The fierce game between "fairness" and "righteousness", using dialectical thinking to look at MEV

source|pdaian.com

Some readers may be familiar with "Extracting MEV is tantamount to theft” Such a view. In this article, I will discuss in depth why, in my opinionWithdrawing MEV in the Cryptocurrency WorldNot akin to theft; why MEV is a key metric for cybersecurity in any distributed system secured by economic incentives (yes, alsocentralized systemfirst level title

My MEV Journey

I will first analyzeDistributed Fairness ProtocolRelation to MEV, and how distributed fair protocols relate to true fairness; then argues why MEV is a tricky double-edged sword whose reach is so wide that any attempt to reduce it to "good" versus "bad" All reduced to reductionism; in the end, I will provide a roadmap of how the community should respond to MEV, what I plan to do, and what you should do regardless of your role in the ecosystem.

This article represents only my personal views. Although I am inextricably linked to MEV, either personally or professionally, I hope readers take what follows from a perspective rather than take it all in.

first level title

The existence of MEV is fundamental

First, we believe that the existence of MEV is fundamental to cryptocurrency experimentation. It is not going away. It has the following logical relationship with cryptocurrency experiments:

Transcriptability of encryption technology: All distributed systems that can function as currency must have "auditability (auditability)" This key feature, or the ability to verify system state transitions and/or user actions. Auditability is a key feature provided by the Merkle transaction root of the Bitcoin blockchain and many Ethereum smart contracts. We create encrypted Recording or proof arguments (proof arguments) achieve auditability, they can be verified by others. This is the key to the construction method of distributed economic systems, without it, no one can verify anything of the system. However, recordable Both openness and auditability open the door for MEV. There will always be some users who prefer one version of an event to another, and there will always be some users who find certain records more valuable than others.These users will be able to express preferences for selected outcomes through conditional payments, and these payments (and the preferences themselves) provide a form of MEV by changing the incentives given to those who choose that record.

Interoperability across different trusts: There will never be a single blockchain architecture that will rule them all (sorry, fanboys). In systems with multiple trust assumptions, the boundaries of those systems always provide value to those in privileged positions at those boundaries. For example, even if ETH itself does not have MEV, there may be room for MEV in Binance Smart Chain arbitrage with ETH, especially for those who do verification on both sides. Therefore, the person who controls the order of Bitcoin transactionsCentralized exchange arbitrage during periods of high market activityIt is the same reason. For MEV to disappear completely, the world would have to operate within a single trust zone, which seems...unlikely.

This image I shamelessly took fromA presentation by Charlie Noyes on interoperability and MEV, then modified, that lecture was very instructive

Permissionless: A common design pattern for cryptocurrencies is to provide payments that any user can receive to provide useful action to the network. This is the cornerstone on which countless distributed applications can run. For example, in Uniswap, if there is no MEV formed with an arbitrage market with external or centralized exchanges and other decentralized exchanges, the price cannot reflect market changes, nor can it provide users with useful trading products. In MakerDAO, if there is noPay to liquidation robotMEV, there is no incentive to pay gas to update the loan status in the system, which directly leads to a systemic collapse. In Cryptokitties, no kittens are born without the MEV used to pay for the gas to breed. It is the existence of "bribery of all" (permissionless bribery) that supports the functioning of these systems, and is not always harmful. It's often decentralized or "permissionless" protocols that work, how secure they are, and why they can work without requiring permission.achieve some balancenecessary condition.

first level title

MEV must be withdrawn for economic security

Assuming that you are convinced by my argument so far, that is, you agree that there will always be a certain amount of MEV in a distributed system. Now let's take this as an axiom and only consider worlds where MEV exists and can be extracted (For example, our experience world today)

Think about the economic security model of cryptocurrencies,Satoshi Nakamoto defined it like thisCryptocurrency: "As long as the honest nodes collectively control more CPU power than the attacker group controls nodes, it is secure." More than attacking nodes? This is where it comes to see if Satoshi’s model of honesty holds up to the test.

One reason for the success of cryptocurrencies is thatTheir security assumption is weaker than the honesty assumption. If we can accept relying on an honest majority assumption, say a public key, we can simply useLicense Consensus Protocol, and create a currency abstraction. Instead, cryptocurrencies allow for more abstract economic assumptions than this "permission".

Such an economic assumption is that due to the long-term desire to convert mining revenue into legal currency such "Incentives outside the agreement”, miners have sufficient incentives not to engage in dishonest mining behavior. Another assumption is that CPU power costs money, so Nakamoto consensus makes the attack cost not to be underestimated. No matter what security assumptions lead you to believe Bitcoin works, what makes you believe is probably the economic hypothesis, not the honest one. While honesty may be the foundation of technical analysis, economics is its raison d'être, and I bet that's what a lot of people reading this are Reasons (direct or indirect) for people to believe that Bitcoin works.

Economics are important, we don't want to rely on identity or trust alone. In this model, we as a community utilize the concept of financial rewards in order to motivate people to behave in desired ways. Such incentives provide security for any interesting cryptocurrency system that cannot simply be replaced by a database. In a system where cryptocurrencies are secured by validators, the presence of MEV (whether withdrawn or not) fundamentally changes validator incentives. Validators participating in the economy who ignore MEV will systematically lose out to validators who utilize MEV.

If we want any form of strong economic security, each validator must withdraw available MEV at roughly the same rate. Any validator who withdraws MEV at a higher rate than everyone else is essentially concentrating economic gain, thereby compromising the security of the system. Any miner who withdraws MEV at a higher rate than everyone else is able to centralize CPU control, which is more straightforward in a proof-of-stake system where stakers can withdraw MEV more efficiently in a centralized staked fund.

