I recently finished a section on faith & reason in my Introduction to Philosophy class. As per tradition, I included a discussion of the problem of evil and used David Hume’s writings on the subject. Condensing down his argument, he contends that we cannot reasonably infer the existence of an all powerful, all knowing and supremely benevolent being from the nature of the world. After all, there seems to be a significant tension between all the evil in the world and the existence of such a perfect being. Hume does note that the existence of evil is consistent with God having the qualities commonly attributed to Him, but he thinks that this is not what we would expect.
I think this condenses down the argument a stage too far. Hume's argument is not -- and cannot be, given how he actually crafts it -- that we cannot reasonably infer the existence of an all-powerful, all-knowing, and supremely benevolent being from the nature of the world, but that we cannot do so from natural phenomena on the basis of an analogical design inference. And that's a key issue, because not all inferences from empirical evidence are analogical design inferences. Cleanthes (and in a different context the Epicurean) is quite clear that it is the 'experimental inference' applied to the contrivance of the world, and it alone, that is in play: Like effects have like causes, therefore since the world is like an artifact the cause of the world is like an artificer, and so forth. And Philo's point is that, contrary to Cleanthes's claim, the design inference is not adequate for religion. Indeed, there are good reasons to think that this is the point of much of the argument in the Dialogues: Philo's aiming not at the design inference itself but at Cleanthes's assumption that it is a sufficient foundation for religion. Of course, there is reason to think that Hume allows Cleanthes to confine himself to this single type of inference because he thinks no other inference has genuine promise. But, while Cleanthes gives some slight argument in favor of it, we hardly get any systematic discussion of it.
And it is in fact something that not everyone would have agreed with; even Butler in the Analogy does not argue solely on analogy, and Butler would have replied to Hume with much the same argument he gave prior to Hume, i.e., that if we don't artificially reduce our reasoning to simple inferences, if we allow ourselves much more complicated forms of reasoning, we can get much farther. If, to give one very Butlerian instance, we have reason to think there is some sort of moral system in place (Butler argues at some length that we do), and that God is responsible for it, we cannot draw conclusions merely from where the system happens to be at particular stages in its existence but only from the whole system, which requires focusing not on the world at large but on the moral system itself in order to determine its "natural tendencies", which give us an idea of what will happen when the complete moral system is in effect. Butler notoriously regards good reasoning on the basis of the evidence even in this area of thought as an extraordinarily complicated thing, making use of many different kinds of inferences; he would not accept that it is something that can be reduced down to a single briefly stated causal maxim. It's doubtful Hume would accept such a line. Demea in the Dialogues makes a brief attempt to go in this direction, and is sharply and vehemently attacked for it by Cleanthes on the ground that it gets us into mere speculation. Cleanthes trounces Demea; but Demea is no Butler, either, and one can hardly move forward on the assumption that people will not question the key claim by Cleanthes that the one and only way to argue for divine benevolence is "to deny absolutely the misery and wickedness of man." It is a point on which Hume makes the argument much easier for himself than it might have been; not arbitrarily, it should be said, since he had many excellent reasons for taking Cleanthes in this direction. But it does place limits on how generally applicable the argument is, because most people will not concede all the crucial claims that get Cleanthes in his bind: that analogical design inference is the only possible way we can know anything about God, and that therefore we can only conclude that God is good if the design of the world as we can discern it is obviously good overall, and that therefore we can only conclude that God is good if the design is such that it obviously is currently such that human beings are happy and good overall. Indeed, I don't think I've ever actually met any theist who would concede all three of these.
And, of course, that really puts the argument back one step to the reasonableness of views in which these Cleanthean principles are rejected.
