Book Review: David O’Connor, God and Inscrutable Evil: In Defense of Theism and Atheism, Lanham/London, Rowman & Littlefield, 1998, pp. xiii + 273, $28.95 paper, $78.50 cloth.
(To purchase the book, click on the image of the book cover to the right.)
“There is a great deal of evil in the world, much of it seemingly pointless” (p.1). Does this fact provide good evidence against the existence of God? O’Connor’s important book explores this question. God and Inscrutable Evil is frequently technical–probably too technical for the casual reader–but the thoroughness of argument and O’Connor’s interaction with the latest work on the problem of evil make it required reading for anyone with at least some background in philosophy and an interest in the issues explored.
In part I O’Connor develops a two-stage argument from evil and deploys it against “a central, arguably the central, version of theism,” which he calls “orthodox theism [OT]” (p.7); OT is the claim that God–“understood as the omnipotent, omniscient, perfectly good designer, creator, and sustainer of the universe”–exists, has “a divine plan for human beings and their world,” and so on (pp.7-8). The first stage of the argument comprises a “reformed logical argument” from natural evil (p.7, my emphasis). The reformed argument claims that certain facts of natural evil–in particular, the existence of (large amounts of) natural evil resulting entirely from natural processes (NERNP)–are logically inconsistent with the existence of God. (O’Connor focuses on natural evil–diseases, earthquakes and the like–because God bears responsibility for it; moral evil, resulting from the free choices of moral agents, is less of a problem for OT.) The argument is ‘logical’ because it “aims to demonstrate that theism could not possibly be true [in the logical sense of possibility], given certain facts of evil” (p.5). The argument is ‘reformed’ because, unlike other logical arguments from evil, it does not claim to prove God’s existence is logically inconsistent the facts of evil in question; merely that we have reason to think they are logically inconsistent (p.10). O’Connor argues, futhermore, that OT’s presumptively best defensive arguments–“Richard Swinburne’s greater-good defense, Alvin Plantinga’s free-will defense, and George N. Schlesinger’s no-best-possible-world-defense” (p.11)–fail to show that God’s existence is consistent with those facts of evil (p.7).
The second stage, an “indirect empirical argument” from evil, claims that OT’s failure in this regard “significantly undermines [OT’s] epistemic standing” and “gives us good, although not compelling, reason to regard [OT] as either unjustified or false” (p.10). What makes the argument “empirical” is precisely that it “aims to show that, given certain facts of evil, there is good…reason to think theism is either false or unjustified” (p.5). What makes the argument “indirect” is that its “evidential base,” comprising “facts about [the failure of certain defensive] arguments and not facts about evil, is at a remove from the phenomena of evil themselves” (p.179). The conclusion of the indirect argument is that, so far as the facts of evil in the world are concerned, “atheists are justified in [remaining atheists], while theists and agnostics have good reason to move a significant amount in the direction of atheism,” so that “belief in God is intellectually troubled to a significant degree” (p.178).
This is the defense of atheism referred to in the subtitle. But it comes with a caveat: the “two-stage argument in part I, comprising the reformed logical argument and the indirect empirical argument issuing from it” (p.41), is developed under the “standard model of philosophical debate” on the problem of evil (p.39). The standard model assumes: (1) that humans are in a position to judge whether the actual world is or is not the sort of world God would be likely to make, so that (2) we can in principle reach a verdict as to whether the facts of evil support atheism or theism (p.40). As both these assumptions have been challenged (an issue discussed in part II), “the conclusion of the indirect argument is interim and provisional” (p.180).
