How secure is Ethereum 2.0's consensus?

Editor's Note: This article comes fromBlue Fox NotesEditor's Note: This article comes from

Blue Fox Notes

(ID: lanhubiji), author: Alex Stokeso, compilation: SL, released with authorization.

One of the main changes of Ethereum 2.0 is the change of its consensus mechanism. It moves from PoW to PoS (Casper). What is the difference between PoW and PoS in terms of finality? Will it be more secure than PoW?

The normal operation of the Casper proof-of-stake consensus protocol leads to the generation of finality attributes on the data, and the participants reach a consensus during the participation process. We refer to this data as finality, because once reached, the data cannot be changed without burning more than 1/3 of the pledged tokens that participants deposit in the system. It turns out that this property adds some nice security guarantees that current PoW systems lack.

One way to think about Casper is that validators in the system (similar to miners in PoW) are tasked with voting for blocks in the blockchain that they consider "canonical". Individual validators are rewarded for reaching consensus with other validators (encouraging consensus on one chain), and punished for proven malicious behavior (such as trying to vote for two different chains). Penalties are enforced through secured deposits (staking), staked tokens locked in the system by validators prior to participation.

The security of Ethereum 2.0 stems from the total amount of these deposits and its rules (such as "slash conditions", etc.), which ensure that validators will lose their deposits if they try to attack the consensus. To add some fuel to the margin of safety, an Ethereum 2.0 consensus failure would require more than 1/3 of the active validators on the network to violate the condition for mitigation. Assuming a healthy level of participation in the Ethereum system of 10 million ETH, a successful attack would mean burning 3.3 million ETH, a total value of over $500 million at the time of writing. (Blue Fox Notes: The current figure has exceeded 700 million US dollars, that is to say, according to the current situation, the cost of its attack needs to be at least 700 million US dollars.)

Given the size of this safety margin, the Casper process considers any data normalized to be final, since the cost of undoing this "normalization" is a large amount of self-destruction. The finality of this consensus is in contrast to the Nakamoto consensus guarantee, which is only a probabilistic statement about the security of a given data, based on the proof of work performed on it.