Dissecting Molinism

I'm going to post an email exchange I had with a friend on Molinism. His comments are indented. 

1) I find Molinism is self-contradictory. Molinist assumptions generate an internal dilemma for Molinism. Let's take two of Craig's claims:

i) there is an intrinsically possible world in which Peter freely affirms Christ in precisely the same circumstances in which he in fact denied him

ii) logically prior to His creative decree, God knew that if Peter were in circumstances C, he would freely deny Christ three times…given the counterfactual truth that if Peter were in precisely those circumstances he would freely deny Christ, then the possible world in which Peter freely affirms Christ in those circumstances is not feasible for God. 

Seems to me that Craig is making two diametrically opposing claims:

i) Peter denies Christ under those circumstances

ii) Peter both affirms and denies Christ under those circumstances.

Craig seems to say that Peter only does one of those while saying in the same breath that Peter does both! 

2) Likewise, I don't see where the "counterfactual truth" is coming from. If, as Craig admits, there's a possible world for each alternative, then what privileges Peter denying Christ as the "counterfactual truth"? 

3) Notice that at this stage we're still talking about would-statements rather than will-statements. Obviously, if God instantiates one of those scenarios, then there's only one choice which Peter will make in the actual world. 

But is it the case that Peter will deny Christ because God instantiates that choice, or does God instantiate that choice because Peter will deny Christ? Craig's position commits him to the latter. 

But how does he get from plural would-choices to the singular will-choice? There's more to it than the shift from possible to actual, for there's more than one would-choice that's initially available to God. 

Yes, Craig says the "counterfactual truth" moots one of those possibilities, but I don't see how he derives that from his set-up. 

4) Which brings me to the next point. I think rewinding the tape is a good way to illustrate libertarian freedom. Assuming the agent could do otherwise given the same antecedent conditions, then if God put him in the same situation 20 times, would we expect him to make the same choice each time? Isn't that implausible? If he's truly free to do otherwise in the same situation, and he's put in the same situation repeatedly, why would he invariably make the very same choice? Doesn't that uniformity indicate some factor that's swaying the outcome? 

Seems to me that if he has the freedom to do otherwise whenever he's put in that situation, then if God rewound the tape 20 times, there shouldn't be a consistent pattern. He might do the same thing three times in a row, then do something different, then alternate. 

I think Craig's definition of freedom clashes with his counterfactual truth. Pick one or the other, but I don't see how he can insist on both. 

At the very least, your statement of the claims elides the distinction between would and could. This distinction is crucial. Would-claims are about actual truth, whereas could-claims are about possible truth.
Would-claims are about the actual world. They indicate brute facts true from eternity. They could be otherwise, but they are what they are. They are just about the only element of the actual world that God doesn't actualize. They are actualized and 'in place' from eternity.
Craig is not claiming that Peter "both affirms and denies" Christ under those circumstances. Craig claims that it is possible that Peter affirms Christ under those circumstances, and Craig claims that it is possible that Peter denies Christ under those circumstances. But those are mere could-claims. They don't tell us (and aren't intended to tell us) what Peter would actually do if he were created and placed in C. That's what claim (i) tells us. Claim (i) is better described as the claim, not that "Peter denies Christ under those circumstances," but that Peter *would* deny Christ under those circumstances.

I don't think the could/would, possible/actual dichotomy is metaphysically sustainable at this hypothetical stage. And I don't think we can adjudicate the issue before we address the preliminary question of what possible worlds are. What's our ontology of possible worlds? Seems to me that, in principle, there are basically four models:

i) Fictionalism/nominalism

This denies the existence/reality of possible worlds. But aside from the general fact that fictionalism/nominalism fall prey to tenacious objections, this doesn't seem to be a viable option for Molinism, which is heavily invested in possible worlds. At the very least, it would require a drastic revision of Molinism. Can Molinism survive an eliminativist ontology of possible worlds? 

ii) Concrete entities

According to David Lewis, possible worlds are alternate spatiotemporal scenarios. If you're an atheist, that's probably your best bet to ground modal distinctions. But it has no appeal to a Christian. 

iii) Divine ideas

In the Augustinian/Leibnizian framework, possible worlds are God's exemplary ideas. Compete conceptual alternate scenarios. And if God is timeless and spaceless, so are possible worlds. 

On a related note, Peter Geach defines a possible world as a nominalization (in the grammatical sense) of divine omnipotence. In that respect, a possible world is a description of what God can do rather than what human agents can do.  William Young, "What is Truth," defines a similar model. 

