This book tackles a question which is now unpopular amongst professional philosophers – even amongst professional metaphysicians – but which continues to grip the layman's imagination: why is there anything at all? More precisely: (1) is there anything contingent, and (2) if there is, why is there? O'Connor's treatment of this question starts from a way back, with (1), and thus with an analysis of the metaphysics and epistemology of modality. This then forms the framework for him to develop and defend, in response to (2), a contingency version of the cosmological argument – buttressed at a late stage with a fine-tuning version of the design argument – for a particular sort of entity whose necessary existence and choice to create provides, O'Connor argues, the best answer to the question of why there is the contingent stuff that there is. The book closes with a chapter suggesting that there's no good reason not to identify this necessary entity with the God of Abraham and Isaac.

As this will have indicated, Theism and Ultimate Explanation is a work of classic natural theology – O'Connor himself traces his influences to Aquinas and Scotus (as well as, most obviously, Leibniz (though, significantly, his argument does not depend on such a strong form of the Principle of Sufficient Reason as Leibniz endorsed and is more defensible thereby)). But it is also a work of analytic metaphysics, especially at its earlier stages. Whilst the first part of O'Connor's argument, on the metaphysics and epistemology of modality, will be relatively hard-going for those who are not already at least somewhat familiar with the ground it traverses, O'Connor is an accommodating as well as reliable guide: the views he dismisses as he passes them by are given brief, but entirely self-sufficient, descriptions. And, in following this part of his route, the non-specialist is helped by the fact that the position O'Connor ultimately ends up at is the one which the non-specialist might well have been inclined to start from in any case. This position O'Connor calls ‘modal realism’, although the choice of this name is not an entirely happy one in that it is definitely not the position of David Lewis, which commonly goes by it. Unlike Lewis's modal realism, O'Connor's modal realism will strike many non-specialists as common-sense, even obvious.

Once one moves on from the first (and shorter) part of the book, the second part, which is concerned with laying out the argument from contingency to God, may strike readers of this journal as easier ground, easier as being ground of which the landmarks are already familiar to them, although – by the end of the journey – O'Connor is pushing the argument in novel and surprising directions. All in all then, Theism and Ultimate Explanation gives a thorough and careful defence of a contingency version of the cosmological argument, casting new light on some old problems, and, especially towards the end (e.g., when considering what sort of ‘super-universe’ – a super-universe being a set of universes – the theist is committed to seeing God as creating), making some new and striking claims in the philosophy of religion.

The first part of Theism and Ultimate Explanation argues that there are ‘real’ modal truths: we should not be modal nihilists (like Quine); reductionists (like Tarski, Armstrong, and Lewis); deflationists which are very similar to reductionists (like Rosen and Sider), or anti-realists (like Sidelle and Blackburn). Why not? Well, there are problems peculiar to each of these attempts to do away with ‘real’ modality, but there is also a general difficulty which O'Connor diagnoses: none of these attempts allows us to continue with our ordinary interrogating of the world through the sciences and explaining bits of it. So, to Quine, for example, O'Connor says that is

Modal truths – both core and non-core – are ones which we can have rationally justified beliefs concerning, but we reach this justification not in the manners that Yablo and Peacocke have suggested. Rather, such ‘beliefs are justified by a process of reflective equilibrium on the jumble of basic and often implicit modal beliefs that we find ourselves with at the outset of theoretical inquiry’ (ix). These modal beliefs may be divided into two categories: ‘theoretical’ (e.g., it's impossible for a given proposition to be both true and false at the same time) and ‘objectual’ (e.g., it's impossible for there to be water in a universe with no hydrogen). Some such beliefs then – as already seen – are more ‘core’ than others, and of the non-core ones we can, of course, hold that it is at least epistemically possible that we might be rationally led to abandon one or two as we seek to find a new equilibrium in the face of some recalcitrant experience. But at each stage we'll be able to form some suitably justified modal beliefs concerning the world. So, for example, I know with a fair degree of certainty right now that it really is possible that I could have been a better philosopher. And I know with even greater certainty that it is not really possible that I could have been a philosophy. It really is contingent then that I'm as poor a philosopher as I am whereas it really is necessary that I'm not an abstract object and, albeit fallibly, I can know these and a multitude of other things about modal reality.

