A. Rowe’s Arguments

I turn first to an argument William Rowe has been proposing and developing for the past twenty years.⁵⁷⁷⁵⁷⁷   See his “The Problem of Evil and Some Varieties of Atheism,” American Philosophical Quarterly (1979), pp. 335–41, reprinted in The Evidential Argument from Evil, ed. Daniel Howard-Snyder (Bloomington: Indiana University Press, 1996), pp. 1–11; “Evil and the Theistic Hypothesis: A Response to S. J. Wykstra,” International Journal for Philosophy of Religion 16 (1984), pp. 95–100; “The Empirical Argument from Evil,” in Rationality, Religious Belief, and Moral Commitment, ed. Robert Audi and William J. Wainwright (Ithaca: Cornell University Press, 1986); “Evil and Theodicy,” Philosophical Topics 16 (1988), pp. 119–32; “Ruminations about Evil,” Philosophical Perspectives 5 (1991), pp. 69–88; “William Alston on the Problem of Evil,” in The Rationality of Belief and the Plurality of Faith, ed. Thomas D. Senor (Ithaca: Cornell University Press, 1994); and “The Evidential Argument from Evil: A Second Look,” in Howard-Snyder, The Evidential Argument from Evil (hereafter EAESL). Consider some particularly horrifying cases of evil or suffering: a five-year-old girl’s rape and murder (E₁) or a fawn’s lingering and painful death in a forest fire (E₂). Rowe’s argument goes as follows:

P: No good we know of is such that we know that it justifies an omnipotent, omniscient, perfectly good being [a perfect being, for short] in permitting E₁ and E₂;⁵⁷⁸⁵⁷⁸   Rowe actually states P as “No good we know of justifies an omnipotent, omniscient, perfectly good being in permitting E₁ and E₂”; neither the theist nor the neutral bystander, however, can be expected to accept this premise because it could be that some good we know of does justify a perfect being in permitting E₁ and E₂, even though we don’t know that it does. Indeed, Rowe countenances conjunctive goods such as G, the conjunction of all the goods there are. But G (one supposes) is a good state of affairs; and if theism is true, G justifies a perfect being in permitting E₁ and E₂. Alternatively, the theist might think that the unthinkably great good of incarnation and atonement—a good that we know of—justifies E₁ and E₂. This could happen as follows: God selects for actualization one of the best worlds; but all the best worlds include incarnation and atonement (see below, p. 489), and hence also a great deal of evil—if not specifically E₁ and E₂, then others just as bad. The most the atheologian can sensibly claim, therefore (if he is hoping for agreement from theist and neutral bystander) is that no good we know of is such that we know that it justifies a perfect being in permitting E₁ and E₂. If Rowe insists on his premise as originally stated, then, it seems to me, the theist should respond that there is no reason to think it true and good reason to think it false.

Therefore, probably

Q: No good at all justifies a perfect being in permitting E₁ and E₂;

Therefore probably

not-G: There is no perfect being.

Here we are thinking of goods and evils as states of affairs. A state of affairs can be actual or nonactual; only an actual good, says Rowe, could justify a perfect being in permitting E₁ and E₂ (or, indeed, any other evil). So the idea behind P is that we do not know of any good that is actual and is such that we know that it suffices to justify a perfect being in permitting E₁ and E₂.

There are several problems with this argument. At the simplest level, however, the main problem, once the others are straightened out or ignored, is with the inference from P to Q. I look inside my tent: I don’t see a St. Bernard; it is then probable that there is no St. Bernard in my tent. That is because if there were one there, I would very likely have seen it; it’s not easy for a St. Bernard to avoid detection in a small tent. Again, I look inside my tent: I don’t see any noseeums (very small midges with a bite out of all proportion to their size); this time it is not particularly probable that there are no noseeums in my tent—at least it isn’t any more probable than before I looked. The reason, of course, is that even if there were noseeums there, I wouldn’t see ’em; they’re too small to see. And now the question is whether God’s reasons, if any, for permitting such evils as E₁ and E₂ are more like St. Bernards or more like noseeums. Suppose the fact is God has a reason for permitting a particular evil like E₁ or E₂, and suppose we try to figure out what that reason might be: is it likely that we would come up with the right answer? Is it even likely that we would wind up with plausible candidates for God’s reason? A series of important recent papers by Stephen Wykstra, William Alston, and Peter van Inwagen argue (among other things) that it is not.⁵⁷⁹⁵⁷⁹   Wykstra: “Difficulties in Rowe’s Argument for Atheism, and in One of Plantinga’s Fustigations against It,” read on the Queen Mary at the Pacific Division Meeting of the American Philosophical Association, 1983; “The Humean Obstacle to Evidential Arguments from Suffering: On Avoiding the Evils of ‘Appearance’,” International Journal for Philosophy of Religion 16 (1984), pp. 73–94; “The ‘Inductive’ Argument from Evil: A Dialogue” (co-authored with Bruce Russell), Philosophical Topics 16, pp. 133–60; Alston: “The Inductive Argument from Evil and the Human Cognitive Condition,” Philosophical Perspectives 5, pp. 29–67; van Inwagen: “The Place of Chance in a World Sustained by God,” in Divine and Human Action, ed. T. Morris (Ithaca: Cornell University Press, 1988); “The Magnitude, Duration, and Distribution of Evil: A Theodicy,” Philosophical Topics (1988); “The Problem of Evil, the Problem of Air, and the Problem of Silence,” Philosophical Topics (1991). One hopes these pieces will put the final quietus to the “I can’t see what reason God could have for p; therefore, probably God doesn’t have a reason for p” form of argument. (But of course they won’t.) The main reason is the epistemic distance between us and 467God: given that God does have a reason for permitting these evils, why think we would be the first to know? Given that he is omniscient and given our very substantial epistemic limitations, it isn’t at all surprising that his reasons for some of what he does or permits completely escape us. But then from the fact that no goods we know of are such that we know that they justify God in (serve as his reasons for) permitting E₁ or E₂, it simply doesn’t follow that it is probable, with respect to what we know, that there aren’t any such goods, or that God has no reason for permitting those evils. The arguments in these papers seem to me to be conclusive; I shall not repeat them here.

