“Basically, all it takes is one node in our network, to prevent a sybil attack. Literally if one node has not been compromised or been a part of some kind like collusion attack, this would mean that operators would have to get together and deploy code that is like, you know, not the correct source code. All it takes is one, because, the fact that any kind of action that could be seen as invalid to invalidate a chain state, is stored directly in chain. So, if you try to read back the state of nodes who perform a sybil attack, it will be read back as an invalid state.
So, if all it takes is one node to keep the cluster online, and then we can rejoin nodes to that one node, or we can join extra peers to that one node to kick consensus off again, because that node will have forked from the other peers. So, that’s also kind of part of how the redownload works as well. Now I’m thinking maybe I should spend some time talking about that, and maybe going over hashgraph as well I think had something similar, but we took it to the next level…So yeah. Wyatt Technical AMA Timestamped link https://youtu.be/aagTpET88qM?t=107
Link to full talk https://youtu.be/w6qYVUk2ssA