This post provides a high-level introduction to Maximum Extractable Value (MEV), why MEV focused tools such as Flashbots have emerged in recent times, and finally explores and discusses briefly Eden Network’s whitepaper.
Maximum Extractable Value (MEV) can be understood as “the total value that can be extracted permissionlessly (i.e. without any special rights) from the re-ordering, insertion or censorship of transactions within a block” or even across a series of blocks. In some sense, it can be regarded as a sort of tax on network participants.
Although MEV was initially defined as Miner Extractable Value, it is not the case that transaction ordering is handled by miners in all blockchains. In comparison, Maximum Extractable Value provides a wider and more encompassing definition, which also includes validators in Proof-of-Stake (PoS) systems or rollup providers. For Ethereum 1.0 and similar Proof-of-Work (PoW) systems, both definitions are equivalent. As this article focuses on Ethereum, the term miners is used throughout to refer to the party in charge of transaction ordering.
Early work on MEV had already highlighted transaction ordering as a potential attack vector on decentralized exchanges (DEXes) via front-running. However, it did so from a too broad perspective or without assessing its extent nor security implications at a base protocol level.
Published in 2019, the FlashBoys 2.0 paper formalized and provided empirical evidence of MEV activity on-chain. Furthermore, by evaluating its magnitude, Daian *et al. *identified MEV as a significant and credible risk to PoW and PoS protocols, as it introduces unpredictable incentives which could ultimately destabilize a blockchain’s consensus layer.
For example, if MEV within a block was significantly greater than the block reward plus fees, other miners would be incentivized to create a competing block to capture that MEV for themselves rather than to build on top of the canonical chain. If the MEV bounty was high enough, it could even make economic sense for miners to carry out deeper chain re-organisations (i.e. time-bandit attacks). There’s therefore a tension between block rewards and MEV with deep implications for network security.
Since last year, miners’ revenue linked to MEV capture has been on the rise, driving miners’ total income up and now representing a non-trivial percentage of their earnings. Note that it is nearly impossible to find out what that percentage exactly is, as that ultimately depends on how much of the currently extracted MEV is correctly identified as such. As a rough lower-bound estimate, *at least *more than $700m in MEV have been captured since last year on Ethereum alone.
Although MEV extraction is also present in other blockchains such as BSC or Polygon, most MEV activity is currently happening on Ethereum. This is due to the fact that most MEV opportunities open up as a direct result of users interacting with decentralized finance (DeFi) protocols. So far, Ethereum has had significant success at bootstrapping a vibrant decentralized economy, being the total value locked in DeFi protocols over $100b.
Since Ethereum’s security is somehow dependent on the ratio between block rewards plus fees and MEV, changes in the block rewards schedule or in the way fees are handled are also an important matter.
While block rewards have remained constant since early 20191, miners’ revenue from fees saw a major increase following 2020’s DeFi summer. This was primarily driven by an explosive growth in DeFi activity caused by novel protocol launches and liquidity mining programs (e.g. Yearn Finance). As a second-order effect, MEV capture via priority gas auctions (PGA) took off as well, leading to higher gas prices across the network and therefore increasing miners’ revenue from fees.
This has recently changed with the merge of the much anticipated Ethereum Improvement Proposal (EIP) 1559 as part of the London hard-fork. Among other improvements, this upgrade should translate into several benefits for users and ETH holders, such as better fee estimation or enforced monetary premium. More importantly, EIP-1559 also changes completely the way transaction fees are handled.
Before its deployment, miners accrued the totality of fees paid by network participants. As a consequence of EIP-1559, users’ fees are now split between premium and base fees, being the latter burnt automatically by the system. Premium fees are only expected to be significant when the network is heavily congested, so most fees (i.e. base fees) will be burnt instead of handed over to miners from now on. Hence, as a result of EIP-1559, miners’ revenue per block is expected to drop significantly.
These developments will incentivize miners to maximize the amount of MEV they currently capture even further. This trend can be dangerous, as it implies that miners’ revenue will not be dictated anymore by their hashrate capacity alone but also by their algorithmic competency to identify and exploit MEV opportunities. This poses a new threat: miners sitting at the cutting-edge of MEV capture will be more profitable, which in turn will allow them to increase their market share and put miners who have fallen behind the MEV race out of business. Hence, MEV is not only a security problem from a game-theory perspective but also a credible and significant centralizing force when it comes to hashrate distribution.
To fight against these future developments, a research organization under the name of Flashbots emerged in late 2020. In a nutshell, Flashbots’ mission is to minimize any negative externalities associated with MEV (such as high network fees) while democratizing access to MEV revenue to all miners.
