The OP (me) explicitly pointed out that self-referential systems can work, for a while. The problem is that there's no way to resolve hard forks. You (and Vita

The OP (me) explicitly pointed out that self-referential systems can work, for a while. The problem is that there's no way to resolve hard forks. You (and Vitalik, in that "Nothing At Stake" paper) seem to assume there will never be a hard fork. That's proven wrong by history; hard forks are common. In fact, in the ETC/ETH fork, it was _ETH_ that got a major code change. PoS, by definition, does not provide any way to resolve hard forks that split "stake". After the fork, both chains can appear perfectly consistent with a majority of stakers (in that split) voting "remain".