Leaving MEV alone by validators in a system is tantamount to providing perks to attackers. Under a purely economically rational model, this would amount to suicide and security degradation, as it would lead to centralization or oligopoly. in aIn an economically expensive attack model against nation-states, the consequences of this downgrade are even more serious, as malicious nation-states may be perfectly happy to exploit MEVs to reduce the cost of attacking the system.

first level title

MEV and fairness

Those who are new to MEV naturally ask a question"Why not enforce a fair agreement on a layer protocol and be done with it?"Surely there are cryptographic protocols that build a fairer, less manipulative ordering than existing proof-of-work mining systems?

Herein lies the problem, dear reader. Let us sincerely participate in this quest to "add a fair deal". The first step is to define "a fair agreement". We need to specify a protocol that is integrated into one layer. A good way to choose a protocol is to read theState-of-the-art academic literature on fair rankingimage description

(Red text in the picture: Which protocol would you choose? Who will choose??)

Upon further dissection, it turns out that there is no such thing as a "fair ordering protocol", just as there is no such thing as a "consensus protocol". Instead, what we have is more like a family of protocols that parameterize network assumptions, failure thresholds, set of nodes on which to run them, encryption assumptions, etc. Which to choose, which is more popular?

The interesting conclusion here is that whenever two fair ordering protocols exist, MEV also arbitrages these protocols. Note that the definition of MEV does not require honesty assumptions, but only profit-maximizing behavior. If validators in both protocols are profit maximizing, they can break the fair ordering protocol in a number of ways and profit.

This is similar to market design in many ways. There are many proposals for a fairer market design. They are great. I want to see them deployed, tested and experimented with. But claims that any of these designs are a panacea for improving fairness often stem from their analysis being performed in an isolated model that does not adequately reflect real-world These differences become more salient and more profitable for those in MEV.

first level title

What is fair?

When diving into this complex topic, it is helpful to review how we define fairness in the real world:

Clearly, fairness must be impartial. But how can there be a single fair agreement when there are multiple choices and people (supposedly) have different choices?

I think it's tantamount toA libertarian daydream - having a global property ledger from which all justice in the world originates. I'm sorry, not because I can't empathize, but because I think it's impossible and absurd for all of humanity to agree on such a complex and impactful issue.

Going back to fairness, then we will find that some fairness protocols are useful for some users, and other protocols are useful for some other users. Which agreement do you think is fairer? Which one did Group A or Group B choose? It is impossible to prejudge at all.

Fair ordering protocols may attempt to approximate real-world notions of fairness, but they can never achieve an ideal, non-parameterized definition, because removing these parameters requires humans to agree on "what parameters are fair", while This in turn requires humans to first reach a consensus on "what is fair".

Therefore, fair sorting can only be close to ideal at best. It works, but it's not a panacea for sorting problems, and certainly not universally applicable.

Worse,Most fair protocols rely on honesty assumptionsfirst level title

Fairness sucks, MEV is justified!

Please note that this post of mine is written to counter a narrative that has been prevalent recently: that there is a clear moral divide between quantifying MEV, MEV extraction, and fair ranking, and that the two are mutually exclusive.

I'm not trying to say that fair sorting is useless. Under reasonable assumptions, fair ordering, especially for a specific protocol, can substantially reduce MEV and increase user revenue. This is a key area of ​​research and investigation. However, for the above reasons, fair ordering cannot completely eliminate MEV.

Our complete solution has to go deeper and has to cover more facets of the problem in a more dynamic way.

Clearly, there is a deep relationship between MEV and fairness. Some forms of MEV lookobviously unfair. But there are protocols that actually incentivize users or validators to run a fair order. In fact, running a fair ordering protocol on Ethereum may require rewards for validators, providing MEV directly to a layer. Ironically, the fairness of an economically secure proof-of-work system might not be possible without MEV! (This is just my conjecture, no one has proven it yet...)

From a research perspective, these relationships are still not well understood. However, these relationships clearly exist and deserve considerable research, discussion, and thought.

What's the matter?

honestly….

Now, finally, there is a prescription. Yes, this is for you, you'd better follow your doctor's orders (just to clarify, I'm not from Purdue Pharma 🤡):

For DApp developers: it is necessary to have strict requirements on the extent of MEV restrictions. Carefully consider the MEV setup in your protocol. Where does it exist? How will it harm the interests of users? What benefits can it bring to users? Can it be redistributed or reduced to a minimum? How will interoperability with other systems change it over time? And how is this communicated to the user? Figure out these issues and set them up in the correct way, otherwise your distributed system will crash.

For users: DApp must have the above requirements. Do not use DApps with predatory MEV. Understand what MEV means to you and your transactions. Learn about MEV and gain a solid understanding of the systems you use.

**For L1/L2: Redesign and take MEV into account. Think about the use case you are designing for and how you can better manage MEV. Think about it, if the MEV incentives change after DApps are deployed, how will this change your economic incentives? Do you have a plan to deal with it? And in what way might MEV break assumptions made around network participants?

For miners:Feel free to extract MEV, don't feel shy. For any MEV network, its feature of using game theory to ensure network security requires miners to continuously extract MEV. I would also recommend further aligning incentives and expectations by enforcing a code of conduct for other miners/network participants where necessary, as short-term gain in MEV may harm the long-term interests of all users in the ecosystem.

To everyone: Help us democratize MEV profits while keeping as little damage to users as possible across the system. Get involved in this conversation and keep tryingMitigate 👏 negative 👏 externalities。