By the time one gets to the second part of O'Connor's book, one has been given good reason to believe that there is, in a robust sense, contingent stuff. At the most general level (most general until one gets on to consider super-universes, that is), the universe – we may confidently say (after a bit of reflective equilibrium generation) – is contingent; it has the property of being such that it could have failed to exist. Why then – we may properly ask – does it exist? O'Connor argues in the second part of his book that the best answer to this question is that there is a being that is necessary. This necessary being could be personal and reason-driven – Logos as O'Connor dubs it – or impersonal and/or arbitrary – Chaos. Logos comes out in front as more probable once we bring in evidence provided by the fine-tuning of the universe to life; that's a feature that's more to be expected on the Logos hypothesis than it is on any Chaos hypothesis. The book moves on: Logos may be identified as the God of classical theism, albeit that this identification isn't really argued for by O'Connor (as he himself admits): the identification is, as he puts it, more a ‘working hypothesis’. But it is one which we don't – O'Connor goes on to suggest – find any reason to abandon.

To prevent us sawing off behind ourselves the branch which we're climbing along, we need, of course, to characterize God as only contingently creating the universe or universes that He does, but that is quite feasible. God, whilst necessary in Himself, does not necessarily create whatever it is that He does create; He has choice. But, according to O'Connor, it ‘would be passing strange that God would opt for less than the best when creating the best involves no cost at all! It is implausible that a perfect being should have idiosyncratic preferences for certain kinds of universes, quite apart from their value’ (113). So how then may God's choice be preserved? Fortunately, choice (and thus the contingency of what God creates) may be preserved in that whilst ‘it is inevitable both that God create something or other and that He create at least a countable infinity of universes’ (121), there are an infinite number of such sets of universes – super-universes – that God might create, viz. every super-universe which contains an actual infinity of universes all of which fall on or above a certain threshold of goodness.

That theism commits one to the existence of a super-universe with an infinite number of members, O'Connor observes, helps with the problem of evil. ‘Why did God not create a better universe than this?’, we ask, gesturing towards bits of our universe which seem pretty bad. Answer: ‘He did: He created an infinite number of such universes’. ‘Er, okay’, we respond, ‘so why did He create one as bad as this?’. Answer: ‘Because overall this was good enough to be above the threshold’. Finding bits of our universe which are bad and could have been better without any net loss to the universe as a whole (even if it could be done) wouldn't be finding evidence that O'Connor's God doesn't exist. To do that, one would need to find that the universe overall was below the permissible threshold, and this would be impossible if only because (although not only because) the threshold itself would be hard to find with any certainty.

It is towards the end of the book, where O'Connor pushes his argument towards these conclusions, that this particular reviewer found himself parting company with him to a significant extent (there were one or two niggles earlier on, e.g., on fine-tuning). So, for example, the premise that ‘it is inevitable, then, that a perfect God would create’ (113) gets scarcely a page of justification preceding it and is not one which seems to me irresistible. Surely one might, with some plausibility, argue that it is more obviously compatible with God's perfection to maintain that He was free to create nothing whatsoever as well as free between choices of which super-universe to create. That God created something does not seem to me to flow from His perfection in the ‘Dionysian’ (112) way O'Connor finds inevitable. Similarly, to me, the claim that God couldn't have idiosyncratic preferences seems needful of more defence than it receives.

It might perhaps strike us as strange (but surely not ‘passing strange’) that God might have idiosyncratic non-value-based preferences. There would not be an explanation of such preferences, of course, beyond the fact that such preferences were ones that did not violate value-based reasons. But O'Connor himself has wisely forsaken the idea that adequate explanations have to be, as he puts it, ‘contrastive’ – that for a genuinely adequate explanation of x, one has to be able to explain why every alternative to x failed to obtain. An idiosyncratic God who preferred to create certain types of universe rather than others as a result of idiosyncrasies that do not detract from His perfection in any way is at least as plausible to me as O'Connor's arbitrary God, arbitrary in that He selects for actualization one super-universe from amongst the infinite options available to Him and selects the one He does for no reason at all. If my idiosyncratic God is ‘passing strange’ in His idiosyncrasies, surely I can equally well maintain that O'Connor's is ‘passing strange’ in His arbitrariness.

And what about this claim that God finds Himself compelled to create a super-universe with an infinity of members? This surely is dubious. No denizen of universes God fails to create gets wronged by God's failure to create them, so again, it seems to me, God might well decide not to create such a plenitude, restricting Himself – idiosyncratically or arbitrarily – to a finite number of universes, possibly even one. Sure, there'll then end up being fewer types of good than there could have been, but if no-one's harmed by their absence, so what? After Santa's delivered presents to every inhabited house on Christmas Eve, does he then feel morally compelled to deliver them to every uninhabited house too? So it is that, before O'Connor's train of reasoning pulled into its final destination, this reviewer had alighted from it. Others would, of course, have waved it goodbye at yet earlier stages. But even if one decides that one will not oneself make O'Connor's journey in full, one can certainly enjoy watching him make it and learn much about the metaphysics and epistemology of modality and the contingency version of the cosmological argument by doing so.