To do so, Flashbots has established a private communications channel between users and miners. Users can send highly detailed transaction bundles on Flashbots together with a quote on how much they are willing to pay for inclusion in a block. On the other end, any miner accepting Flashbots bundles can permissionlessly process those for a profit without having to put any effort on their side to identify and capture MEV.
Counterintuitively, Flashbots also allows users to protect themselves - to some extent - against MEV extraction. Instead of publishing their transactions publicly on the mempool and exposing themselves to MEV attacks (e.g. sandwiching), users can privately submit their transactions through Flashbots and define a fee they are willing to pay in order for their transactions to be mined. This gives users state guarantees regarding transaction execution outcomes without unexpected surprises. Evidently, from a user perspective, this only makes economic sense as long as the fee offered is less than the MEV that would have been otherwise exposed.
Although perhaps partly unintended, Flashbots highlighted the value of solutions that offered MEV protection as a service. It also made clear that there was demand for systems that empowered individuals over miners, allowing them to negotiate transaction execution terms as equals and with greater granularity. This has led to the emergence of several products and protocols focused on this matter, such as mistX, MiningDAO or ArcherDAO.
Earlier this year, ArcherDAO developed Archer Swap, a product that allowed users to execute token swaps without exposing themselves to MEV attacks by simply skipping the mempool.
Following the London hard-fork, ArcherDAO has re-branded to Eden Network. At a high level, Eden Network self-defines as a “non-consensus breaking transaction ordering protocol which allows participants to guarantee placement within blocks and protection from arbitrary reordering. […] An accompanying token reward system realizes MEV profits to block producers to maximize network security”.
In other words, Eden Network is a MEV protection as a service protocol aiming to align incentives across different stakeholders through tokenomics design. According to the whitepaper, it has the following three main objectives:
Protect users from MEV extraction and reduce negative externalities on Ethereum.
Improve earnings for miners and increase network’s resilience against time-bandit attacks.
Tokenize access to MEV via the protocol’s native token and re-distribute benefits to the wider network.
The central coordination unit of the protocol is the EDEN token, which is simply a re-branding of the legacy ARCH token. Contrary to the latter, which lacked any sort of use case other than voting, EDEN is used both by users to pay for or access certain network services and to incentivize miners to process Eden blocks.
Eden blocks can be thought of as a subset of valid Ethereum blocks which comply with Eden Network’s ordering rules and present the following structure:
Header:
Consists of three slots. These contain the highest priority transactions, as they are executed before any other transaction in the block and in strict order (i.e. slot 0 transactions are cleared first, then slot 1’s, etc…).
Each slot has a maximum gas limit of 1.5M and is owned by exactly one account at any given time. Ownership is determined through a continuous auction mechanism known as Harberger tax.
In order to own a slot, accounts must place bids denominated in EDEN. Funds from the auction winner are then locked and a 3.3% daily tax rate is applied (i.e. an amount equal to 1/30th of the initial tenant’s principal is burnt daily). After 30 days, all EDEN initially staked by the slot tenant will have been burnt. The only exception to this would be if a new tenant outbid by 10% the previous slot owner’s bid. In such case, any remaining funds would be returned to the original tenant and ownership would be transferred.
Body:
Block producers are allowed to allocate up to 4M gas in block space to transaction bundles just after priority slots. Bundle selection is purely done *à la *Flashbots based on gas efficiency. In essence, the greater the profit per unit of gas spent, the more likely a bundle will be included in a block.
Bottom:
Standard transactions from users ordered by total amount of EDEN staked first and ETH tip (i.e. premium fee) second are included next. Addresses that have staked at least 100 EDEN are able to access some further privacy-enhancing services.
Source: Eden Network whitepaper.
Miners who produce Eden blocks according to the criteria presented above are owed some Eden by the network. After each epoch (i.e. 24h), an admin address must submit the root of the Merkle tree that captures all balances owed to miners, ultimately enabling block producers to claim any outstanding EDEN debt.
The governing body of the Eden Network is the Eden DAO which is composed of Eden block producers that have previously procured EDEN tokens. At any given time, members of the DAO can vote to include or exclude any miner from the list of Eden block producers eligible for miner rewards.
Despite its very recent launch, Eden Network has already shown significant traction in terms of miners engagement. At the time of writing, over 50% of all Ethereum hashpower is participating in the network. For context, more than 85% of it has been onboarded to Flashbots as of this date. This relative success allows for the assessment of how the protocol performs against some of its core objectives and the analysis of some real on-chain examples. Hereafter, Eden Network’s main objectives as described in the whitepaper are commented:
Protect users from MEV extraction and reduce negative externalities on Ethereum.
