This doesn't apply to my coin under development, because I eliminate mining-for-profit, but I don't eliminate mining
Is the entire thing still hinged on the fact that it fails if the algo becomes optimized by over 2x speedup on specialized hardware to create identical pool centralization? Even if you believe it's possible to create such a thing, the fact that you need to launch the coin and have it exist for years before the answer is readily apparent makes it a very unattractive target for adoption, because at any time the fundamentals can just completely disappear when some guy like Wolf slightly optimizes the hashing function...
In another thread, both Smooth and I seemed to feel this more traditional PoW-mail, send your own transactions approach would be the Occam's Razor solution for Satoshi and already would have been explored if it was viable. Instead, Bitcoin went the more Rube Goldberg approach because the other methods weren't deemed feasible.