In the shorter part II, O’Connor moves to discussion of direct empirical arguments from evil (namely, those whose evidential base comprises certain facts of evil), and in particular the argument formulated by William Rowe. O’Connor then considers the skeptical defense of theism advocated by Stephen Wykstra and others, and concedes that this defense not only succeeds to a large degree against Rowe’s argument, but also refutes (or at least justifiably departs from) the assumptions of the standard model of debate on the problem of evil, and thus undermines the indirect empirical argument of part I as well. This is the defense of theism referred to in the subtitle. The facts of evil, O’Connor says, constitute sustaining evidence for atheism (p.211), in the sense that someone who is already an atheist will regard those facts as further reason to remain an atheist (since those facts are just what we would expect if atheism were true); but those facts do not settle or even tend to settle the debate in favour of atheism, since, as the skeptical defence shows, the facts of evil are equally what we would expect if theism were true (or at least are not particularly surprising given theism).
O’Connor thus argues, in the end, for a “detente” between “friendly theism and friendly atheism”–the term “friendly” denoting “each side’s recognition of failure to either refute the other [side] or to gain decisive cognitive advantage over it” (p.227). “[T]heism,” he says, can be justified for certain persons in certain circumstances, atheism for others in other circumstances” (p.xi); thus the need for an “intellectually tolerant, live-and-let-live view” on the issue of God’s existence (p.236). On the other hand, atheism at least has a slight edge in the debate over evil. Once a skeptical defense of theism is adopted–as it must be if OT is to avoid strong disconfirmation–then theism can no longer be classed as an empirical theory, which undermines the entire project of natural theology (pp.219-24). Thus “the cost of the skeptical defense is high, possibly too high for OT to pay” (p.231). Theists must either accept this price or else retreat to the standard model of debate where OT faces serious difficulties.
God and Inscrutable Evil thus presents a subtle, complex and in many ways attractive thesis which can only be fully appreciated by reading the book. O’Connor’s arguments are often persuasive, although (as I will now explain) in some cases they need bolstering and in others he concedes too much territory to the theist. A word of warning: in what follows, I will frequently use acronyms. If this is annoying, so is O’Connor’s book, which is full of them: OT, NENP, NENPi, NERNP, NE~RNP, NEM, GGNE, OGNE, etc. Acronyms allow O’Connor to avoid convolution and repetition, but it does become difficult to keep track of what they all stand for. (Readers of the book may find it useful to write them down for ready reference.)
Chapter 3, the first substantive chapter, considers the relation between theism and gratuitous natural evil, that is, ‘[natural] evil that could be avoided or prevented without thereby forfeiting any greater or outweighing good’ (p.52). The standard view that gratuitous evil (natural or otherwise) is logically incompatible with the existence of God has been challenged by William Hasker and Peter van Inwagen. Hasker argues that if there were no genuinely gratuitous natural evil (GGNE), and we really believed this, then our motivation (say) to help those in pain–thereby instantiating goods such as courage, compassion and the like–would be drastically curtailed. After all, if Socrates’ pain is not GGNE–if there is some greater good this suffering will achieve–then why intervene to stop that suffering? But, says Hasker, one of God’s desiderata in world-making is the instantiation of precisely these sorts of goods–courage, compassion, etc.; hence God must allow GGNE for these goods to occur. Hence God and GGNE can coexist. To this O’Connor replies–rightly, I think–that so long as there is OGNE (ostensibly gratuitous natural evil–natural evil that appears to be GGNE), our motivation to instantiate goods such as courage and compassion need not be curtailed, even if we believe (or know) there is no GGNE. The reason is that we can be equally strongly motivated to instantiate those goods by the knowledge that God wishes such goods to be instantiated, that instantiating those goods prima facie makes the world a better place, and so on (pp.53-63).
van Inwagen argues that there is no minimum level of evil required for God’s plan to succeed: “One might as well suppose that if God’s purposes require an impressively tall prophet to appear at a certain place and time, there is a minimum height such a prophet could have” (quoted on p.73). So, by analogy, for any amount of evil you choose (call it n units of evil), God’s purposes could have been equally well served with slightly less evil (say n-1 units of evil). But then there will always be some evil (at least 1 unit) that could have been avoided or prevented without losing any greater or outweighing good (i.e., without any reduction in the extent to which God’s plan is fulfilled). So, whatever God does, there will always be some gratuitous evil: this is something God cannot avoid. So God and gratuitous evil can coexist (p.73). Hence one cannot blame God for not getting by with less evil, “any more than one can argue that a law that fines motorists $25.00 for illegal parking is unjust or cruel owing to the fact that a fine of $24.99 would have an identical deterrent effect” (quoted on p.93). van Inwagen does concede, however, that if God could have got by with much less evil, he surely would have; hence God’s existence, while compatible with some gratuitous evil, is only compatible with a certain (unspecified) level of it (p.74).
