With the development of the space of decentralized finance in the Ethereum blockchain, the problem of front-running has arisen, when miners, using the opportunity available to them to organize transactions in blocks in any way, outrun the transactions of ordinary users and benefit from them.
This concept has recently taken a step forward with a proposal to launch a modified Ethereum client that would allow miners to get paid to change the history of the last few blocks by reorganizing the blockchain. Thus, even confirmed trades can be fronted if the potential profit is high enough. Ethereum founder Vitalik Buterin, together with Paradigm fund partner Georgios Constantopoulos, explained how the problem will be mitigated with the transition to the second version of the protocol.
Reorganizations in Ethereum 1.0.
Ethereum is now using the Nakamoto consensus, which has the "longest chain rule". It lies in the fact that when there are two similar blockchains, the client chooses the chain with the greatest overall complexity.
Short reorganizations are not a threat and occur constantly due to delays. If two miners find a block at the same time and transmit it to the network, part of the network will see one version of the blockchain first, and part of the network will see another. Usually, to resolve such a situation, it is enough for the third miner to add his block to one of the chains, which is then considered the main one. In unfavorable conditions, reorganizations of 2-5 blocks can occur, and reorganizations of six or more blocks are almost always the result of a critical network failure, client bugs or attacks.
If the incentive for reorganization is high enough, miners may try to add blocks to a competing chain, rather than to the established or "canonical" chain. This would be possible with the aforementioned modified client, with the help of which several miners could pool their resources to help one user for a fee. Researchers call these miners "short-sighted." Their actions may be justified in the short term, but ultimately undermine Ethereum's credibility and harm themselves.
Reorganizations can be attempted even with a small amount of computing power, as the super-profitable capabilities that sometimes arise in DeFi can justify a low success rate of 1-10%. In addition, a medium-sized miner can do this, confident enough that he can add 2-3 blocks in a row. Also, the miner can send part of the profit to the contract, from where other miners can extract it, if they help him compete with the canonical chain.
Reorganizations in Ethereum 2.0.
Unlike the Nakamoto consensus with its longest chain, Ethereum 2.0 uses the LMD-GHOST fork selection rule, which includes block initiators (proposer) and attesters (attesters). The initiator must propose a block, and the group of assessors must vote on which extension of the canonical network they consider to be correct.
Every 12 seconds, a new slot appears in Ethereum 2.0, which makes it possible to offer a block. For each slot, the pseudo-random algorithm selects the 32nd part of the validators, one of which becomes the initiator, and the rest become the attestators. In such a system, a potential attacker will have almost no chance to concentrate the dominant part of validators in a single block consisting of 64 slots.
Thus, Buterin believes that if the miners really begin to massively carry out reorganizations, the most effective way to combat them will be to accelerate the transition to Ethereum 2.0. He acknowledges that the hasty merger of the existing network with a new one entails additional risks, but the willingness to do so alone will make miners think about their actions.
In summary, the development of decentralized finance has raised concerns about front running, where miners can easily perform transaction in blocks and outrun the ordinary users. Reorganizations can be performed even with a small amount of processing power. In addition, a medium sized miner can easily add several blocks in a row. Due to the fact that all transactions beyond two epochs are considered complete, long reorganizations are impossible.