I had a post last year about R. T. Mullins's attempt to argue for a Problem of Arbitrary Creation for Divine Simplicity. The basic criticism was that Mullins's conception of impassibility was incorrect and the key premise ("If God is impassible, God created the universe for no reason") was inadequately motivated; in addition, the whole thing seemed to require a conception of how reasons work in knowledge and action that is false even in our own case. Joseph Schmid has a forthcoming article, "Classical Theism, Arbitrary Creation, and Reason-Based Action", on the argument. Schmid rightly notes that Mullins's version seems to confuse acting on the basis of consideration of X and acting on the basis of X itself. He then attempts to propose a modification of it, based on the idea that intentional actions are dependent on "prior realities", namely, reasons:
1. God’s act of creation is an intentional action (if only analogously so).
2. Intentional actions are dependent on one or more reasons.
3. So, God’s act of creation is dependent on one or more reasons. (1, 2)
4. If DDS is true, then God’s existence is identical to God’s act of creation.
5. If God’s existence is identical to God’s act of creation, then if God’s act of creation is dependent, then God’s existence is dependent. (Leibniz’s Law)
6. So, if DDS is true, then if God’s act of creation is dependent, then God’s existence is dependent. (4, 5)
7. Suppose DDS is true. (Assumption for Conditional Proof)
8. So, if God’s act of creation is dependent, then God’s existence is dependent. (6, 7)
9. God’s existence is not dependent.
10. So, God’s act of creation is not dependent. (8, 9)
11. So, God’s act of creation is not dependent, and God’s act of creation is dependent (on one or more reasons). (10, 3)
12. So, DDS is false. (7–11, Conditional Proof)
13. If DDS is false, then classical theism is false.
14. So, classical theism is false. (12, 13)
This is a very great improvement over Mullins's argument, not least in the fact that it no longer depends on a false account of divine impassibility. There are, however, three points of failure for it.
(1) Divine simplicity does not imply that divine existence is identical to the divine act of creation. It is only taken to do so by a very narrow school of analytic philosophers (and possibly some Cartesians) who try to equate noncomposition with identity. All major proponents of divine simplicity outside these groups (and most proponents are outside these groups) hold that simplicity does not imply that 'divine existence' is logically intersubstitutable with 'divine act of creation' -- they just don't mean the same thing, and therefore you cannot substitute 'divine existence' whenever you have 'divine act of creation', or the reverse.
Even if we did follow the schools that characterize simplicity in terms of identity, there's no obvious reason why we would do so in terms of classical identity, for which Leibniz's Law applies, rather than some other option, like relative identity or modally qualified identity.
(2) While there is more room for controversy on this, intentional actions are not necessarily dependent on reasons as "prior realities"; indeed, they often seem to be only 'dependent on reasons' if we are speaking figuratively. Your reasons don't cause your intentional actions to exist; you do act on reasons, but such reasons are at least often not "prior realities", but simply things you are considering in the acting itself. This is a way reasons differ from motivating needs and cravings. And very often we don't have reasons and then act; we formulate our reasons in the acting itself -- it is in the acting that many of our reasons become our reasons for acting.
What is more, it is not at all clear that all of our intentional actions have any definitely discernible reasons at all. Most people think that you can at least sometimes do things for no identifiable reason. It's always possible to say that there was such a reason despite its being unidentifiable, but it's often the case that even when we are identifying reasons for an action, we are identifying things that make the action one that in context could be reasonable, not anything that we can certainly say the intentional action itself depended on. Likewise, we often talk as if some intentional actions can be their own reason. Thus our own experience of intentionally acting does not really establish that intentional actions are always really dependent on one or more reasons; we would need some stronger foundation. We also can't say how this applies to divine intentional actions, if at all, unless we know why we attribute it to human intentional actions, if we do; some things we attribute to human intentional actions are attributed because they are specifically and distinctively human, not because they are essential to intentional action.. Do we have reason to think that a first cause needs a reason, beyond the sense in which it itself can be a reason, for its action?
But more than this, talk of dependence in discussing intentional acts and reasons seems to be figurative. To say that my intentionally doing something depended on a particular reason R for doing it seems to say no more than that my intentional act was an R-type of intentional action. That is it seems that the reason enters the picture from the fact that in describing what the intentional act is, our description includes the structural or teleological reasons for its being what it is. If I tie my shoes tightly because it's a good idea not to risk falling, the 'because it's a good idea not to risk falling' is not something separable from the intentional action; it more completely specifies its nature. Fully describing the action, of course, does depend on describing the reason, in that it is incomplete if I don't do the latter; but this doesn't imply that the action itself depends on the reason, only that the action is of the kind that has that kind of reason. There is nothing wrong with figuratively saying that God's understanding depends on His existence, for instance; all we mean is that our being able to talk about God's understanding presupposes our being able to talk about His existence. Someone likewise could say, even when accepting Schmid's premise 9, that God's existence 'depends' on His possibility -- lots of classical theists wouldn't, for various reasons, but it wouldn't be a serious problem to say it as a figure of speech. If the dependence in premise 2 is figurative, it is irrelevant to premise 9.
(3) Schmid tries to bolster the argument on these points by arguing that it is a contingent matter whether God takes a particular reason to act, but this really doesn't follow from anything he notes. One could say that all of God's reasons for creating (i.e., as Schmid puts it, considerations that count in favor of creating) are necessary reasons necessarily accepted by God; it's just that 'counting in favor of creating' is not the same as 'requiring that one create' -- God is free to create or not, so anything created is actually contingent rather than necessary.
Schmid attempts to argue that God's reasons cannot be God Himself, on the ground that reasons are considerations that count in favor of something. But this is obviously not going to work in this particular argument. If we assume premise 4 it can only be on the ground that we think that divine simplicity implies that everything in God is identical to everything else. Considerations are acts, and therefore Schmid should conclude, to be consistent with the premises of the argument as he has developed it, that God's considerations are identical to His act of creation. If you're in for a penny, you're in for a pound; you can't use identity in one part of the argument while pretending that it can't be used in another part. Thus one could say, if we assume the identity account, that God's intentional actions have reasons that are identical to those intentional acts; even if that's not true of us, its not being true of us would plausibly be because we are not perfectly simple beings. Certainly it makes a kind of sense that at least at the limit intentional act and its reasons are not separate. While all of this assumes something I think is not right (the identity account), it makes a broader point that I think is important in general, namely, that both Mullins and Schmid in their accounts of the Problem of Arbitrary Creation assume that what is true of human composite beings must be equally true (i.e., stays true without any new qualifications) as we get more and more simple so that it must also be true at the divine limit of perfect simplicity. There is really no reason to prefer this over saying that as things get more and more simple, more and more of the distinctions that we can make in highly composite cases drop out as no longer being applicable, until we reach the divine limit, at which we can only make these distinctions in highly qualified, and sometimes purely figurative, ways.
All in all, though, as I said above, it's a massive improvement over Mullins's argument; Schmid's version of the Arbitrary Creation Problem is not the monster of confusion that Mullins's was.
