Is it hack-proof?
Malicious actions on Ethereum’s proposed PoS network are discouraged due to the staking of tokens. With validators required to lock up a substantial amount of ETH as their stake, they have no incentive to undermine the network. Doing so would see them lose their entire deposit.
However, with a 51% attack seen as a real risk for PoW cryptocurrencies, how would PoS Ethereum stand up to this risk? Ethereum states that the most basic form of 51% attack against Casper is a "finality reversion", in which validators that have already finalised block A then finalise a competing block A, creating a split of the blockchain.
If this were to occur, it would be up to the community to decide which one of the branches to take forward, ignoring the other. Deciding which branch to focus on could be made on a "whichever branch was finalised first is the real one" basis, which could be agreed upon via communication over social media, between corporations or through backchannels. Alternatively, a free-market approach could be taken, which would rely on market consensus to decide on which branch is the most valuable. In this case, both branches would be traded on exchanges and the like, until one branch established itself as more valuable than the other. If a 51% attack did actually occur, community consensus would probably arise through a combination of approaches.
Another potential attack is liveness denial, whereby 34% or more of the validators refuse to finalise any more blocks. While the chain would continue growing, such an attack would result in greatly decreased security.
There are a couple of actions for responding to this:
- Include an automatic feature in the protocol to rotate the validator set.
- Use a hard fork to add new validators and delete the attackers’ balances.
Finally, a censorship attack is where 34% or more of the validators refuse to finalise blocks that contain certain kinds of transactions. If the attacker has 34-67% of the stake, one defence could involve programming validators to refuse to finalise or build on blocks that they believe are censoring transactions.
However, if the attacker has more than 67% of the stake, they can block whatever transactions they want and simply refuse to build on blocks that contain those transactions. There are a couple of key defence mechanisms that can provide protection in this scenario:
- As Ethereum is Turing-complete, it is already naturally resistant to a certain level of censorship.
- Introducing an "active fork choice rule", where nodes can work out whether a given chain is valid by trying to interact with it and verifying that it is not trying to censor them.
