Perhaps one might complain that it isn't clear how God possesses middle knowledge. I would say that God is omniscient, and as such, knows all facts, including counterfactuals. To be honest, I'm not sure how God knows them either. But that's not devastating. After all, one doesn't need to know how God exercises omnipotence in order to believe that he is omnipotent? He's God, and God just is omnipotent and omniscient by his nature. If one wishes to deny counterfactuals have some epistemic grounding, they need to produce a whole theory of grounding to show how facts are grounded in such a way that excludes counterfactuals. As William Lane Craig has noted, we are still waiting for someone to produce a plausible theory that succeeds at showing this.
This, I think, would be quite the right response if the 'how' of the grounding objection were a 'how' of means or mechanism. There is no means or mechanism for omniscience or omnipotence; unless you wish to stretch the words by saying God's own nature is the means or mechanism. Beyond that, one can simply give the reasons for thinking that God is omniscience or omnipotent. But that's not really what the grounding objection is asking; the grounding objection is, effectively, an objection to the inclusion of counterfactuals of freedom among facts to be known in the first place. There are three grounds for including something among the facts that are known to be true:
(1) It actually exists, and is known because of that.
(2) Its causes are known, and they require it.
(3) Its effects are known, and they require it.
But (3) does not come into play in the case of future contingents; (2) does not come into play because we are dealing with cases of free choice (only a compatibilist could appeal to (2); and (1) does not come into play because we are dealing with counterfactuals. Hence we appear to have no reason to regard counterfactuals of freedom as facts (rather than, e.g., mere possibilities). Now, it appears that, if none of (1)-(3) are met by counterfactuals of freedom, all that is needed to know them, to the extent that there is anything to know, is natural knowledge. The Molinist needs to find a fourth option that shows us that there is actually work for middle knowledge to do. So the point made by Plantinga and Craig about the lack of a theory of grounding is a red herring. (And, in any case, even if it were relevant, there is no need for one: we don't need an account of grounding, we just need to know the purported grounds, and there is nothing incoherent or irrational about asking for the latter even if there is a lot about grounding we still don't know. There are many parallels: we don't need an account of belief to ask whether something is believed; we don't need an account of justification to ask why someone thinks a claim is justified. Clearly a full answer would have to get into details, but the fact that it's unclear what would be required for a full answer doesn't justify giving no answer.)
So this is where the analogy with omniscience and omnipotence breaks down: in those cases we eventually reach a point where we can't do more than give our reasons for attributing omniscience and omnipotence to God. But as long as they evade the grounding objection, Molinists give no reasons for thinking that counterfactuals of freedom are the sort of thing that would require middle knowledge. So Molinists must answer the grounding objection.
