozart’s Violin Concerto No. 4, Ste. Michele’s Cabernet Sauvignon (Reserve), and the best sorts of love have in common when compared to Chopsticks, cheap Gallo, and puppy love is that each illustrates the fact that the goodness of a state of affairs is sometimes greater, in part, because it is more complex. Now, since intense, undeserved suffering and horrific wickedness are so bad, it would take correspondingly greater goods to justify God’s permitting such horrors. Hence, it would not be surprising if the greater goods involved in God’s purposes possess a degree of complexity well beyond our grasp. It follows that it would not be surprising if there were greater goods outside our ken.

Of course, while complexity does not always adversely affect our ability to recognize value, it can and sometimes does. To defend this claim, I cannot show you a complex state of affairs whose value we fully recognize but whose complexity hinders such recognition. I must resort to more general considerations.

First, there is the general phenomenon of the complexity of something hindering our view of some important feature it has, e.g. the complexity of an argument hindering our ability to discern its validity, or the complexity of your opponent’s strategy hindering your ability to discern that unless you move your knight to queen’s side bishop 5, her next move is checkmate. But, more to the point, why can a child discern the literary merits of a comic book but not Henry V or The Brothers Karamazov? Why can a child clearly discern the aesthetic value of a toffee but have a difficult time with alder-smoked Copper River sockeye served with pesto, lightly buttered asparagus al dente, fresh greens in a ginger vinegarette, and chilled chardonnay? Why can a child recognize the value of his friendship with his buddy next door but not the full value of his parents’ mutual love? Surely because great works of literature, fine cuisine and adult love at its best involves much more than he is able to comprehend. And this is true of adults as well, as reflection on our progress in understanding the complexity of various things of value reveals. For example, periodically reflecting on the fabric of our relationships with those whom we love most and whose love we most cherish, we might well find strands and shades that when brought to full light permit us to see those love-relationships as more valuable than we had once thought. If the failure to grasp the more complicated aspects of our relationships can prevent a full appreciation of love’s value, surely the failure to grasp the complexity of a state of affairs might well hinder us from discerning its goodness. Value is often veiled in complexity.

The three considerations presented here—Alston’s Analogies,the Progress Argument, and the Argument from Complexity—together constitute good reasons to doubt whether it is highly likely that we would see a greater good if there were one. More specifically, they cumulatively constitute good reason to be in doubt about each of the instances of Rowe’s Noseeum Assumption that we will have to believe reasonably if we are to move from any of the propositions on Our Noseeum List to any of the propositions on Our List of Pointless Evils. For example, here is one of the noseeum inferences that Rowe bids us to make in connecting Our Noseeum List to the corresponding items on Our List of Pointless Evils.

So what? Three paragraphs before stating his argument, Rowe asks this question: "Is there an argument for atheism based on the existence of evil that may rationally justify someone in being an atheist?" He says the answer is "yes" and gives the argument we’ve been examining. As you might expect, Rowe’s argument can rationally justify us in being atheists only if we are rationally justified in believing all of its premises. Unfortunately, his initial defense of premise 1 does not rationally justify us in believing premise 1. So if that’s all we have to go on, we are not rationally justified in being atheists on the basis of Rowe’s argument.

But what if some other defense of premise 1 works? [8] I suspect that any defense will not avoid the objections raised above. If all we have to go on is Rowe’s initial defense, then we are not rationally justified in believing all of the premises of Rowe’s argument; consequently, we are not rationally justified in being atheists on the basis of Rowe’s argument. That strikes me as a point worth understanding, a point that fits snugly with our ignorance of whether there are any pointless evils. If the game is Let’s Make An Argument That Rationally Justifies Us In Being Atheists, then doesn’t this point seem like a winner? [9]

