Minimise MEV

[lzhou1110]

Hi Vitalik,

I'm writing to see if you or your team are interested in working together on the MEV paper (A2MM). I have found your previous comments very inspiring. I am now convinced that certain types of MEV cannot be mitigated, especially if an attacker attempts to deliberately "give away" MEV to attract MEV extractors, which may be similar to a DDoS (as what I replied on Medium). After all, the conclusions heavily depend on the threat model.

I'd like to further expand on the A2MM paper by considering liquidation and other protocols, as well as provide a stronger and more generalised argument that MEV can be greatly reduced. Would be great if you'd like to contribute to the paper as a co-author.

Regards,
Liyi Zhou([email protected])

————————
My response to your previous comments: https://medium.com/research-imperium/a2mm-towards-protecting-defi-users-from-mev-and-flashbots-d0dca2594be0

Hi Vitalik. First of all, I had a great deal of fun conducting research in DeFi and Ethereum, and I'd like to thank you and your team for this.

Before we debate whether MEV is fundamental, I'd like to say that I hope reading this article prompts the reader to consider whether the entire community wants more resources (money, time, and research) to maximise MEV, or to minimise MEV. Have we made enough efforts to minimise MEV before the community continues to provide resources for Flashbots to make MEV more "democratic"?

Even if MEV is not fundamental, I believe MEV can be greatly reduced, which is beneficial to the Ethereum ecosystem as a whole. Our A2MM paper (https://arxiv.org/abs/2106.07371) mitigates against arbitrage and sandwich attacks. Back in December 2019, our HFT paper proposed a solution to the sandwich attack (and we did disclose it to Uniswap back in 2019) (https://arxiv.org/abs/2009.14021). Now, as a security researcher, I want to speak out, and I hope that more people will join us in making the entire ecosystem fundamentally better and fairer. This is what I actually want to say.

Now back to your comment:

As what I claimed in the article, A2MM or similar designs only mitigate one type of MEV, which is back-running. If we think about this as a state transition system, back-running basically tries to extract revenue after a state transition from a benign user. So why don't the benign user extract it directly, instead of leaving the money on the table?

So I find it useful to consider who is the creator of an MEV opportunity, and can the creator close the MEV itself cleanly.

• Arbitrage against external exchanges (including centralized ones)

Because the centralised exchange is the creator of MEV in this case, the exchange should ideally close the arbitrage opportunity onchain, without the miner even noticing the price change in the exchange (theoretically, but there are regulations and such, so I'm not sure if it's practical).

• Providing fraud proofs

In this case, the fraudster should be closing the MEV after he or she created it. If you're talking about layer 2 optimistic protocols, this makes sense if the fraudster intends to "hit-and-run." However, if the fraudster is betting that no one else will be able to submit the fraud proofs in time when Ethereum is congested, the situation changes and MEV may appear.

• Claiming bounties to "poke" any on-chain contract that requires it

I'm not sure what you mean, but I'm assuming some adversary creates a contract that rewards anyone who can call the contract after a certain block number. This is essentially a distributed denial of service (DDoS) attack on Ethereum, which we discussed in one of our previous papers (clogging attack, https://arxiv.org/abs/2101.05511). In this case, MEV does exist, and I'm not sure how to get rid of it (shame on me).

• Buying from a descending price auction once the price gets low enough (on exchange A)

This purchase by itself, in my opinion, is not an MEV. It is an MEV if there is another exchange B where the adversarial miner can immediately sell the purchased asset to perform an arbitrage. Instead of letting the miner extract the revenue, the previous trader should perform the arbitrage (atomically in one transaction, similar to A2MM actually).

Creating/Extracting MEV consumes on-chain space, network layer bandwidth, harms of benign DeFi users, and may render the entire ecosystem prone to centralised power capture (see Flashbots). That is because DApps are not designed in a way to consider the limitations of their "environment". Analogous to how burning fossil fuels and emitting CO2 ignores the unpriced externality, which is the health of the earth's atmosphere, DApps should be mindful about their MEV mitigation responsibility. Once again, I hope reducing MEV gets more attention from researchers, investors, protocol designers, and, of course, the Ethereum core team.