O’Connor accepts this line of argument, and thus sets himself the task of showing (what is then required to disconfirm God’s existence) that there is reason to think there is a great deal of gratuitous natural evil in the world, so that there is at least some natural evil that falls above the unspecified level (the “van Inwagen line”) and hence is inconsistent with OT. But this concession is unnecessary. To see this, consider two possible worlds P1 and P2 where God (in accordance with his purposes) creates an impressively tall prophet. In P1, let us say, the prophet is slightly taller. Would God’s purposes, as van Inwagen claims, be equally well served in both worlds? Well, no, because the prophet in P2 is less impressively tall: God’s purposes in P2 are less well served. The analogy therefore fails: it does not point to a case where God’s purposes would be equally well served by a lesser degree of tallness (or, by analogy, a lesser degree of evil). (I owe this point to Jim Stone.) What then of the parking fine analogy? It does show, I think, that if one does not know the minimum level of parking fines (or, by analogy, evil) required for the desired effect then one may nonculpably allow fines (or evil) somewhat in excess of that level. But ‘God does not know the minimum level of evil required to fulfil his purposes’ is true only if there is no such level (for if there were such a level God would know it). Hence the analogy gets God off the hook for failing to get by with less evil only if it has already been established that there is no minimum level of evil required to fulfil God’s purposes (where ‘fulfil’ means ‘fulfil to a particular degree’). As this has not been established, the analogy is beside the point. So van Inwagen has not shown that God and gratuitous evil can coexist.
Chapter 4 presents the core of the reformed logical argument. O’Connor distinguishes between moral evils, that is, “the morally culpable intentions, actions, or omissions of sufficiently mature, free, moral agents” (e.g., Smith’s murdering Jones), and natural evils, those “for which human persons are not [morally] responsible” (e.g., “the painfulness of pain,” “diseases…[and] birth defects for which no human is culpable,” “natural disasters,” etc.) (p.16). He grants that any God-created world will contain some natural evil, namely, NEM–natural evil necessary for moral development (p.82). But, he says, there is a possible world Wp which is exactly like the actual world except that it does not contain NERNP; that is, “nobody (or any animal either) ever experiences pain or suffering due only to natural processes” (p.86). “Wp,” O’Connor explains, “would contain NEM even though it does not contain NERNP…: in Wp, NEM is provided by moral evil or NE~RNP [natural evil not resulting solely from natural processes], or both” (p.89). Thus the divine plan (human moral development and so on) would be accomplished in Wp, but with much less NERNP than there is in the actual world. So all (or a great deal of) NERNP in the actual world is gratuitous; that is, much GGNE exists. But if much GGNE exists then God does not exist. So God does not exist (pp.77-91).
One difficulty with this argument concerns the distinction between NERNP and NE~RNP. O’Connor stipulates that NERNP comprises those natural evils that do not result from any act of human agency, whereas NE~RNP comprises natural evils such that “morally nonculpable agency is a crucial part of their causal history” (p.89, my emphasis). But what does it mean to say that X ‘results’ from Y, or that X is ‘a crucial part of the causal history’ of Y? Perhaps ‘results’ expresses counterfactual dependence: X results from Y if and only if the counterfactual ‘if Y had not occurred, X would not have occurred’ is true; and perhaps ‘a crucial part of the causal history’ expresses something similar. But this creates problems. Many Christian theists maintain that if the Fall had not occurred then all the natural evil we see around us would also not have occurred. Such theists, then, would hold that all natural evils ‘resulted’ from an act of agency–the Fall–so that strictly speaking there is no NERNP. The reformed argument’s assumption that NERNP exists would then be question-begging. It is plausible, moreover, that every natural evil suffered by humans is counterfactually dependent on, and in this sense ‘results’ from, an act of agency: if one’s parents (or grandparents, or great-grandparents, or…) had not decided to have intercourse–an act of agency–then one would never have existed, in which case any natural evils one suffers would not have existed either. It follows that if ‘results’ expresses mere counterfactual dependence, there is no NERNP suffered by humans–in which case the reformed logical argument fails (since it claims there is NERNP suffered by humans). The lack of precision in O’Connor’s analysis of NERNP thus leads to difficulties. We need to be told what ‘results’ expresses, if not mere counterfactual dependence.
