Good thread!
A few comments below:
A few comments below:
There are two other anti-centralization techniques that the post misses. The first is to have forgiving protocol rules that do not require nodes to have especially high performance. Make sure even 70% uptime validation is net-profitable, don't have ultra-fast block times, etc.
The second is to do sharding, so the amount of data you have to process scales linearly with how many coins you are staking, at least up until the maximum where if you have a large fraction of staked coins you are processing everything. This cuts down economies of scale.
I disagree with the idea that in the long run cryptography will fully supplant cryotoeconomics! Particularly you cannot ZK prove that you did not sign a conflicting message, and so consensus finality will only ever be cryotoeconomic.
However, there are lots of good things ZKPs *can* do. Data availability checks arxiv.org/abs/1809.09044 are a great example.
And once you have data availability checks, all else can be done with fraud proofs, even in a sharded network.
I also think research around automatically identifying who was at fault for all classes of 51% attacks, including censorship, is important.
