I'm posting some comments I gave to an email correspondent:
1. First I’ll make a preliminary point: open theism is the only theistic libertarian position I think is even coherent. It makes some sense within open theism to say a human agent has the freedom to do otherwise in the actual world, for in open theism, while the past and present are actual, the future is not actual. The future is the realm of possibilities, and our libertarian choices actualize the future by instantiating one possibility rather than another.
Of course, open theism suffers from its own liabilities. The position is heretical. And it shares a common problem with all libertarian schemes: the motivation for libertarianism is to ground human responsibility. But can libertarianism really furnish the degree of personal control which an agent needs to be morally responsible?
2. By contrast, I don’t see how a human agent can be free to do otherwise in the actual world given Arminianism or Molinism. They can be free in possible worlds, but the Arminian or Molinist God only instantiates one possible world to the exclusion of others.
3. I believe that possible agents are free to do otherwise in possible worlds. Indeed, not only to I think they can to otherwise in different possible worlds, but I believe they do otherwise in different possible worlds.
However, I don’t think they’re free to do otherwise in the actual world. And I don’t think possible agents originate their own possibilities (i.e. alternate courses of action).
4. This brings us to the ontology of possible worlds. What is a possible world? What is a possible agent?
Here I’m puzzled by the position of Craig, Plantinga, et al. They treat possible persons as a given. Given possible persons, with determinate character traits, God chooses which world to instantiate in light of what possible agents would do in different possible worlds. He chooses the possible world which achieves his objective.
But it’s unclear to me how Craig, Plantinga, et al. account for the given. How do possible persons subsist, with determinate character traits, such that God’s choice is responsive the free choices?
To me, the only logical way to embed this notion, consistent with libertarianism, is to go the route of Richard Creel. There’s a platonic plenum which is populated by possible agents in possible worlds. This exists independently of the divine nature or will. It’s like a mail order catalogue from which God can make his selection.
But, of course, the notion of a coeternal, self-subsistent plenum, alongside God, is profoundly heretical. It’s also metaphysically bizarre. What is the plenum? Is the plenum a mind-like entity?
5. My own position falls within the divine ideas tradition. Possible agents inhere in the mind of God. A possible agent is a divinely conceivable agent. A possible agent is free to do whatever is conceivable for God.
Conceivability is sometimes criticized because not everything that’s conceivable is possible. But that’s because we’re usually talking about what’s conceivable for human thinkers. And we are shortsighted. What strikes us as conceivable may be incoherent since we can’t keep track of all the variables.
But for God, whatever is conceivable is possible or compossible. God’s concepts are logically self-consistent.
6. I think that in creating the actual world, God selects one of these possible timelines to instantiate.
The question is whether that’s sufficient to ground human responsibility. Of course, that’s a purely intuitive judgment. But, offhand, I don’t see why that would be insufficient to ground human responsibility.
God isn’t making us do something opposed to what we would otherwise have done, if left to our own devices, for there’s no one thing we would have done. Put another way, there’s nothing in particular which we were going to do. It’s not as if, by making this world, God prevented us from doing what we intended to do–had he not “interfered.”
To the contrary, there are any number of things which God imagines us doing. In making this world, God simply actualized one of those forking paths.
Seems to me that this framework does justice to Biblical predestination, but it also provides a different framework for human responsibility.
Possible agents lack the freedom to do otherwise within a given possible world, but that's because their freedom to do otherwise is cashed out by having a plurality of possible worlds which correspond to different courses of action. An ensemble of possible worlds represent their freedom to do otherwise.
But if there's only one actual world, then, of course, there is no otherwise at the concrete level.
The only way to create an analogy between the possible and the actual is if you had more than one actual world representing more than one alternative.
In a sense, that's what Lewis does. However, Lewis doesn't have two different domains: possible and actual. Rather, he collapses the possibilities into multiplicity of actual worlds.