Perhaps the answer is to stipulate that NERNP comprises those evils–of which there are clearly very many–such that human agency is only a very small part of their causal history; of course, we would then have to say that NERNP means natural evils resulting almost solely from natural processes. One difficulty with this move is that ‘very small’ is vague, meaning that in some cases it will be indeterminate whether a given instance of evil is NERNP or not. Vagueness is not always a problem, but we would not, I think, want it creeping into our description of Wp. To get around this, we could stipulate that Wp contains no evil that is very clearly NERNP. The reformed argument can then (I think) proceed.
Another problem with the reformed argument is that O’Connor provides no more than a cursory description of Wp (pp.85-88). He tells us only that Wp is like the actual world except that it contains no NERNP, and that the absence of NERNP from Wp is effected by a “circuit-breaker” mechanism built into natural processes so that they do not (without human action) result in natural evil (p.86). This is not a very detailed picture. And, surely, the less detailed our picture of Wp, the less certain we can be that Wp would be better than the actual world. Hence O’Connor’s failure to provide a detailed description of Wp significantly reduces the force of the reformed argument. In addition, the following questions might be asked of Wp:
(1) Does the absence of NERNP from Wp mean natural disasters simply do not occur in Wp, or do they perhaps occur without people suffering? If the former, would not Wp therefore have to have substantially different causal laws to the actual world? And would this not mean we must be somewhat uncertain as to what Wp would look like? If the latter, would people still die prematurely (albeit painlessly) in natural disasters? Surely not: as premature death is an evil, premature death due solely to natural disasters is NERNP–which by hypothesis is absent from Wp. Perhaps, then, people in Wp do not die prematurely in natural disasters, but are miraculously saved–e.g., the law of gravity is conveniently suspended while I get out of the way of a falling building. But would this not produce many irregularities in nature–irregularities which, besides producing considerable confusion, could also seriously undermine the scientific enterprise (since scientists seek, in general, to discover (near-)exceptionless natural laws)? If so, isn’t it understanable that God might create the actual world rather than Wp?
(2) To reinforce this point, suppose you encounter a forest fire in Wp. If it was caused by lightning, you do not get burnt, or at any rate do not suffer (for this would be NERNP, and Wp contains no NERNP). If it was lit by humans, you do get burnt (for this is NE~RNP, which can occur in Wp). Similarly for other potentially dangerous situations. But, as forest fires look the same either way, you (or an animal in the same situation) cannot know whether to run or stay put. The actual world, by contrast, is far more regular: when there is a fire, you know you will get burnt regardless of how it was caused. Is the certainty produced by this greater regularity not a good thing for humans and animals alike? If so, doesn’t the actual world (to this extent at least) look like a better choice than Wp?
I do not claim such questions are unanswerable. But O’Connor’s failure even to address them constitutes a significant weakness in his treatment of the reformed logical argument. Until such questions are answered, the claim that Wp is a better world than the actual world lacks support, so that theists may rationally decline to accept a key premise in the argument.
