> Two different metrics would have to describe two different, disjoint surfaces, which would have no connection to each other, which would mean the second one, being unobservable, would be scraped right off the model by Occam's razor.

The theory states that the two surfaces are not disjoint:

> What is important to note however, is that the two field equations are coupled, i.e. a mass always creates a positive curvature in spacetime according to its own metric (where the mass appears visible), and it also always induces a negative curvature in the conjugate metric (where the mass appears invisible).

I do find it confusing why this would be called two metrics, however, since it seems to be describing a unified spacetime "surface".

As for your second critique:

> One spacetime can only have one curvature at any given ponit. Considering the simpler example of a 2-d surface should make this intuitively obvious--the same surface can't have two different curvatures at the same point.

But if the surface has two sides, can it not? Consider the classic "gravity well" demonstration of a heavy ball placed on an elastic membrane. It forms a concave surface. Now turn your head upside down and look at the membrane from underneath. You'll see a convex surface. Same membrane, different curvatures.

Anyhow, I'm certainly not equipped to analyse this theory, but I do like that it attempts to make falsifiable predictions. I guess it's an attempt to explain dark energy? It would appear to leave problems of dark matter unaddressed, however.