It isn't too difficult to formulate a mathematical system where two subsets are completely identical when considered in isolation, but when interacting, one subset operates as if the other doesn't exist, while the other is easily perturbed by any values in the mirror subset. I'm not sure if you can do it without at least one dimensional basis unit that squares to zero, though.
Imagine, if you will, that there are dark-matter people. We have no evidence whatsoever that they exist, any they are constantly annoyed by all the real-matter people that wander between their dark-couches and dark-televisions, and the real-matter bicyclists that pedal through their dark-bedrooms at the worst possible times.
Or did I just invent a crappy layman-author's explanation of where inertia comes from?
Anyway, the way the math works out, it is impossible for a known-universe person to even measure the effects the known-universe has on the dark-universe. The numbers can't come back to the known-universe without an additional input from it, and they would be explainable with another hypothesis that did not include a dark-universe. It way be that all effects we observe to be random are actually perfectly predictable to a dark-universe observer.
Sorry for ruining your dream. :/
I can still dream, though.
Never the less, Verlinde's ideas are more interesting and reasonable than other proposed modification of gravity, but that's just my opinion.