The trouble with the atheological argument against the existence of God from the occasion of evil(s) is that it is hopelessly speculative. I was explaining this recently to a good friend via an email exchange. I wrote:
Perhaps you could imagine that God wouldn’t do X (that would be incompatible with his all-good nature), and an X will be something like an instance of gratuitous evil (an evil which is not, in the final calculus of things, redemptive – i.e., for which God has no morally sufficient reason). However, there’s no way to empirically verify X. Some Naturalists do give a probabilistic empirical argument against the existence of God this way, by arguing that there are some surprising occasions of evil E, which seem to be gratuitous G. We can formulate this as follows:
- Ga (probably)This isn’t a deductive-nomological argument, but a statistical, probabilistic inductive argument. Still, nobody thinks, post-Plantinga, that you can have any deductive-nomological argument against the existence of God from evil. All such arguments have to be inductive and probabilistic.The trouble with this argument is that it is hopelessly speculative. What kind of criteria can we have for measuring the probability claim of the first premise? We would presumably have to know what God knows, or at least know most of what God knows, in order to know that God ‘probably’ doesn’t have any morally sufficient reason for allowing some instance of evil. The argument from evil, even when it appeals to things like the sanctioned death of all the Canaanites who stubbornly remained in the land, is just speculative. Any argument for there being a Gx will be less plausible than the argument(s) for believing that Theism is true, and since mutual exclusivity obtains between Gx and God’s existence, we have better grounds for denying Gx than we can have for affirming it.
Here’s a thought: suppose an Atheist were to append her argument from evil with an index of defeaters for other arguments for the existence of God, such that if we take any argument for the existence of God, like the ontological, teleological or cosmological arguments, we find that not a single one increases the probability that God exists. Take the set of all such arguments in Natural Theology to be symbolized as NT, and let G represent the proposition “God exists”. Then the claim will go as follows:
With this claim in place, one can proceed to argue that, even if the argument from evil is extremely speculative, it does make it somewhat more probable that God does not exist. Then one can argue that all things are equal, and run the argument in the way I formulated it in the email exchange. Technically one doesn’t even need P(G|NT)=P(~G|NT), one may only need the following, as Paul Draper (whose work on this I highly recommend) sometimes seems to put it (PE represents ‘Problem of Evil’):