Whereas in Byzantine agreement (of which Byteball is an example), then when sufficient witnesses stop responding (or collude to respond towards divergence), then the system can't become unstuck even if the token value declines.
Bad witnesses can be replaced.
With Bitshares' DPoS (also in Steem), the whales can vote in new witnesses at any time, so it can't get stuck unless the whales decide to destroy it.
Same with byteball, only replacement mechanism is different.
No, they can't in practice be replaced. It is not the same. I already refuted that:
Note there is a mechanism for voting in new witnesses, but because it requires a total order so it will in practice never work out because total orders don't exist in nature.
I am sorry if you don't understand the terminology, e.g. the implications of the term "total order". It is indicative of why you are not an expert.
When I discussed this in the Byteball thread, the Byteball author Anthony realized there is a problem and started to talk about avoidance of needing to replace witnesses and how to have a community wide vote and other mechanisms for overcoming the inherent flaw (which means really he will end up needing whales same as for DPoS to avoid chaos so then you are right back to centralization and the failure of Bitshares and Steem again).
And someone wrote upthread that Anthony current controls 87% so we can see once again centralization is employed to avoid the truth that (without centralization) chaos is the only outcome if we don't use PoW.
We don't have decentralization in any blockchain in existence (Monero claims otherwise but I doubt it and the ASICs and mining farms will come if the market cap grows some more). That is a fact. And that is one reason that blockchains are not going mainstream, because the world isn't going to trust some whales to handle the world's blockchain.