I propose that we treat individual possible worlds or scenarios as (metaphysically and epistemically) fundamental. Doing so has interesting implications in either case. First, it grounds an attractive and plausible theory of ontological deflationism. On this view, disputes about what exists are only substantive if they involve carving up the space of possible worlds in different ways. If metaphysicians agree about which world is actual, but disagree about its constituents (e.g. whether it contains tables or merely "particles arranged tablewise"), their apparent disagreement is empty. It is whole worlds that are fundamental; we might break them down in any number of ways, and it is pointless to argue that any one of these is the One True Ontology. (See the linked post for further explanation.) Second, I think the epistemic fundamentality of possible worlds has important implications for meta-modal conceivability. But this will take a bit more explanation...
Let us begin with Yablo. In ‘Is Conceivability a Guide to Possibility?’, he isolates the particular kind of conceivability that best serves as evidence of metaphysical possibility. Note that any number of necessary falsehoods might seem conceivable in some sense. Consider the statement (P): "There is a greatest prime number." Mathematicians can easily prove the impossibility of P, yet you might have thought it conceivable in some sense. For example, you might not be certain that it isn't actually true. (As humble fallibilists, we might even be inclined to grant that absolutely anything is "conceivable" in this weak sense!)
However, Yablo suggests that it is whole scenarios (epistemically possible worlds) that are fundamental for modally-relevant conceivability claims. Used in this sense, the "conceivability" of a statement X consists in there being a scenario which we take to verify X. (That is, we endorse the indicative conditional "if scenario V is actual, then X is true", or we judge that the conditional probability of X given V is near 1.) This allows us to deny that the aforementioned P is conceivable in the relevant sense, since there is no scenario which we take to verify the claim. There is no way the world might turn out, such that if things did turn out that way, there would be a greatest prime number.
At best, we might imagine a scenario in which mathematicians report discovering such a "fact" -- but there is nothing in the given description to rule out this being a scenario in which the mathematicians are simply deluded. So even the mathematically ignorant should not take this scenario to verify P. Instead, they should suspend judgment; they are not sure whether P is true in the scenario or not. That is, they are not sure whether P is conceivable. Hence, their imaginings provide no evidence about the possibility of P -- which is just as it should be. Yablo's account seems spot on. (See also Chalmers' discussion of positive conceivability in section 2 of Does Conceivability Entail Possibility?)
Yablo works down from scenarios to the first-order claims that they verify. He thus treats worlds as fundamental, at least in this respect. But he could go further. I think we should also work upwards, from scenarios to the meta-modal claims that are true of all of them. Let me explain. First, some background context:
Yablo objects to modal rationalism -- or the claim that ideal conceivability entails (primary) possibility -- on grounds of meta-modal conceivability. Even if the modal rationalist is right about the scope of possibility, there are (Yablo claims) coherently conceivable alternatives. For example, it's conceivable that a necessary being exists, and the negation of this is likewise conceivable. But they cannot both be possible, for by S5, the possibility of a modal claim entails its actual truth! Conflicting meta-modal possibilities would thus entail actual contradictions. So there cannot be conflicting meta-modal possibilities. The conceivability of conflicting meta-modal claims thus provides a counterexample to the thesis that conceivability entails possibility.
However, I want to suggest that meta-modal claims are, in a sense, similar to mathematical claims like P above. Considered in isolation, you might mistakenly consider them to be conceivable, but this error is remedied by recalling the fundamentality of worlds (or scenarios). First-order claims require us to work down from the total scenario of which they are part. Similarly, I suggest, meta-modal claims require us to work up from the individual scenarios that comprise epistemic modal space. By treating scenarios as fundamental, we close the gap between epistemically possible necessity and epistemic necessity tout court. (S5 applied to the intersection of epistemic and metaphysical modal space?)
On this proposal, for a necessary being to be coherently conceivable, it must be the case that the being exists in every scenario one can coherently conceive of. The failure of the latter condition suffices to render the necessary being inconceivable. (Assume I can conceive of a Godless world. That possible world doesn't disappear when I turn to the question of whether God might necessarily exist. It can rightly influence my modal reasoning. In particular, it rightly precludes the possibility of God existing in every possible world, for I can see all along that he doesn't exist in that one.) More generally: there are not multiple conceivable modal spaces, for that would require, impossibly, that there be more than one maximal space of individually coherent scenarios.
This view effectively rules out from the start the challenge from meta-modal conceivability posed to modal rationalism. One might thereby complain that it is question-begging. But I think the picture I've presented is independently attractive, and so might be better described as disarming the challenge. At the very least, it may be employed defensively by the modal rationalist to explain why they are not troubled by the meta-modal arguments, even if the challenger remains unconvinced. What do you think?