What Is MEV? Ethereum’s Invisible Tax Explained
Alongside scaling and hacks, MEV is arguably the most serious issue Ethereum and similar smart contract blockchains face today.
- MEV stands for “Miner Extractable Value” or “Maximal Extractable Value.”
- It refers to the extraction of value from Ethereum users by reordering, inserting, and censoring transactions within blocks.
- MEV is one of Ethereum’s biggest issues, with more than $689 million extracted from users of the network year-to-date.
By leveraging their discretionary power to sequence transactions within blocks, miners can extract value from decentralized application users on Ethereum, greatly diminishing the user experience and threatening the stability of the network.
MEV, The Invisible Tax On Ethereum Users
MEV is an abbreviation of “Miner Extractable Value” or “Maximal Extractable Value.” It refers to profits that can be made by extracting value from Ethereum users by reordering, inserting or censoring transactions within blocks being produced. It typically affects DeFi users interacting with automated market makers and other apps.
Interestingly, the problem of MEV in Ethereum was first identified in 2014—a year before Ethereum launched—by an analyst coder and long-time algorithmic trader operating under the pseudonym Pmcgoohan.
Horrified by what happened in 2008 and the outfall of the global financial crisis, when Pmcgoohan first heard about Ethereum and the idea of a programmable blockchain promising distributed and equitable markets, he became enamored. To use his own words, it “blew his mind,” and he was “so excited about it,” but when he looked at Ethereum’s pre-Genesis draft documents, he was taken aback to find a critical flaw. Pcgoohan recognized that miners had total control of the transaction inclusion and ordering process, which meant that they could leverage this power to extract value from unsuspecting users of the protocol went it went live.
While some instantly recognized the shortfalls of Ethereum’s proposed design, Pmcgoohan was, unfortunately, ahead of his time, and his warning fell largely on deaf ears. That is until, in 2019, a group of researchers highlighted the issue by publishing a paper called Flash Boys 2.0, where the “MEV” term was first coined to describe the problem Pmcgoohan had referenced years earlier.
Subsequently, Georgios Konstantopoulos’ and Dan Robinson’s Ethereum is a Dark Forest, and Samczsun’s Escaping the Dark Forest articles, published in Aug. and Sep. 2020 respectively, cemented MEV as a fundamental concept in crypto-economics and highlighted its importance as one of the most challenging and pressing issues the Ethereum research community faces today.
These texts revealed that MEV was not merely a theoretical issue, but a real phenomenon already occurring at a significant scale with concerning consequences for Ethereum users.
Why MEV Occurs
In Ethereum, miners are responsible for selecting and aggregating transactions into blocks. Crucially, they have full autonomy in deciding which transactions from the mempool—an off-chain space where pending transactions await confirmation—they’ll include in the blocks they mine.
As miners, validators, and sequencers optimize for profit, they tend to select and order transactions by the highest gas price or transaction fees. However, the protocol does not require transactions to be ordered according to fees. Miners can leverage their discretionary ability to reorder transactions to extract additional profits from users. This “irregular” stream of revenue is MEV.
Although MEV is most frequently associated with miners, it is neither a Proof-of-Work nor an Ethereum-exclusive issue. Moreover, “miner extractable value” is a somewhat misleading term. In reality, the majority of MEV extraction today comes from so-called “searchers”—usually arbitrage traders and bot operators—actively seeking and identifying MEV opportunities on-chain and capturing them in different ways, whereas miners only indirectly profit from these traders’ transaction fees. MEV exists on all smart contract-enabled blockchains with a party responsible for transaction ordering, including validators in Proof-of-Stake-based systems like Ethereum 2.0 and rollup providers on Optimistic Rollups.
Understanding the MEV Game
The best way to understand the MEV game is to look at it through the lens of the key players, including miners, searchers, users, decentralized applications, and protocol developers.
The miners or block producers are responsible for sequencing transactions and deciding which transactions to include in blocks and in what order. Miners can profit from the MEV game in two ways: first, by selling scarce block space to non-miner MEV extractors through so-called Priority Gas Auctions (PGA) in exchange for exorbitant transaction fees, and by capturing MEV directly through reordering, including, or censoring transactions to profit from on-chain liquidation or arbitrage opportunities for themselves.
The End Users
MEV also involves the end-users, such as people taking out on-chain loans or trading on decentralized exchanges. Users are the most exploited party in this game as they emit some amount of value that can be captured by miners and non-miner MEV extractors.
