[captainmuon]

I mean that the underlying "real" theory would be for example just a manifold that obeys certain equations. There is no mention of particles, space, or time. Then you'd have a procedure to get from this manifold to the observables of this universe. One point in the manifold does not map to one point in physical space. Rather, for example, the state of the manifold at many different points collectively determines a certain physical thing in spacetime.

So the fundamental description of the world would be some strange configuration space, which is neither classical nor quantum. And the physical world would sit "on top".

If the universe was like this, and you'd try to form a hidden variables QM theory using the wrong, i.e. physical variables x, p, t, you'd automatically get either nonlocality + infinitely many variables, or very convoluted dynamics (like Bohmian mechanics). Which is exactly what you see.

Another reason for me to believe that a description below QM and spacetime is possible, is because QM looks like a statistical limit of something. I find it very odd to identify that QM is the extension of probability theory to imaginary numbers, and then not ask: Probability theory of what?