Bayesians care a lot about unfalsifiability, a theory can only gain probability mass by assigning low probabilities to some outcomes (if you don't believe me then go read Eliezer's technical explanation of technical explanation).

Nitpicking; I meant that falsifiability-in-practice-as-such-the-way-most-people-use-the-word is not a necessary precondition for determining which hypotheses to pay attention to. Apparently unfalsifiable hypotheses (which are nonetheless probably actually falsifiable with enough computing power) like the existence of a creator God are thus fair game for Bayesians, and pointing out their apparent unfalsifiability isn't scoring a point for the atheists.

"Anything not made from subatomic particles" is a poor definition of the supernatural, since it leaves is irrationally prejudiced against the idea that subatomic particles could be made out of something else, which is a perfectly reasonable hypothesis (currently one with no evidence for it, but we still shouldn't be prejudiced against it).

Right, but you have to use a poor ontology in order to get a concept that even looks like supernatural in the first place... this is an argument against using the word supernatural at all. God is just not supernatural if you are using the right ontology. I don't know what an ontologically fundamental state would look like (when I think of people who believe in the supernatural that does not seem to describe their beliefs at all), and I don't see how that conception is at all relevant to gods, witches, or ghosts. We can follow that digression, as I'm really curious as to what people are trying to explain when they talk about supernaturalism as belief in ontologically fundamental mental states, but it doesn't seem relevant to the OP.

Hypothesis: Most if not all of this epistemological support of which you speak is bad philosophy, possibly based on the mind projection fallacy, which could just as easily have been constructed to defend witches or ghosts if someone had had enough reason to do so.

Most? Yes, of course yes. To a first approximation, everyone everywhere always has always been wrong about everything, including all of atheism and science. But all? Not even close.

Here's a basic argument for a somewhat vague Creator God: the universe exists. Things that exist tend to have causes. Powerful things like superintelligences or transcendent uploads are good at causing things. This universe might have been caused by one of those really powerful things.

That we feel better when we call those powerful things 'superintelligences' instead of 'gods' just says something about our choice of ontology, not about the righteousness of our epistemology.