I think this is the best model of possible worlds, and it dovetails with Calvinism. But by the same token, it's a poor candidate for freewill theists, inasmuch as God is the ultimate source of possible worlds on this model. They totally depend on him, whereas freewill theism demands a degree of autonomy for the "counterfactuals of freedom."

iv) Abstract objects

This is a variant of Platonic realism, like Richard Creel's plenum. On this view, possible worlds would be complete, timeless, spaceless alternate scenarios. 

I think that's the best bet for freewill theism. But it erases the would/could dichotomy. 

Maybe it's implausible. But as long as the conditions for libertarian freedom are met (source and leeway), where's the problem? Libertarian freedom isn't some kind of empirical inference from trial runs or something! We don't have access to that.

I think that's too facile. If a dealer assures us that the deck is randomly shuffled, yet each time he shuffles the deck, the cards have the same sequence, then we'd be forced to conclude that the deck is stacked. Odds are, an indeterminate outcome will occasionally parallel a determinate outcome. If, however, the outcome is always the same, then that's not coincidental. 

I think Molinists would want an argument for thinking that. What metaphysically precludes there being brute facts about what creatures would choose? Many critics of Molinism have raised this issue, and they make interesting points. But Molinists have directly engaged them on these issues. So we have to see more than just a claim that it's not metaphysically sustainable. Why can't there be these facts?

You're substituting "fact" for "actual," as if those are synonymous. 

And I'm not sure what you mean by "this hypothetical stage"

At this stage we're referring to unexemplified possibilities, where what's "actual" typically stands in contrast to what's possible, viz. physical, instantiated in time and space. 

What does it matter what possible worlds are, or if there are any possible worlds in some metaphysically robust sense, as long as those claims are true?

It matters when you attempt to distinguish between what an agent would do and could do. 

Molinists are unique in holding to the contingent truth of CCFs. But these are claims about the actual world.

That's ambiguous. You mean about the actual world in the conditional, roundabout sense that if God instantiated a counterfactual, that would refer to something in the actual world?

How is Molinism "heavily invested in possible worlds," where the latter are entities in some metaphysically robust sense? 

Is it your contention that prominent Molinists like Plantinga and Almeida aren't heavily invested in a metaphysically robust conception of possible worlds?

All Molinism needs are facts about possibilities, and facts about subjunctive truth.

With all due respect, isn't that hopelessly superficial? "Facts about possibilities" don't subsist in a metaphysical vacuum. There needs to be some metaphysical framework in which they inhere. 

Molinism can survive an eliminativist ontology of possible worlds just in case there doesn't need to be such worlds in order for claims about possibilities to be true.

"Just in case" is evasive. Do you or don't you think that's tenable?

PWs are necessary but CCFs are contingent. Holding that "God is the ultimate source of possible worlds" because PWs are "divine ideas" doesn't tell us anything at all about CCFs.

Since I explicitly said this is a poor model for freewill theists, the fact that "doesn't tell us anything at all about CCFs" is what we'd expect if it's a poor model for freewill theists. Remember that I set this in contrast to a more promising model for freewill theism. So it's strange that you criticize my statement by raising an objection that's consistent with my statement.

Some people think that possible worlds are just giant propositions

Which pushes the issue back a step. Propositions about what? Presumably, about alternate scenarios. 

Why would the fact that possible worlds are "complete, timeless, spaceless alternate scenarios" somehow "erase the would/could dichotomy"? I'm not getting the connection here.

If possible worlds are abstract objects, then these don't merely represent what an agent could do, but what he did or will do in that world. Like an abstract actual infinite, we're dealing with complete world histories. Complete alternate timelines. In that event, there's more than one thing the agent would do.  

In the actual world -- that is, in the only possible world that is in fact actualized -- a particular set of claims is true, including would-claims. If these contingent claims about the actual world are indistinguishable from possibly true claims, on the grounds that possible worlds are "complete, timeless, spaceless alternate scenarios," then there can be no contingency at all.

There's contingency in what God chooses to instantiate.

I don't see why it's "too facile," when all you've done is repeat the claim I've disputed: that epistemological concepts -- of what choice we would "expect him to make," of the uniformity 'indicating' some outside factor of influence, that there are 'odds' that are relevant here -- have any bearing on the fact of libertarian free will. 

No, I haven't simply repeated my original claim. I tried to advance the argument by pointing out that if a supposedly indeterminate outcome is consistently indistinguishable from a determinate outcome, then we're dealing with determinism rather than libertarianism.

You don't have an argument that the counterfactual must be necessary.

It was never my intent to argue that the counterfactual must be necessary. Rather, I've challenged the postulate that there's only one available counterfactual for God to instantiate.