Erik Wielenberg has a well-known paper, "A morally unsurpassable God must create the best" [Religious Studies Vol. 40, No. 1 (Mar., 2004), pp. 43-62]; the title gives the point. 'Morally unsurpassable' is obviously a key idea here. He defines it as follows:
[D1] x is a morally unsurpassable agent in w =df. (1) x is an agent in w, and (2) it is not the case that there is some world w' and some being y such that y is a better moral agent in w' than x is in w.
This is not an entirely satisfactory definition, since there are obvious problems with what it means to be a "better moral agent". Are we to take it to be "a moral agent who does better things", "a moral agent with better character (i.e., better facility in doing good things)", "a moral agent with capacity to do better things", "a moral agent with better capacity to do moral things", "a moral agent who is better at doing moral things", or what? Wielenberg considers this, although in a somewhat roundabout way, discussing Wierenga's notion that 'better moral agent' is to be understood as 'more virtuous', which he then reglosses as "more disposed to pursue intrinsically valuable states proportionally". This is not a standard way to understand 'virtuous' in virtue ethics, since virtue is itself a proportional intrinsically valuable state, and many virtues pursue extrinsically valuable states, and most virtues don't pursue anything at all (they just do things) unless we're speaking very loosely. This is perhaps not surprising; he is adapting Hurka, and Hurka is a consequentialist, so we get a consequentialist account of virtue. But Wielenberg also recognizes this and builds an elaborate ad hoc workaround to leave out cases in which the morally unsurpassable agent wouldn't be more disposed to pursue intrinsically valuable states proportionally. It seems to me that this should raise red flags immediately, but let us set this worry aside, and just go with the interpretation of 'better moral agent' as 'more disposed to pursue intrinsically valuable states proportionally'. As we will see, this still doesn't end the problem.
In any case, he adds to this two other definitions:
[D2] Possible world w is surpassable for x =df. there is some possible world w' such that (1) x can actualize w' and (2) w' is better than w.
[D3] x is creator =df. x actualizes some world or other.
Possible worlds are not 'actualized' except as a loose figure of speech; it's the actual world that is 'actualized' (by something existing and doing something), and the actual world is not a possible world in the sense of possible world semantics. Wielenberg at one point says that possible worlds are states of affairs. This is also not strictly true -- possible worlds are logical objects mapped to claims with truth values, interpreted as describing ways the actual world can be -- whereas states of affairs, whatever else they may be, involve mappings to gerund clauses, not complete propositions. But if one did hold that possible worlds were states of affairs then, 'actualizing a possible world' would mean 'being the cause that is the reason why a world-sized state of affairs obtains'. The real idea here is that the actual world would be surpassable for x if x actualizes the actual world and there is some possible world (which describes a possible way the actual world is) that is better than the possible world(s) whose description is closest to the description of the actual way the actual world is. x, on the other hand, is creator if x actualizes the actual world, whatever way the actual world might be.
But take the definitions as they are. Then Wielenberg reasons as follows to conclude that a morally unsurpassable agent must create an unsurpassable world. Suppose A creates a world whose value has measure v and B creates a world whose value has measure v', and suppose that v' > v. "Unsurpassably proportional pursuit" requires that when you have two things that you can bring about, and you can't bring about both, and one is better than the other, you bring about the better one. Therefore B is more disposed to pursue intrinsically valuable states proportionally. Unfortunately for this line of reasoning, there is a slip of scope here; the argument doesn't prove that B is more disposed to pursue intrinsically valuable states proportionally, but that B is disposed to pursue intrinsically valuable states more proportionally. Just going on (D1) as interpreted in terms of disposition to pursue intrinsically valuable states, we have no reason to regard these as equivalent; it seems entirely possible that A could be more disposed (or as disposed, since Wielenberg is assuming that A and B are both omniscient and omnipotent and that nothing else is relevant to how they are disposed) to pursue intrinsically valuable states proportionally even though B is disposed to pursue them more proportionally. My point above about the different ways to interpret 'better moral agent' was not a quibble; even with the definitional work Wielenberg does, it resurfaces again. If we slim down to just A, we run into the same problem. Wielenberg thinks that if A creates a surpassable world, then, since A is "otherwise identically disposed" whether he creates this world or a better one, then he is more disposed to pursue intrinsic valuable states proportionally if he creates an unsurpassable one. But the 'otherwise' is a little sleight of hand here whose vague suggestion that there is some difference relevant to the disposition is not motivated by anything in the set-up; given Wielenberg's assumptions about the dispositions of creators, A's disposition in both cases is identical. In either case A is disposed to pursue intrinsic valuable states proportionally, and in both cases it is exactly the same disposition. Now if dispositions and actions necessarily have to match up one-to-one, so that any difference in action at all is a difference in disposition, it's true that we would then have to say that, since the actions are not the same, the dispositions are not the same. But even if that extremely substantive and controversial assumption were made, we have no particular to reason to say that they would not be the same specifically in that one was a greater degree of the actual disposition to pursue intrinsically valuable states proportionally, as opposed to some other way of being different -- such as being equal as to disposition but one being disposed to a slightly different way of handling proportionality. That is to say, the same ambiguity arises with A alone as when we were comparing A and B.
