On Donald Turner's theistic multiverse theory (recently ably defended by Klaas Kraay), God, out of his perfect goodness, necessary creates all possible universes that are "good enough", i.e., whose value is above a cut-off line, below which a universe is not worth creating. Thus, the world is a multiverse consisting of infinitely many universes. A God who actualized the best possible world can only create such a multiverse. The metaphysics is similar to that of David Lewis, but with only "good enough" worlds there. This gives a lovely answer to the problem of evil: "Ah, you think this universe would be better without evil E? It sure would—and there is another universe, a better one, in existence that lacks E, but it is better that both this universe and the other one should exist, than that just one of the two should. We see what we would expect to see on the theistic multiverse theory, a world with a lot of evil, but still worth creating."
I want to make a bit of trouble for the notion of a universe worth creating: Whether a universe is worth creating can depend on what other universes, if any, there exist.
Imagine a possible universe U that is just around the cut-off line between good-enough and not-quite-good-enough, and which contains a thousand philosophers who care about little but whether the multiverse hypothesis is correct, and who come to a justified[note 1] belief that there is only one universe concretely in existence. Assuming the notion of a cut-off line makes sense, it seems we could imagine that the universe is so close to the cut-off line that whether the universe falls above the cut-off line or below it depends on whether the philosophers' justified belief that there is only one universe concretely in existence is true. If so, then U is worth creating on its own, since then that belief is true, but it is not worth creating as part of a multiverse, since then that belief is false, and U is so close to the cut-off line that having this belief be wrong pushes it below the cut-off line.
So, the notion of the cut-off line is not so clear: whether a universe is below or above the cut-off line can depend on what other universes there are. Moreover, it seems one can have a pair of universes U1 and U2 such that each is worth creating as part of a multiverse only if the other is not included in the multiverse, and such that neither of the two is better than the other. If we can have such a pair of universes, then there is no unique optimal multiverse. For there will, presumably, be one possible optimal multiverse (i.e., multiverse than which there is no better) that includes U1 but not U2, and another possible optimal multiverse that includes U2 but not U1. How to construct such U1 and U2? Well, again, suppose that the two universes are marginal—very close to the cut-off line—and what decides whether U1 is above it or below it is whether the justified beliefs of U1's philosophers that U2 is non-actual are true, and what decides whether U2 is above it or below it is whether the justified beliefs of U2's philosophers that U1 is non-actual are true.[note 2]