No, my approach does not require much extra computation.
Could you please point me to some literature on the well-known trade-off? I've read some things about that, but they seemed unpractical. I'd like to know if we're talking about the same thing.
By its very definition activated "Economic clustering" can't open new attack vectors in technical domain. Socioengineering attacks - yes, political attacks - maybe, but technical ones - unbelievable.
Is there a clear definition of how will "Economic clustering" be implemented? if you have a new consensus algorithm, there will be new strengths and new flaws. I'd rather not go into deeper discussion of something that is still not defined.
All this stuff is not systematized and I, unfortunatelly, don't have links ready.
Anyway, my approach does not rely on heavy computation, and it also has the potential of more than a 200% increase: no asymptotes here.
Regarding the double spend every 30hs, I have given all the technical details (which were never a secret, that's just how POW and POS work), so you'll have to admit it's possible: the only problem is acquiring the forging weight (be it leasing or buying/scamming/stealing/whatever).
edit: well, not the ONLY problem, you have to be able to forge a private chain, calculate the probability of actually double spending and have the ability to make the transaction that you want to reverse, still, it's possible.