https://pdaian.com/blog/mev-wat-do/
Some of you may be familiar with the perspective that MEV extraction is theft. In this article, I’m going to go deep into my personal arguments for why extracting MEV in cryptocurrencies isn’t like theft, why it is a critical metric for network security in any distributed system secured by economic incentives (yes, including centralized ones), and what we should do about MEV in the next 3-5 years as a community. I will argue that MEV is fundamental, and that no known magic wand exists for remedying the problem.
my MEV journey
I will cover the relationship between distributed fairness protocols and MEV, and the relationship between distributed fairness protocols and real fairness. I will argue that MEV is a sticky double-edged sword, broad enough that all attempts to reduce to “good” or “bad” are reductionist. Lastly, I will provide a roadmap for what we should do about MEV as a community, what I plan to do about it, and what you have to do about it, regardless of your role in the ecosystem.
This piece represents my opinion alone. While I have numerous stakes in the MEV game, personal and professional, I encourage you to read the below AS AN OPINION PIECE and WITH A BUCKET OF SALT.
Buckle up, Buckaroo!
First, we will argue that MEV is fundamental to the cryptocurrency experiment. It is not going away. This logically follows from the cryptocurrency experiment for several reasons:
Cryptographic transcriptability: All distributed systems that can function as currencies must contain the key property of “auditability“, or the ability to validate system state transitions and/or user actions. Auditability is the key property provided by the Bitcoin blockchain’s chain of Merkle-transaction-roots, as well as by many Ethereum smart contracts. We achieve auditability through the creation of cryptographic transcripts, or proof arguments, that can be validated by others. This is key to the way distributed economic systems are built, and without it, nobody would be able to validate anything about the system. Transcriptability and auditability, however, also open the door to MEV. There will always be users that prefer one version of events to another, and there will always be users to whom one transcript is more valuable than another. These users will be able to condition payments to express their preferred outcome, and these payments (as well as the preferences themselves) furnish a form of MEV by altering the incentives of those selecting the transcript.
Interoperability under heterogeneous trust: There will never be a single blockchain architcture to rule them all (sorry, Maxis). In a world where multiple systems with multiple trust assumptions interoperate, the boundaries of these systems always provide value to those who are in privileged positions at those boundaries. For example, even if ETH had 0 MEV, there would likely be MEV available arbitraging Binance Smart Chain with ETH, especially available to those validating both. The same goes for MEV controlling Bitcoin transaction order and therefore centralized exchange arbitrage in periods of high-market-activity. For MEV to go away entirely, the world must operate within a single trust zone, which seems… unlikely.
Charlie Noyes’s thought-provoking presentation on interoperability x MEV.
“Permissionlessness”: One of the common design-patterns in cryptocurrency is to provide a payment that can be received by any user to provide an action useful to the network. This is the backbone of how many distributed applications operate. For example, in Uniswap, without the MEV provided arbitraging markets with external/centralized exchanges and fellow dexes, the price would not reflect the market and not provide a useful trading product for users. In MakerDAO, without the MEV that pays liquidation bots, there would be no incentive to pay gas to update the state of loans in the system, leading to immediate systemic collapse. In Cryptokitties, there would be no birthing of bred kitties without the MEV that pays for the gas of birthing. The existence of a “bribe to anyone” (permissionless bribe) runs these systems, and indeed runs the blockchains they operate on themselves. MEV is therefore fundamental, and not always harmful. It is often an essential part of how decentralized/”permissionless” protocols operate and how they are secured, and why they do not require permission to achieve certain equilibria. Sometimes, we want MEV!
Note that other than permissionlessness, these properties can also exist in centralized systems. This should be obvious to anyone who thinks through MEV on a system like Binance Chain, and thinks about who controls who extracts what, but it also applies to distributed databases, non-financial systems, messengers, and any system in which a user may want to securely bribe a participant to perform a certain action. These bribes can be seen as generalizations of MEV.
Assuming you buy my arguments up until here, some amount of MEV will always exist in distributed systems. Let’s take this as an axiom, and reason only about worlds in which MEV exists and can be extracted (such as, empirically, the world today).
Consider the economic security model of cryptocurrencies. Cryptocurrencies were posited by Nakamoto as “secure as long as honest nodes collectively control more CPU power than any cooperating group of attacker nodes.” When can we expect “honest” nodes (nodes that run the protocol without deviating) to outnumber attacking nodes? This is where the rubber of the Nakamoto-honesty model meets the road.
One reason why cryptocurrencies have been successful is their security under weaker assumptions than honesty. If we were fine on relying on a majority-honesty assumption, say of public keys, we could simply use permissioned consensus protocols and establish a money abstraction. Instead, cryptocurrencies allow for existing in a world with more abstract economic assumptions than such “permission”.