DApps and Protocol Developers
Decentralized applications and protocol developers play an auxiliary role. The former create MEV opportunities through their design and the incentives they produce, while the latter establishes the game’s base rules such as giving block producers power to sequence transactions, which is what makes MEV possible.
The Searchers, DeFi traders and Bot Operators
Finally, central to the MEV game are the searchers or the DeFi traders and bot operators who seek to identify MEV opportunities and capture them in different ways. The two primary ways searchers participate in the MEV game are by bidding exorbitant gas prices in on-chain PGAs to have their transactions strategically placed at specific positions within blocks by miners, and by expressing transaction ordering preferences to miners off-chain using novel MEV extraction tools like Flashbots.
The Searchers’ Typical MEV Extraction Process
Searchers start their MEV journey by monitoring the Ethereum blockchain using bots and automation tools for potential profit extraction opportunities.
When they spot an opportunity, searchers analyze the logic behind the trade, conceptualize the attack vector, and create a bundle—one or more transactions grouped and executed in the order they’re provided—designed to materialize its MEV extraction goal when mined. Searchers’ transaction bundles can refer to other users’ pending transactions in the mempool and target specific blocks for inclusion.
Once a bundle is created, a searcher will usually send it to a miner using off-chain networks like Flashbots’ MEV-Geth. This allows them to avoid the public transaction pool and express their transaction ordering preferences fast and risk-free (they save on gas fees when their transactions are rejected) directly to miners.
As searchers in aggregate submit a huge amount of bundles and block space is limited, miners auction their block space through a Flashbots Auction—an off-chain first-price sealed-bid auction where searchers can privately communicate their bid and granular transaction order preference directly to miners without paying for failed bids—and only include the most profitable transactions in their block.
When a miner includes a searcher’s bundle or a transaction in their block, the MEV extraction process is complete. The searcher’s transaction gets confirmed on-chain and, if the MEV strategy was well-designed, the searcher would have extracted some amount of value from other traders on Ethereum.
The Most Common Attacks
Front-running involves getting a transaction first in line in the execution queue ahead of a known pending transaction. In Ethereum, searchers run specialized front-running bots that scan the network for large orders on decentralized exchanges and submit competing transactions with higher gas fees to get them mined before the victim’s transaction.
A sandwich attack is a variation of front-running whereby a predatory trader places two transactions, one before and another right after a pending victim transaction. Searchers typically use sandwich attacks to extract MEV from unsuspecting traders on decentralized exchanges by manipulating the price of an asset. For example, a trader can identify a token a victim is about to buy and make a trade to push the price up, then sell the token straight after the victim’s buy order has further increased the price.
Back-running is the practice of getting a transaction ordered second in line or immediately after a known pending target transaction. Searchers typically employ back-running bots to monitor the mempool for new token pair listings or liquidity pools created on decentralized exchanges like Uniswap. When a bot finds a new token pair listing, it can place a transaction order immediately after the initial liquidity and buys as many tokens as possible, leaving only a small amount for other traders to buy later. The bot can then wait for the price to go up after other traders have purchased the tokens and sell at a higher price for a profit.
Liquidators are searchers that specialize in extracting MEV through liquidations of over-collateralized loans on decentralized on-chain protocols like Compound, Maker, Aave, and dYdX. Liquidators run specialized bots to monitor the network for transactions presenting liquidation opportunities and act to either front-run or back-run transactions to be the first to liquidate a loan. Liquidators extract MEV from unsuspecting borrowers by liquidating their loans before they can repay the debt, then profit by selling the borrowers’ collateral.
Time-bandit attacks are a novel type of attack only miners can execute that retroactively reorganize blocks to capture MEV opportunities in previously mined blocks. When MEV is high enough compared to block rewards, it can be rational for miners to destabilize the consensus to capture MEV in older blocks. For example, suppose a miner with significant mining power spots a $20,000 arbitrage opportunity in block 100 that is three blocks deep. Instead of mining the latest block to earn a much smaller block reward, they may decide to re-mine block 100, as well as blocks 101 and 102, to capture the arbitrage opportunity and have a longer chain than the miner who originally mined the block.
How Bad Is MEV?
According to Flashbots’ data, which only measures the lower bound of total extracted MEV and tracks only eight DeFi protocols, more than $689 million has been extracted from unsuspecting users of the Ethereum network since Jan. 1, 2021.
In addition to scaling and attacks, MEV is one of the biggest issues Ethereum and similar smart contract blockchains face today. Pmcgoohan argued that MEV auctions would kill the Ethereum network. While Pmcgoohan takes a pessimistic view, the negative implications of MEV extraction are many and varied. The biggest one is that MEV represents an invisible tax that miners and searchers collect from users. Every dollar extracted through MEV is a dollar lost for users. Some would go as far as to describe it as theft.
