My primary problem with the NPW argument is that I tend to think (P1) highly implausible:
(P1) If it is possible for the product of a world-actualizing action performed by some being to have been better, then, ceteris paribus, it is possible for that action to have been better.
(There is an interpretation of this on which it is trivially true, as Kraay notes; I am only considering it in the more substantive sense, since it is only in this sense that it really has force.) I see no reason to accept this principle, since I see no reason why there should be any straightforward relation between action and product: whether the action was as good as it could be depends in part, for instance, on whether the imperfection in the product was due to deficiency in the action or due to deliberate intent that was good in its kind. As a matter of pedagogy, for instance, I might deliberately make an imperfect product in order to have students improve it, or in order for students simply to see what to avoid; this might be done well enough (note that how well it is done would not correspond to how good the product was) that it could not be done better for the intended pedagogical purpose. Thus (P1) does not seem to follow from any general principle.
Could (P1) follow not from general principles, but from some requirement created by unsurpassability? I don't see how. For instance, Aquinas, in responding to the contrariety argument from evil (there is a reason I brought it up), replies (ST 1.2.3 ad 1):
As Augustine says (Enchiridion xi): "Since God is the highest good, He would not allow any evil to exist in His works, unless His omnipotence and goodness were such as to bring good even out of evil." This is part of the infinite goodness of God, that He should allow evil to exist, and out of it produce good.
In other words, if Aquinas, following Augustine, is right, the unsurpassable case is analogous to the pedagogical case I noted above. This highlights the fact that unsurpassability makes (P1) even harder to support: if someone held that (P1) were the result of a requirement of unsurpassability, this would effectively require them to argue that being unsurpassable makes it impossible to use something surpassable for an unsurpassably good and rational intent. In other words, it would effectively be a claim that God is so unsurpassable he can't produce "bring good even out of evil"; which appears to be necessarily false. Whatever unsurpassability is, it can't entail the surpassability of the goodness and rationality of God's intentions, nor his ability to effect them even with a surpassable product. (If it is consistent, that is, and as a reductio the NPW argument can't really presuppose that it isn't without begging the question.) Indeed, without being explicit about God's intentions, we can't be sure that the unsurpassable goodness and rationality of the action doesn't require surpassability in the product (in a way analogous to the pedagogical example).
So (P1) cannot be supported as a specific case of a principle about action generally, and it can't be a requirement of unsurpassability. Could it, then, be the result of a principle not about action generally, but about a certain kind of action? But unsurpassability seems to throw a wrench in the works here, too; if we take unsurpassability seriously, then Aquinas's response to the contrariety argument seems plausible; and if it does, world-actualizing action is a kind of action of which it is false to say that surpassability in its product precludes unsurpassability in the action itself, because again this would actually depend on what was actually intended in the action. If the unsurpassable being has an unsurpassably good and rational intention deliberately involving the making of a surpassable product, then it appears to be the case that the unsurpassable being's action is itself unsurpassable; to make a better product would not (as (P1) explicitly requires) make that action better, but would instead be an entirely different action, because it would require an entirely different intention. This action, of course, would also be unsurpassable; unsurpassability, understood in this way, has the implication that all God's world-actualization actions, and indeed, all God's actions at all, involve intentions that are equally unsurpassably good and rational; but (P1) would be false. (This, it is worth noting, coincides with Aquinas's view in ST 1.25.6 ad 3.) If the NPW argument is to be a reductio of (1), it has to give unsurpassability full play - it must not arbitrarily limit it. But if it does give unsurpassability full play, I think it is dead in the water. And this is confirmed by the fact that there are good reasons both to hold that there is an unsurpassable being, and that there is no best possible world (in the sense we have been considering).
But, to return to the paper, I agree with Klaas's conclusions about the particular defense he's considering, which attacks a different premise by a very different sort of route; and recommend it highly.