Does $VP \neq VNP$ imply $P \neq NP$?
In his GCT papers Mulmuley first starts with the #P/NC question. And then gets into setting up a roadmap for how to show obstructions ( some sort of Weyl modules I gathered, if they can be constructed) could potentially rule out embedding of a class variety into another.
Is it clear, and almost obvious how that chain of argument ( if and when fleshed out) will automatically extend for P/NP separation? I am not a TCS expert - so I wonder if I am missing something obvious.