I haven't read Wieseltier's review of Dennett's Breaking the Spell, since I don't generally read New York Times book reviews, and don't have a subscription to get behind the subscription wall. I notice, however, from the extracts that are occasionally put forward in responses to it, that the review briefly discusses Hume's Natural History of Religion. NHR is a very tricky book, so I thought I would say a few things to aid people in evaluating that part of the review. I have previously summarized the argument; I recommend you read that first.
(1) Wieseltier is right that in NHR Hume explicitly insists that the design inference to God's existence is rational and obvious. There appear to be three options for interpretation:
(a) Hume is, for some reason or other, lying, or, as we usually say, 'being ironic';
(b) Since Hume is explicitly setting aside rational arguments for God's existence in order to look at the original history of theistic opinions, he is simply conceding as much to the rational side as he can, without committing himself to anything in particular;
(c) Hume does accept the design inference in some form, and is just not discussing here the sense in which he accepts it.
The position suggested by (a) is very implausible, since part of Hume's argument that the original religion was polytheistic won't work if it is not true that the conclusion to a designer from the frame of nature is rational and obvious. (b) has some difficulty with this as well, but can, I think, be formulated in a way that can handle the problem. So either (b) or (c) or some combination of the two seems the right way to interpret NHR's comments on the subject.
(2) The question of whether Hume was, strictly speaking, an atheist, is one of the bugbear problems of Hume scholarship; I have a whole box of issues of Hume Studies for the past twenty years that underline how complicated this controversy is. We can distinguish, as a number of people have, a spectrum of possible opinions:
(a) strong theism: The most notable examples of philosophical strong theists in the early modern period would have been the Cartesians. A strong theism not only involves belief that there is a God, but belief that this God has a robust set of properties (e.g., infinity, omniscience, omnipresence).
(b) weak theism: The weak theist believes in God in some sense, but very little commitment to further attributes.
(c) weak atheism: The weak atheist believes there is no God, but is willing to allow that the theist is probably on to something. From this atheistic perspective, the problem with theism is exaggeration -- to put it roughly, the theist recognizes some genuine facet of the universe, but makes too much of it.
(d) strong atheism: Strong atheists believe there is no God and are not willing to allow to the theist as much as the weak atheist is.
Both (a) and (d) are very improbable interpretations of Hume. For one thing, in his public writings Hume always positions himself as a theist; but he argues at great length against strong theism. Russell's article at SEP on Hume's religion has a good discussion of exactly where Hume falls. My own sense is that Hume wavers between (b) and (c) (in the Dialogues he argues that their only difference is one of emphasis); but any position here is underdetermined by the evidence. One of the reasons Russell is one of the top scholars doing philosophy of religion is that he has managed to come up with some rather ingenious arguments that expand the field of evidence; but he takes a stronger position on Hume's leanings toward atheism than seems to be generally accepted. For instance, the point he makes in the article about the alternative analogies in the Dialogues is only telling if we assume a certain approach to analogical inferences. If Hume still has in mind an account of such inferences like the one he gave in the Treatise, however, the alternative analogies don't eliminate the design inference. Indeed, if Hume still accepts such an account of analogy, he is committed to the analogy he gives in the Dialogues, and the only question is how much can actually be concluded on the basis of it (this is, in fact, exactly the way he sets up the discussion). Likewise, Russell's mention of the end of NHR (the "riddle, enigma, inexplicable mystery" passage) isn't really relevant, because it is clear in context that Hume is referring to the conclusion he has just finished coming to: that both monotheism and polytheism are bad for morals, but that the more rational view (monotheism) is worse for morals than the less rational view (polytheism), and that none of these would be expected just from surveying the doctrines in question. Russell is right, though, that 'irreligion' is a better label for Hume's position than 'atheism' (it is also closer to what Hume's contemporaries meant when they called him an atheist than our own term 'atheist' is).
