So more analysis is still in order, but overall, I don't think these dynamics encourage the formation of big miners.
This is encouraging... it sounded yesterday as if you had almost regretted making this thread and were about to pull your own support from the proposal because of this and now it looks like it might be less of a problem than initially thought.
At the moment I'm confident that the claimed centralization issue does not invalidate the method - whatever effect there might be, it's nullified with a correct parameter choice. I'm not as confident in my ability to convince others of this, but I can try.
Note, though, that superlinear mining rewards is a problem in general, I just don't see how this proposal contributes to it.
- Do the 2 25% miners (or 2 of the 10% miners) have a higher-than-in-current-system incentive to collude?
- Is Menis proposal making it easier for the 2 25% miners to try to drive out small (as in bandwidth) miners by mining disproportionally large blocks?
Both questions boil down to
- Does Menis proposal encourage centralization more than the current system?
Before I go on... am I asking the correct questions?
To be honest, I'm not sure. There are different elements at play here with complex interplay at different timescales. The usual question I'd ask is "do big miners have a superlinear advantage over small miners?". I claim above that the answer is no, but I'm now sure that answering this question alone is sufficient.
Obviously if you can find someone to work a PoC for your proposal, that would be fantastic.
That was one of the purposes of this thread.
If the proposal finds enough support, I'm sure we can crowd-fund a PoC.
FWIW, I'll be happy to contribute to it.