Very nice essay, Roger. You’ve put your finger on a key internal tension within Molinism.
While Molinism is *officially* committed to a libertarian view of creaturely freedom (and thus soft determinists like Ware are *not* Molinists, even if they co-opt the label), such a view of freedom requires that middle knowledge counterfactuals of actual creatures be explanatorily *posterior* to actual creaturely free choices. Thus, if Adam and Eve are free (in the libertarian sense) to eat or not eat the forbidden fruit, then it must not be fixed *independently* of their actual choices that IF they were to be placed in such-and-such circumstances that they would eat the forbidden fruit. For it the truth of that conditional were independently fixed, then they would have no say about whether it is true, and so couldn’t act so as to bring about its falsity. This means that they couldn’t do otherwise than eat the fruit in those circumstances, which in turn means that they weren’t free in a libertarian sense, contrary to hypothesis. Hence, the truth values of middle knowledge counterfactuals must be explanatorily *posterior* to actual creaturely free choices. But this is a huge problem for Molinism because the providential usefulness of middle knowledge is predicated on its being explanatorily *prior* to actual creaturely choices. That’s the only way it can inform God’s creative decree. So Molinism is internally inconsistent. Its alleged reconciliation of creaturely libertarian freedom and meticulous divine providence depends on both affirming and denying that the truth values of middle knowledge counterfactuals are explanatorily *posterior* to actual creaturely free choices.