O’Connor’s next task is to show that theism’s presumptively best defences fail to block or defeat the reformed logical argument. In Chapter 5 he considers Swinburne’s greater good defense. Swinburne argues (i) that NERNP is necessary for practical knowledge of good and evil (i.e., knowledge of the morally relevant consequences of our actions, such as causing pain: p.99), hence (ii) that moral development is impossible without NERNP, hence (iii) that, pace the reformed argument, God’s purposes (which include moral development) would not be as well served in a world such as Wp which contains no NERNP (p.104). O’Connor replies that, on the contrary, Wp (like the actual world) contains a great deal of moral evil and NE~RNP from which we can gain practical knowledge of good and evil (i.e., Wp contains an abundance of NEM); and there is no reason to think the evil in Wp would be insufficient for a degree of practical knowledge of good and evil sufficient to satisfy God’s purposes (pp.106-8).
Chapter 6 deals with Plantinga’s extension of the free will defense to cover natural evil. To show the natural evil we see around us is consistent with OT, Plantinga proposes, as a mere logical possibility, the “Satan hypothesis”. On this hypothesis, God creates a supernatural person, Satan, who has the power to become “GS, good Satan” (a Satan who remains well behaved) or “ES, evil Satan” (a Satan who falls and causes all the natural evil we see around us) (p.120). As Satan has libertarian free will, God cannot control whether Satan becomes ES or GS. As it happens, Satan becomes ES, and the ‘natural’ evil we see around us (which is really a species of moral evil) is the result. O’Connor accepts this as a logical possibility, but claims it is inconsistent with OT. The reason is that there could not be “a God-justifying reason for actualizing a world containing free human beings, in which the Satan hypothesis is true” (p.117). Why not? Because, in preference to creating a world containing Satan and his cohorts, God could have created a world without them, such as Wp.
To see that God should have created Wp, O’Connor says, first suppose God does not have middle knowledge. Then, nevertheless, God at least has substantial knowledge of the psychological traits of each possible person. God would therefore know, prior to creating Satan, that Satan has the capacity “to cause and enjoy causing pain and suffering on a huge scale over the whole of history [i.e., to become ES]” (p.126). (That God would know this is, unfortunately, little more than assertion on O’Connor’s part; but perhaps he can point out that one surely does not cause, and enjoy causing, untold amounts of suffering over the course of millennia unless one has a fairly strong tendency towards extreme evil: such a tendency must be a significant part of Satan’s character, hence would indeed be known to God. As O’Connor maintains, Satan had the choice to become GS or ES, but it must have been quite likely that he would become ES.) Since other possible beings would have a lesser tendency towards, and lesser capacity for, evil, God would have created those other beings in preference to Satan (pp.126-8). On the other hand, if God does have middle knowledge (and foreknowledge), his creation of a world containing Satan and his cohorts, in full knowledge of the massive evil that would result, amounts to a culpable failure to actualise a world with less evil, namely, Wp (pp.128-37). Either way then, Plantinga’s defense fails: it provides no excuse for God to create the actual world rather than Wp.
Chapter 7 considers Schlesinger’s “No Best Possible World” defense. Schlesinger argues, in effect, that as there is no best possible world, any world God creates is such that he could have created a better one; hence, we cannot criticise God for failing to have created a better world (such as Wp), since God cannot avoid failing to have created a better world (p.152). Against this defense, O’Connor presents three persuasive objections (pp.161-73). The most important is that “even when nonculpably I do not do my best, I do not have moral carte blanche to do, or not do, anything at all” (p.167). Thus, while God cannot be faulted for creating a world with a surpassable number of good features (for this is something God cannot avoid), he can be faulted for creating a world with so many bad features–in particular, so much NERNP. God may not be culpable for failing to do his best (for there is no ‘best’), but he is culpable for failing to do less badly than he did. So he can be culpable for failing to create the better (or rather, less bad) world Wp.
With these components in place, Chapter 8 formulates the indirect empirical argument from evil already mentioned: on the standard model, the failure of OT’s presumptively best defenses to block the reformed logical argument seriously reduces OT’s epistemic standing (pp.175-80). While the argument of chapters 5-7 is, I think, persuasive, the aforementioned difficulties show that the indirect empirical argument fails unless O’Connor can provide a more complete description of Wp.