More recently (and partly under the pressure of some of the works mentioned in footnote 10), Rowe has himself come to view this argument with a jaundiced eye: “I now think this argument is, at best, a weak argument.”⁵⁸⁰⁵⁸⁰   See EAESL, p. 270. He therefore sets this argument aside in favor of one whose prospects he thinks are brighter: “I propose to abandon this argument altogether and give what I believe is a better argument for thinking that P makes Q more likely than not” (p. 267). After giving that argument, Rowe goes on to say that “we can simplify the argument considerably by bypassing Q altogether and proceeding directly from P to -G” (p. 270). This new argument goes as follows. First, we must note that Rowe intends P in such a way that it is entailed by not-G; P is equivalent to

P’ There is no perfect being and known good such that the latter justifies the former in permitting E₁ and E₂.

Rowe then assumes that P(G/k) and P(P/G&k) both equal .5 (where k is our background information—what all or most of us know or believe.) It then follows by the probability calculus that P(G/P&k) is considerably less than P(G/k); hence P disconfirms G. The argument thus simplified is Rowe’s new evidential argument from evil. I regret to say, however, that this new argument is, if anything, weaker than the old. That is because an analysis of purely formal features of the argument shows that it is counterbalanced by other arguments of the same structure and strength for a conclusion inconsistent with Rowe’s conclusion (and hence for the denial of Rowe’s conclusion).⁵⁸¹⁵⁸¹   For details, please consult my “Degenerate Evidence and Rowe’s New Evidential Argument from Evil,” Noûs 32, no. 4 (Dec. 1998); see also Rowe’s reply in “Reply to Plantinga,” Noûs 32, no. 4 (Dec. 1998). In essence, the problem is twofold.

First, Rowe’s argument really depends on the fact (as already noted) that the conclusion he proposes to support, i.e.,

not-G There is no perfect being

entails P, the premise of his argument. Now the probability calculus tells us that if a proposition A entails a proposition B, then B confirms A in the sense that the probability of A on B conjoined with our background information k will exceed that of A on k simpliciter (unless either A or B has an absolute probability of 1). Thus any contingent consequence C of not-G will confirm not-G with respect to any body of background information k (k, of course, cannot include or entail C).

But then by the same token, any contingent consequence of G will confirm G with respect to any body of background information k. This means that Rowe’s argument will be counterbalanced by other arguments—for example, one that takes as its premise any of the following propositions:

P* Neither E₁ nor E₂ is such that we know that no good justifies a perfect being in permitting it.

P** No evil we know of is such that we know that no perfect being is justified by some good in permitting it.

P*** No evil we know of is such that we know that no perfect being would permit it.

Presumably there will be as many arguments of this sort for G as there are arguments of Rowe’s sort against G.

The second problem is like unto the first. Rowe’s argument is really an “argument from degenerate evidence”—an argument in which you take as your new evidence, not the new proposition you learn, but a weaker consequence of it. We can see this as follows. Rowe’s premise P is equivalent to

P' Either not-G or no good we know of is such that we know that it justifies E₁ and E₂,

where a good g justifies an evil e iff if there were a perfect being b, and g and e were actual, then b would be justified by g in permitting e.⁵⁸²⁵⁸²   For the argument, see “Degenerate Evidence.” (For example, perhaps a certain kind of moral growth on my part requires a certain amount of suffering; and perhaps we can see that a perfect being would be justified by that moral growth in permitting the suffering in question.) Now what we learn by reflecting on E₁ and E₂ (and other evils) and their relation to a perfect being is really

-J No good we know of is such that we know that it justifies E₁ and E₂.

Clearly enough, -J entails and is stronger than P’, the premise of Rowe’s argument. And the problem with arguments of this sort is that, once again, there will be other arguments of the same structure and strength for an incompatible conclusion. For example, suppose I win the Indiana 469lottery (W). The probability of W with respect to k is very low, say one in a million. Now suppose I take as my new evidence not W, but

W or -G.

By an argument just like Rowe’s,⁵⁸³⁵⁸³   Again, for details see “Degenerate Evidence.” we can show that the probability of -G on this premise together with the relevant background information is very high indeed—something like .999999. Of course there is a similar argument for G; here the premise will be

W or G.

Clearly, neither of these arguments makes any real advance, and that is because they counterbalance each other.

Rowe’s argument from P to -G displays the same structure as this lottery argument. He proposes to argue for -G; our “new evidence” is really -J; but to get his premise P he weakens this new evidence by adding the conclusion of his argument, -G, as a disjunct, so that P is or is equivalent to the proposition -J or -G. That makes this an argument from degenerate evidence. To construct the counterbalancing argument we simply weaken -J by adding as a disjunct G, the proposition that there is a perfect being, rather than -G; this counterbalancing argument will be for the denial of Rowe’s conclusion and will be as strong as his. Arguments from degenerate evidence, clearly enough, do not serve to advance the discussion.