I pose a problem, offer without endorsing a solution, and then evaluate Paul Manata's objection to the solution.
Suppose a creaturely agent freely performs an action A. He files his tax return, say, by the April 15th deadline. Suppose that the freedom involved is not the compatibilist "freedom of the turnspit" (to borrow Kant's derisive phrase) but the robust freedom that implies both that the agent is the unsourced source of the action and that the agent could have done otherwise. The performance of A makes true a number of contingent propositions, all of them known by God in his omniscience. Now if S knows that p, and p is contingent, then S's knowing that p is an accidental (as opposed to essential) property of S. If God is omniscient, then he knows every (non-indexical) truth, including every contingent truth. It seems to follow that God has at least as many accidental properties as there are contingent truths. Surely some of these are not properties with which God could be identical, as the simplicity doctrine requires.
Consider the property of knowing that Tom freely files his tax return on April 14th, 2014. Assuming that Tom actually performs the action in question, this property is an intrinsic property contingently had by God. (A property can be intrinsic without being accidental.) If God were identical to this property, then he could not be a se. For if God were identical to the property, then God would be dependent on something -- Tom's libertarianly free action -- that is external to God and beyond his control. Now anything that compromises the divine aseity will compromise the divine simplicity, the latter being an entailment of the former. So it seems that an omniscient God cannot be simple if there are free creaturely agents.
The problem is expressible as an aporetic triad:
1. Every free agent is a libertarianly-free (L-free) agent.
2. God is ontologically simple (where simplicity is an entailment of aseity and vice versa).
3. There are contingent items of divine all-knowledge that do not (wholly) depend on divine creation, but do (partially) depend on creaturely freedom.
Each limb of the above triad has a strong, though not irresistible, claim on a classical theist's acceptance. As for (1), if God is L-free, as he must be on classical theism, then it is reasonable to maintain that every free agent is L-free. For if 'could have done otherwise' is an essential ingredient in the analysis of 'Agent A freely performs action X,' then it is highly plausible to maintain that this is so whether the agent is God or Socrates. Otherwise, 'free' will means something different in the two cases. Furthermore, if man is made in the image and likeness of God, then surely part of what this means is that man is a spiritual being who is libertarianly free just as God is. If a man is a deterministic system, then one wonders in what sense man is in the image of God.
As for (2), some reasons were given earlier for thinking that a theism that understands itself must uphold God's ontological simplicity inasmuch as it is implied by the divine aseity.
An example of (3) is Oswald's shooting of Kennedy. The act was freely performed by Oswald, and the proposition that records it is a contingent truth known by God in his omniscience.
But although each of (1)-(3) is plausibly maintained and is typically maintained by theists who uphold the doctrine of divine simplicity (DDS), they cannot all be true. Therein resides the problem. Any two limbs imply the negation of the third. Thus: (1) & (3) --> ~(2); (1) & (2) --> ~(3); (2) & (3) --> ~(1).
To illustrate, let us consider how (1) and (3), taken together, entail the negation of (2). Being omniscient, God knows that Oswald freely chose to kill Kennedy. But Oswald's L-freedom precludes us from saying that God's knowledge of this contingent fact depends solely on the divine will. For it also depends on Oswald's L-free authorship of his evil deed, an authorship that God cannot prevent or override once he has created L-free agents. But this is inconsistent with the divine aseity. For to say that God is a se is to say that God is not dependent on anything distinct from himself for his existence or intrinsic properties. But God has the property of being such that he knows that Oswald freely chose to kill Kennedy, and his having this property depends on something outside of God's control, namely, Oswald's L-free choice. In this way the divine aseity is compromised, and with it the divine simplicity.
It seems, then, that our aporetic triad is an inconsistent triad. The problem it represents can be solved by denying either (1) or (2) or (3). Since (3) cannot be plausibly denied, this leaves (1) and (2). Some will deny the divine simplicity. But an upholder of the divine simplicity has the option of denying (1) and maintaining that, while God is L-free, creaturely agents are free only in a compatibilist sense. If creaturely agents are C-free, but not L-free, then Oswald could not have done otherwise, and it is possible for the upholder of divine simplicity to say that that Oswald's C-free choice is no more a threat to the divine aseity than the fact that God knows the contingent truth that creaturely agents exist. The latter is not a threat to the divine aseity because the existence of creaturely agents derives from God in a way that Oswald's L-free choice does not derive from God.
The proposal, then, is that we abandon (1) and maintain instead that only God is L-free, creatures being all of them C-free. And this despite the reasons adduced for accepting (1), reasons that are admittedly not absolutely compelling. But Paul Manata, in an e-mail, raises an objection to the proposed solution:
I was wondering what you think about this argument that such a solution might not be possible. It goes like this:
Libertarian free will = Incompatibilism + someone is free (does a free action)
Compatibilism = determinism is true in some world w, and someone is free (does a free action) in w.
Incompatibilism = there does not exist a world, w, where determinism is true in w and someone is free (does a free act) in w.
With this quick set up, we can see that compatibilism and incompatibilism contradict each other (the former is scoped by '<>' and the later scoped by '~<>').Thus, to affirm both <>(S is free in some w and determinism is true in w) and ~<>(S is free in some w and determinism is true in w) is not possible. But that is what the solution affirms, i.e., it affirms incompatibilism by affirming that God has LFW and it affirms compatibilism by affirming we have compatibilism freedom.This was quick and there's more to say, but that's the gist of the idea. Thoughts?
An agent X is libertarianly free =df X is the agent-cause of some of its actions.