Chapter 9 switches to Rowe’s direct empirical argument from evil. Rowe argues: (1) there exist instances of suffering (or, on O’Connor’s emendation, many instances of suffering) that God could have prevented without thereby losing some greater good or preventing some greater evil (i.e., that lack a God-justifying reason), (2) the existence of such instances (or, on O’Connor’s emendation, the existence of many such instances) logically precludes the existence of God, hence (3) God does not exist (p.188-9). The crucial premise here is the first. In support of it, Rowe offers the example of a “fawn, trapped in a forest fire caused by lightning, horribly burned, lying in terrible agony for days before dying” (p.190). So far as we can see, the fawn’s suffering (an example of NERNP) is pointless: it is inscrutable in the sense that we see no God-justifying reason for it. The same applies to many other instances of suffering. Yet–and here is the crucial premise–if there were a God-justifying reason we would likely see it. Hence, these instances of inscrutable evil are probably gratuitous: they probably lack a God-justifying reason. Hence (1) is (probably) true.
Chapter 10 considers the skeptical theistic response to Rowe offered by Stephen Wykstra, William Alston, and others. The response claims that the crucial premise in Rowe’s argument lacks support: pace Rowe, there is no good reason to think that if there were a God-justifying reason for the fawn’s suffering (or for any other instance of inscrutable evil) then we would likely see it. After all, God’s ways are beyond us: perhaps every inscrutable evil is outweighed (and thus given God-justifying reason) either by goods we cannot grasp or by goods that have not yet occurred (pp.199-204). Indeed, as the hypothesis that every inscrutable evil is so outweighed explains the facts of evil just as well as the hypothesis that some (or many) inscrutable evils are not so outweighed, the facts of evil do not count in favour of the latter hypothesis or, consequently, in favour of atheism. Given our intellectual limitations (finite humans can hardly comprehend God’s vision), we are simply not in a position to judge whether or not the facts of evil support theism or atheism. This means (1) that Rowe’s argument fails to offer strong evidence against theism (i.e., evidence that would oblige theists and agnostics to move significantly towards atheism), and (2) that the standard model of debate on the problem of evil (which assumes we are in a position to judge which way the facts of evil point) is flawed (pp.230-1).
O’Connor mostly accepts this skeptical defense, and thus argues in Chapter 11 for the aforementioned detente. But, he says, the defense has hidden costs. If, to avoid disconfirmation, an empirical theory must appeal to evidence to which nobody in this universe could ever in principle have access–as OT does when conjoined with the skeptical defense (the evidence being the occurrence of goods beyond our ken or goods in another lifetime)–then that theory is not really an empirical theory at all (pp.221-4). Hence a skeptically defended OT is not an empirical theory, which undermines the project of natural theologians such as Swinburne (who claim OT is a well supported empirical hypothesis) and leads to the dilemma I mentioned at the outset: either accept OT is not an empirical theory, or else retreat to the standard model of debate on the problem of evil.
The view that skeptical theism undermines natural theology seems plausible, and it will be interesting to see how theists respond to O’Connor’s argument. However, is O’Connor right to accept the skeptical defense in the first place? If we consider only Rowe’s argument, I think he may be. But other formulations of the problem of evil, not considered by O’Connor, may overcome that defense. For example, the hypothesis (call it H) that every inscrutable evil is outweighed either by goods beyond our ken or by goods that will occur in an afterlife might be judged (other things being equal) to be far less probable than its denial (not-H) on the basis that not-H is much simpler. Since not-H entails atheism, it would follow that, in the absence of outweighing evidence, atheism is probably true. It remains to be seen whether such formulations are successful; but, as this seems a live possibility, O’Connor’s call for a detente is at least premature.
So God and Inscrutable Evil is not the last word on the problem of evil; it will, however, set the stage for much future discussion. Despite some weaknesses, O’Connor’s arguments are on the whole sufficiently plausible, and the issues covered sufficiently interesting, to make this a highly rewarding read for anyone who wishes to stay at the forefront of debate on the problem of evil.