Fault count assumptions are not appropriate for public blockchains, because public blockchains exist in an oligopolistic setting, where a very small number of miners or coin holders (or reputation holders, in more exotic architectures) control a vast majority of the weight in the consensus.
1. The cartel cannot hide the fact that the censored validators are missing.
2. The cartel had to be punished whenever validators appeared to be missing.
3. It was therefore also necessary to penalize validators who go offline, because it could not be clear to the protocol whether they were being censored, or whether they were offline of their own accord.
consider a game in which Bitcoin miners compete for a reward of each solved puzzle in a sequence of them.
model it as a sequential game with imperfect information, in which miners have to choose whether or not to report their success.
show that the game has a multiplicity of equilibria and we analyze the parameter constellations for each of them.
In particular, the minimum requirement to find it optimal not to report is decreasing with the number of miners who are not reporting, and increasing the heterogeneity among players reduces the likelihood that they choose not to report
And so the equilibrium is that the system gets "captured" by one "political party" that bribes everyone and engages in some combination of parasite-helping-the-host-survive and rent extraction, kind of like dictatorial governments
EOS's scalability is NOT because of DPOS or anything similar; its claimed scalability comes entirely from the fact that it requires each node to have a much higher computational capacity, making it impossible for anyone but large businesses to run full nodes. We could do that too, but won't because it's contrary to the goals of decentralization.