MEV also leads to network congestion and puts upward pressure on gas prices. The game theory involved generates a self-reinforcing loop of circular dependencies: arbitrage and liquidation opportunities create MEV opportunities, MEV-extracting bots compete for the opportunities via gas price bidding wars, and fee estimators use these bot-inflated gas prices as a reference, leading to users overpaying for transactions.
MEV also destabilizes Ethereum on a protocol level because it puts transaction finality and immutability to question. If MEV is bigger than the block rewards, miners are incentivized to destabilize consensus. If miners can reorder transactions in previous blocks for profit, the entire premise of blockchains as secure, predictable, and permissionless ledgers falls apart.
In light of the recent debate in the U.S. Senate on whether miners and validators should be defined as brokers, if these instances become commonplace, it will become increasingly more difficult to defend the role of miners as mere “passive and neutral transaction processors” on blockchain networks.
MEV erodes the usability, neutrality, transparency, decentralization, security of Ethereum today. It creates an environment where miners who are better at extracting MEV grow at the expense of honest ones, effectively skewing the core incentive structure at the heart of Ethereum’s security in the wrong direction.
Ethereum recently launched EIP-1559 and plans to move to Proof-of-Stake, but neither update will solve MEV. In fact, some MEV researchers worry that the upgrades may exacerbate the problem.
While EIP-1559 is primarily designed to improve the predictability of transaction fees, the upgrade also features a fee burn function that negatively affects miners’ profitability, which, in turn, may lead miners to ramp up MEV extraction to compensate for the reward reduction. In response to EIP-1559, Ethermine—a mining pool accounting for roughly 20% of Ethereum’s hash power— has already introduced an MEV extraction program to redistribute the profits extracted through MEV between all miners in the pool.
Announcing #Ethermine MEV beta! 🥳— Bitfly (@etherchain_org) March 17, 2021
In order to compensate the upcoming mining reward reduction caused by the adoption of EIP-1559 we have launched our MEV beta program.
80% of our MEV revenue will be distributed alongside the mining block rewards.
Concerning the move to Proof-of-Stake, MEV extraction will work nearly the same way on Ethereum 2.0 as it currently does on Ethereum, except it will be done by validators instead of miners. Flashbots MEV researchers Alex Obadia and Taarush Vemulapalli believe that the introduction of MEV in validator rewards could be a “centralizing force,” and worry that “MEV could amplify oligopolistic dynamics in Ethereum 2.0 by enriching the entities with the most 32 ETH stakes faster than the ones with less (rich-get-richer dynamics).”
Is MEV Inevitable?
Some have concluded that MEV is inevitable. There are two schools of thought when it comes to this topic. The first school maintains that MEV is unavoidable, so the crypto community should try to alleviate the symptoms and subdue the negative externalities. The other school believes that the MEV problem is solvable, and hence the community should focus our efforts on trying to prevent it.
Flashbots, the leading research and development organization in the field, belongs to the first camp. It focuses on building tools such as MEV-Geth that “democratize access to MEV revenue and bring transparency to MEV.” In that regard, MEV-Geth is effectively a product offering Front-running as a Service (FaaS) to miners and MEV extractors.
Proponents of the first school argue that, given the inevitability of MEV, FaaS is net beneficial because it eliminates negative externalities such as high transaction fees and network congestion while making up for the lost revenue from Ethereum’s EIP-1559 fee burning update. Thus, it indirectly funds Ethereum’s security as miners compete for MEV with higher hash power.
On the other hand, some believe that FaaS is theft. Cornell University researchers have long advocated for an alternative solution, while computer science professor Edward Felten has claimed that MEV auctions increase centralization and exacerbate the problem for Ethereum users. Pmcgoohan also identifies with the second camp, arguing that MEV can be avoided. Critiquing Flashbots’ approach, Pmcgoohan suggests that MEV could be solved if the community builds “a consensus view of the mempool ordering transactions by time where it is discoverable.”
In the second camp, researchers are already gaining ground in minimizing or removing MEV by designing protocols that order transactions fairly. Current application-level solutions include ChainLink’s Fair Sequencing Service, Offchain Labs’ Aribtrum, and Automata Network’s Conveyor. While all of these protocols approach the MEV problem in varying ways, they depend on DeFi applications implementing them on a case-by-case basis. An ultimate, protocol-level panacea is yet to be found, let alone implemented.