As far as I can tell the reason Hashgraph is not permissionless and therefore not centralized is because of possible sybil attacks: if it wasn't permissioned anyone could spin up nodes and look like a large portion of the network.
If that is true then I assume the simplest way to solve that problem of to keep the network permissioned. But My thinking is, why must that permission come from a centralized authority?
Surely the network itself could come to consensus on who is allowed in by many possible means. As an example, the simplest means to achieve this end would be to just start the network with a million whitelisted public keys and no others are allowed in ever. That gives the network a short life span but technically it would be permissioned and decentralized while it lived. There are more sustainable solutions too.
In short, if the members of the network all shared the same whitelist of other members, the network is permissioned, but not centralized.
Hedera's solution is different, keep the power centralized in the governance council but make it appear to be decentralised by looping in proof of stake into the consensus algorithm. (Of course, the argument that the governance council acts as the devs do for Bitcoin is perhaps a valid one, I don't really know).
In any event, I like the speed at which Hashgraph operates but not its centralization. And it seems to me quite simple to make it decentralised (though still permissioned, which has it's own drawbacks).
Am I missing something here? Am I oversimplifying it?
No comments:
Post a Comment