I think part of the answer here might have to do with the difference between epistemological possibility and ontological possibility, but I'm not an expert on the difference -- maybe someone can help me out. As I understand it, epistemological possibility is depicted in statements like "for all I know ___ exists or ___ doesn't exist". But ontological possibility identifies whether it is logically consistent for ___ to exist. Ontological impossibility requires something like logical incoherence, and I don't think that's been shown here regarding MGB's existence.
The difference is that metaphysical possibility (or ontological possibility, as you say), simply refers to possibility. It's what we're talking about when we say things like, "the Nazi's could have won WW2". In this way metaphysical possibility refers to some way things actually could be, or could have been. Epistemic possibility isn't actually a sort of possibility, rather it refers to a state of unknowing. To say that something is epistemically possible is simply to say that, for all we know, it is the case. We sometimes use "could" in this way to denote uncertainty, e.g. "there could be extra-terrestrial life, who knows".
Metaphysical possibility doesn't refer to logical possibility or coherence, that's something else entirely. Logical possibility simply refers to consistency. There are a lot of things that are consistent but impossible (note that both atheism and theism are both consistent, but only one could be true).
Regarding my question about necessity, is it logically possible for a contingent, non-necessary being to exist if that entails that a necessary being can't exist?
Yup, it's possible. When we say God is a necessary being, we're saying that by his nature he has the
property of existing necessarily. Similarly, we would say that a triangle by its nature has three sides. But, there have to actually be triangles for there to be three sided triangles. It doesn't make any sense to talk about non-existing things having properties. Likewise, God has to actually exist if he exists necessarily. And, whether or not he exists is exactly the question being debated.
As such, defining God as a necessarily existing being only entails that:
if God exists,
then God exists necessarily. Similarly,
if there is a triangle,
then it has three sides.
That doesn't seem like a coherent solution. So, perhaps it is being implied that either the knowno is a necessary being, or that there is a quasi-MGB that knowno does know. Is that where this is going?
Hopefully I've explained clearly enough why it's not incoherent. No one would want to say that a Knowno necessarily exists or that there is a quasi-MGB (especially not the atheist!).
Furthermore, I'm not sure how you can state that it is equally plausible in light of the other arguments for theism. I think Craig's position is that while these arguments are independent, they work together to increase the warrant for each other, thus reducing the charge of being ad hoc. I've never heard other arguments for knownos.
But there are also some very strong arguments against the existence, and even the possibility, of God. Think of the problem of evil, divine hiddenness, poor design and so on.
As a side question what is the understanding behind "know" in Inwagen's argument? Does it entail the usual "justified/warranted true belief" idea? It seems like it would have to, because otherwise the knowno is simply wrong if MGB exists and they aren't logically incompatible. I've probably answered my own question, but there it is anyway. I just want to make sure we're all using the same terms here.
Yes, I think that's exactly what Van Inwagen wants to say. It's possible for a Knowno to exist, and if a Knowno exists then God does not (since knowledge requires true belief).