[1] Michael Peterson and William Hasker deny premise 2 in Evil and the Christian God (Baker, 1982) and "The Necessity of Gratuitous Evil," Faith and Philosophy (1992), respectively. To see why their arguments fail, see my "Is Theism Compatible with Gratuitous Evil?," American Philosophical Quarterly (1999), co-authored with Frances Howard-Snyder. Peter van Inwagen, on the other hand, has sketched a fascinating objection in "The Magnitude, Duration, and Distribution of Evil: A Theodicy," Philosophical Topics 16 (1988), 167-68, and "The Problem of Evil, the Problem of Air, and the Problem of Silence," Philosophical Perspectives 5 (1991), 164, note 11. Both are collected in God, Knowledge, and Mystery (Ithaca: Cornell University Press, 1995). In God, Knowledge, and Mystery, pages 15-16 are particularly instructive. Also see "Reflections on the Chapters by Draper, Russell, and Gale," in ed. Daniel Howard-Snyder, The Evidential Argument from Evil (Indiana University Press, 1996), 234-35. Frances and I assess the merits of van Inwagen’s objection in "Vagueness and the Compatibility of God and Gratuitous Evil" (forthcoming).

[2] It is a useful exercise to come up with your own list of specific horrors that you think God would able to prevent without thereby losing a greater good or permitting an equally bad or worse evil.

[3] Richard Swinburne would challenge this claim. See Providence and the Problem of Evil (Oxford, 1998).

[4] The noseeum lingo is Stephen Wykstra’s. See "Rowe’s Noseeum Arguments from Evil," in The Evidential Argument from Evil, 126-50.

[5] I omit the phrase "nor any other equally bad or worse evil that would been permitted" here and in what follows since it is not germane to my argument.

[6] "The Argument from Inscrutable Evil," in The Evidential Argument from Evil, and "The Argument from Divine Hiddenness," Canadian Journal of Philosophy (1996).

[7] William Alston, "Some (Temporarily) Final Thoughts on Evidential Arguments from Evil," in The Evidential Argument from Evil, 316-19.

[8] Rowe initially defends premise 1 by identifying some particular horrible instances of suffering and asserting of at least some of them that they are pointless. In the penultimate paragraph of section I of his essay, he offers a different argument. Suppose it really is reasonable to believe that some greater good would have been lost if God had prevented the fawn’s suffering, he says; indeed, we might suppose the same thing about the other horrors on our List of Pointless Evils. Of course, the particular horrors that we’ve identified are but a mere drop in the vast ocean of all the evil there is. And here an important point emerges: even if we cannot reasonably say of any particular horror on our List that it is pointless, it is reasonable to believe that no greater good would require God to permit so much evil—the vast quantity of horrific suffering and misery and wickedness that has stuffed terrestrial history and shows every sign of continuing unabated. Surely a being who is perfect in power and wisdom would not have to permit all of this enormous variety and profusion of intense human and animal suffering in order to achieve His purposes. Surely He could have prevented some of it without losing any greater good, just some. The idea that none of it could have been prevented by God without thereby losing some greater good seems an extraordinarily absurd idea, quite beyond belief. In that case, at least some instances of evil or other are pointless, even if the instances on our List are not pointless and even if we can’t say which instances are pointless.

The main thing to see about this second defense of premise 1 is that all of the worries about he initial defense arise for it as well. First, why suppose it is reasonable to believe that there is no greater good that wold require God permit so much horrible suffering in the world? Presumably because, so far as we can tell, there is no greater good that would require God to permit so much. So we have a new version of Rowe’s Noseeum Assumption: we would very likely be able to see a greater good that would require God to permit so much horrible suffering, if there were one. Isn’t it pretty clear that Alston’s Analogies, the Progress Argument, and the Argument from Complexity jointly constitute just as good a reason to be in doubt about this new version of Rowe’s Noseeum Assumption as they do its counterparts about particular horrors? If so, then the new defense of premise 1 is no better than the initial one. I leave it as homework to the reader to try to come up with a more promising option.

[9] For useful reminders on an earlier draft of the chapter, I thank William Rowe. I am indebted to William Alston and, especially, Stephen Wykstra for the main lines of thought I develop.

1. What sort of initial plausibility would you apply to Rowe’s No-Seeum inferences?

2. Consider Howard-Snyder’s counter-examples to Rowe’s No-Seeum inferences. Are they sufficiently powerful to overcome the initial plausibility of Rowe’s no-seeum inferences?

3. If Howard-Snyder’s arguments are sound, what rational options are open with respect to belief in God? Are any closed off?

4. Can you come up with a defense of premise 1 that avoids Howard-Snyder’s worries?