More disposed to do something Y-ly is in general not the same as disposed to do something more Y-ly. Being more disposed to cross a street with care is not the same as being disposed to cross a street with more care; for instance, in the first case you could be more eager to cross a street, as long as it is with care, whereas in the second you might not be particularly eager to cross a street, but if you do you will be very, very careful. If Joe is really enthusiastic at crossing streets with basic reasonable precautions, for instance, this is not the same as if he were crossing streets with more than basic reasonable precautions. The point is actually an important one for ethics, since Wielenberg is not the only person to confuse the two, by any means. The necessity of distinguishing them falls fairly from a number of common ethical positions, such as the Doctrine of the Mean. (Wielenberg in a footnote actually tries to link proportionality to the Doctrine of the Mean, but has not, I think, thought through carefully enough the way in which disposition relates to the mean.) Being more disposed to speak truthfully is one thing; being disposed to speak more truthfully is another. If you are an honest person or a blabbermouth truthteller, you are more disposed to speak truthfully than a liar; if you are a blabbermath truthteller, you are disposed to speak more truthfully than an honest person would (which means you have a vice, not a virtue), but you are not necessarily more disposed to speak truthfully than an honest person is (this is purely a matter of how intensive or deeply rooted your disposition has become). You don't become more honest by saying things that are more true (which would perhaps imply that just reading a list of tautologies aloud is maximal honesty), nor by saying more truths; you become more honest by becoming better disposed with regard to saying truths.
This illicit pattern is in fact found in pretty much all "the best agent must make the best thing" arguments, of whatever kind. Maybe you could have one without this illicit move, but I haven't come across it. And we know it's illicit because there are many, many counterexamples. The best writer might not write the best essay; for instance, the writer might be deliberately writing a bad essay in a writing workshop for teaching purposes. The best advisor might not give the best advice; for instance, she might settle for lesser advice that is the best to which people will listen. The best drawer might not make the best drawing; for instance, it might be a case of playing around a bit with different kinds of sketches just for the fun of it. The best chess player might not make the best move; for instance, he might find the game itself interesting enough that he wants to see where his opponent is going, strategy-wise, rather than starting a crushing endgame now. (Chess is an interesting example, because there is an entire history of virtuoso chess in which great chessplayers make suboptimal moves to show that they can still win even when giving their opponents absurd advantages; virtuoso-style play was very popular in the eighteenth century, and you have examples of things like great chess players deliberately losing half their pieces before coming back to win.) And so forth and so forth and so forth. And we get this result in part because being the best at something typically means you have a wider, not a narrower, scope of action; if you are the best painter, the range of ends you can achieve by painting is greater than the range of ends that can be achieved by the merely good painter, and massively greater than the range of ends that can be achieved by the bad painter. To be sure, you could say something like, "The best writer in writing the awful essay for teaching purposes is writing the best awful essay for those purposes", but this changes the subject, because now we need to know exactly what those purposes are, and the very same thing can be the best thing for one purpose and not the best thing for another, and the bare statement, "the best agent must make the best thing" is no longer true simpliciter, but only with significant qualifications, if it is even really true at all.
We should thus be extraordinarily skeptical of the idea that the best of all best agents, with an omniscient ability to consider ends that we cannot even conceive, and with an omnipotent power to achieve them, is capable of making one and only one kind of thing, rather than being capable of making many different kinds of things suitable for many different purposes.
