William Rowe's evidential problem of evil has been a very influential work to modern theology, so much so that, as a response, a new brand of theological opinion has emerged; skeptical theism. I hope to establish that skeptical theism cannot be a successful answer to the evidential problem of evil, and that ultimately it comes down to the success or failure of theodicies. The bare bones of William Rowe's argument can be summarized as follows.
(1) Gratuitous evil exists
(2) If God exists then gratuitous evil does not
(3) Therefore, God does not exist
Gratuitous evil is understood by Rowe to be any instance of evil for which God, if he existed, would not—or could not—have justified reason for permitting. There are three conditions given by Rowe in which God might be justified in permitting an evil E; In this, premise (2) states that, if God exists, then every instance of evil falls under one of these three conditions. This strikes me as very reasonable; after all God, being omnipotent, cannot be thwarted by anything less than logical necessity, and, in his moral perfection, would strive for maximal goodness.
Turning our attention to premise (1) one must ask; how does Rowe seek to support this assertion? Rowe offers a sub-argument which, at face value, appears to employ the very same reasoning we use in our every day lives.
P: We are, upon careful reflection, unable to find any God-justifying reasons for E,
Thus, it's probably the case that
Q: There are no God-justifying reasons for E.
This inductive inference from P to Q (which I will further refer to as Rowe's inductive step) has been hotly contested by theologians of the likes of Wykstra and Alston arguing that, due to human limitations, we are unwarranted in moving from the one to the other. This position that both Wykstra and Alston hold has come to be known as skeptical theism. Skeptical theism can be broken down into two forms, the first of which is called strong skeptical theism for it's complete denial of Rowe's inductive step. The strong skeptical theist does not trust that our ability to infer from P to Q is even somewhat reliable. Weak skeptical theism, on the other hand, gets it's name from making the much weaker claim that, while our ability to infer from P to Q is somewhat reliable, it's not nearly reliable enough for us to justifiably come to any conclusions about Q. It should be clear, then, that both strong and weak skeptical theists seek to defend a complete agnosticism of Q and, with this, a complete agnosticism of the existence of truly gratuitous evil. This approach is different, though not exclusive to that of the theodicist's. Where the theodicist seeks to give a plausible explanation of the justifying reasons God might have in permitting particular evils, the skeptical theist seeks to undercut ones ability to arrive at the belief that these evils require explaining. More about theodicies will be said at the tail end of this paper, but for now I'd like to turn our attention to strong skeptical theism.
Let's suppose, for arguments sake, that the strong skeptical theist is correct and that our ability to infer from P to Q is not even somewhat reliable. In other words the epistemic confidence we ought to have in Q given P should not be any greater than prior to considering P. In mathematical notation we can understand this to mean ¬ Pr(Q|P) > Pr(Q) which is, of course, identical to Pr(Q|P) ≤ Pr(Q); let's take this as a first premise for a reductio ad absurdum. By a simple manipulation of Bayes theorem it follows that the epistemic confidence we ought to have in P given Q is less than or equal to the epistemic confidence we had in P prior to considering Q, or in other words; Pr(P|Q) ≤ Pr(P).
But Pr(P|Q) can't be any less than 1. We know with certainty that we wouldn't be able to find any God-justifying reasons for an evil which had no God-justifying reasons to begin with; after all you can't find what isn't there. But if this is the case, and surely it is, then it follows that Pr(P) likewise cannot be any less than 1. In other words it follows that we, upon careful reflection, would certainly be unable to find any God-justifying reasons for E. But theologians, contrary to what we've just discovered, have found some very good God-justifying reasons for many of the evils we experience in our day-to-day lives. Clearly this conclusion is not something anyone should accept be they theist or atheist and so we must reject each move we've made all the way up to assuming strong skeptical theism.
If we are to accept weak skeptical theism, as I am prepared to do, then Rowe's inductive step would seem to be blocked and, as such, his argument defeated. Is this the end for the evidential problem of evil? I think not. While I concede that we cannot justifiably infer that any particular evil is truly gratuitous, I don't believe this to warrant a complete agnosticism of Rowe's first premise. As I see it, Rowe's first premise need not even be supported by the controversial inductive step—at least not directly.
With this I present a new argument—a modification of Rowe's evidential argument which I believe sidesteps weak skeptical theism all together and strongly supports the denial of orthodoxy theism.
(1) There are an overwhelmingly many instances of seemingly gratuitous evil (observation)
(2) Our ability to recognize gratuitous evil is at least somewhat reliable (denial of strong ST)
(3) Thus, probably gratuitous evil does in fact exist (from 1 and 2)
(4) If God exists, then gratuitous evil does not (assumption)
(5) Therefore, probably God does not exist. (from 3 and 4)
It may not be immediately obvious how the third premise follows from the first and second, to get an idea of how it does we will need to use a bit of math. Suppose that our ability to recognize gratuitous evil is at the very least 1% reliable, any less and we begin to venture back into the realm of strong skeptical theism. It then follows that the chances of us being wrong about a particular evil's being gratuitous is 100% - 1% = 0.99. If we then considered a mere six hundred cases of seemingly gratuitous evil it turns out there would be a 0.99^600 = 0.00240500929 or 0.2% chance of us being wrong about every single case of seemingly gratuitous evil. This means that it's overwhelmingly probable (99.8%) that we'd be correct about at least one of the evils. Now due to weak skeptical theism we could never actually know which of the evils were likely to be gratuitous and which were not, but that doesn't make a difference to the argument. I must now stress that the numbers I provided were solely for example, clearly the strength of our inference will not be the same for all evils—and yet the numbers do make explicit the sort of reasoning I'm using here. If we had only taken into consideration three seemingly gratuitous evils, it wouldn't have been enough to warrant a belief that at least one was probably gratuitous. The more evils we have to consider the stronger our inference will be and, likewise, the fewer evils the weaker the inference.
I had mentioned theodicies before and it comes into relevance now. Premise (1) will not and should not be accepted by anyone who believes themselves to be able to explain most of the evil that occurs in this world. Since many theists will be warranted to deny the first premise in this way, it might be unwise to call this argument a “problem of evil” for orthodoxy theism in general. That being said, there are still a great many theists and atheists alike who do not consider theodicies to be very successful and to these people my argument should be particularly compelling.