Klaas Kraay (of Ryerson) has an interesting paper (PDF) called "Theism and the Multiverse" up (H/T: OPP). In it he argues that theists, at least of the particular sort who hold that God is unsurpassable and creates by surveying the set of all actualizable possible worlds and picking one based on its axiological properties, are plausibly committed to a multiverse. It's an interesting paper, as all of Klaas's work is.
The basic idea is this. You have three basic options. Either:
(EOUW) There is exactly one unsurpassably good world.
(NUW) There are no unsurpassably good worlds.
(IMUW) There are infinitely many unsurpassably good worlds.
On each of these suppositions people have run arguments for atheism. Given EOUW some have argued that an unsurpassable being would choose that unsurpassably good world and that this world is not it. Given NUW some have argued that any being that chooses any world would be surpassed by a being that chooses a better world. Given IMUW some have argued that God would not have sufficient reason to actualize any world rather than any other.
It's no secret that I think all of these arguments are naive and (to be frank) borderline silly -- The arguments all seem to me to involve naive and dubious assumptions about the relations between the axiological properties of products, producings, and producers. So I don't think any of them are defensible, at least without a lot of supplementary argument that I doubt can be made plausible. But it's interesting to set this aside and look at what happens if you assume that they are good arguments. We then get a trilemma.
Klaas's suggestion is that the theist can (possibly) evade the trilemma by positing a multiverse -- a single possible world containing multiple universes (where a universe is a "self-consistent, spatiotemporally continuous, causally closed aggregate"). This allows you to run a parallel between universes and possible worlds. Just as we have three possibilities for possible worlds (EOUW, NUW, and IMUW), we have three possibilities for universes in the one possible world of the multiverse that correspond to each of these three: exactly one, none, and infinitely many. In a theistic multiverse, God creates all and only the universes worth creating, and those universes are the only universes (there are no uncreated universes or universes created by demiurges other than God). Klaas plausibly argues that the theistic multiverse is one, and only one, possible world (at least one possible world is the theistic multiverse, and at most one possible world is the theistic multiverse). He also argues that if the theistic multiverse is possible, NUW and IMUW are false. I find this argument rather less plausible, in great part because I don't see how piling more worthwhile universes on top really makes the possible world an unsurpassable one. Quantity and quality are like oil and water: they're good when you shake them together to put on your salad, but they tend to separate on their own. Having more chocolates doesn't give you better chocolate taste; and having more universes with good-making properties doesn't give you better good-making properties. This is why some people might have the sense that a lot of these universes, while worthwhile on their own, are simply redundant in the theistic multiverse. Having more good things doesn't correlate in any simple fashion with having a better set of things; the former only regards good-making properties of the things, whereas the latter requires that we consider the good-making properties of sets of things, and we can't assume that the latter reduce to the former. Likewise, we can't assume that the good-making properties of the theistic multiverse as a possible world are simply the sum of the good-making properties of the universes it composes; and it is only the good-making properties of possible worlds as such that are really in play when we consider whether there is an unsurpassable world (and if there is at least one, how many). So here again I think the argument considered is dubious, all things being taken into account. But if we assume, as the theist Klaas is considering assumes, that some form of the principle of plenitude is right, then it's certainly true that the theistic multiverse is the only game in town. So I think the argument that only EOUW is consistent with the possibility of the theistic multiverse, given the assumptions made so far, is quite sound.
But this, of course, brings us right back to the atheistic argument based on EOUW. Note, however, that the second premise (we are not in an unsurpassable world) becomes somewhat tricker to defend in the theistic multiverse, because we are only in one universe of this possible world. That's not to say that the atheistic arguer is in dire straits, of course. What it primarily brings out is that the atheistic arguer needs to show not that this world as such is surpassable, but that this universe is not worth creating (that it be worth creating is, of course, a requirement of its being a universe in the theistic multiverse).
So we seem to have run a sort of Red Queen's race -- running as fast as we can to stay in one place. Things are more or less as they were before. But if a theist is actually going to accept the soundness of the atheistic arguments on NUW and IMUW, and is committed to a principle of plenitude, and things like that, that 'more or less' is a bit of wiggle room that would not otherwise exist. For the problem really has shifted. Once the atheistic arguer was arguing that this possible world is surpassable; but the theistic multiverser (for lack of a better label) has shown that the only way to show that is to show that this particular universe is not worth creating. The theistic multiverser has moved the dispute from the surpassability of this possible world to its worthiness for being created, even granting the assumptions that I said I find dubious. And that, I would point out, is exactly where the dispute should be. And so people like myself who think the theistic multiverser has conceded a lot of silly things that need not have been conceded can rejoice at this victory: It's not a small victory to get to the right place even on the wrong assumptions.