Eden Network users can be broadly classified as one of the following types: large players renting top priority slots or individuals staking EDEN for higher priority execution and optional privacy features.
As of launch, the three top slots had been secured by Band protocol, Bancor and Sushi.
The motivation for oracles like Band to secure a privileged block position is clear: by ensuring that any data updates needed are included at the top of the relevant block, oracles can improve their core product’s security against front-running attacks and ensure that their customers get accurate and updated data feeds when required.
In some instances, the total gas allowance linked to a slot may be excessive for an oracle, so sharing its remaining bandwidth with another protocol could be a sensible option to reduce costs. Following this line of thought, Band protocol has announced that it will be sharing its slot with Alpha Finance, showcasing that pooling resources and “sharing” a slot between multiple protocols may also be a viable option.
Given that oracles are a critical piece of DeFi’s infrastructure and indirectly secure billions of dollars, it does not seem far-fetched to expect them to be among the top bidders for Eden Network’s slots.
In a similar fashion, DEXes’ reasoning behind securing a top slot has to do with mitigating MEV extraction for their users. However, dynamics in Eden blocks show that the current implementation can be easily exploited in some situations by MEV extractors to siphon funds from those very DEXes’ users at a very low cost. A slightly simplified version of this attack vector is presented below together with data from a real on-chain example:
Honest user Alice sends a public swap transaction to SushiSwap’s router, paying the required base fee and an optional premium fee A.
MEV exploiter Eva, who is continuously monitoring the Ethereum mempool, sees Alice’s transaction before it is included in a block. Eva then submits a transaction to SushiSwap’s router so to front-run Alice by targeting the same coin pair, trade direction and paying the required base fee plus a premium fee marginally higher than Alice’s B.
Both Alice and Eva’s transactions will be included in Sushi’s slot. Because transaction ordering within a slot is determined by premium fees, Eva’s will be placed just in front of Alice’s, thus effectively front-running the latter. Note how, in the on-chain example referenced, Eva’s gas price is just 0.000000002% higher than Alice’s.
In order to profit from the sandwich attack, Eva submits a second transaction to be included in a lower priority position within the same block. This targets the same coin pair but in reverse order, effectively selling the assets acquired in her first transaction C.
Hence, while renting a top slot does indeed protect Sushi as a protocol from being front-run by other actors, it does not protect its very own users from being attacked from within, suggesting that Eden Network slots may be of very limited use to DEXes in their current form.
In contrast, entities focused on identifying and capturing MEV (commonly known as searchers) may be better suited to own priority slots. This is due to the following two reasons. First, since searchers can benefit from any MEV extraction opportunity, they should be willing to spend more for priority space than most protocols. Second, searchers are in constant competition against each other: for each opportunity, only the one whose transaction lands first on-chain is able to capture the exposed MEV. Thus, being able to ensure execution priority over competitors presents an unparalleled competitive advantage for this group. This creates a centralizing force on the searcher side: only the best-funded teams would be able to cover the upfront cost of renting an Eden slot, which would result in higher MEV accrual and therefore earnings, leading to a vicious cycle where very few searchers would remain profitable if Eden Network was to really take off.
Overall, from the different types of candidates considered, oracles and searchers are the ones who seem to have the strongest incentives to secure a slot, while the current transaction ordering implementation should discourage DEXes from owning any. When examining the protocol’s promise around “protect[ing] users from MEV extraction”, it certainly is not always the case. While potentially true for oracles, it clearly does not hold for DEXes’ end users. More importantly, searchers may employ it to achieve the exact opposite end: instead of using it for protecting themselves against MEV attacks, Eden Network can become a centralizing and pure MEV extraction tool on their hands.
Furthermore, it must be strongly stressed that not all slots are equivalent and that their relative position is also paramount. If a searcher owned slot number 0, then tenants of slots 1 and 2 would still be exposed to MEV extraction by their predecessor.
On the other end of the spectrum, some smaller size accounts may find Eden Network’s value proposition appealing. By staking at least 100 EDEN, users can directly send transactions to Eden’s relayer privately, thus essentially protecting themselves from malicious MEV attacks as advertised. It is worth noting that a similar outcome can be achieved via Flashbots instead without the need for acquiring any EDEN. However, in order to ensure inclusion, users would need to pay some premium fee for each transaction, while none is necessary for EDEN stakers.
Hence, individuals interested in MEV protection have the choice to either “stake and forget” 100 EDEN tokens once to access Eden Network’s services or to achieve so via Flashbots and spend some extra ETH on each transaction.
Improve earnings for miners and increase the network’s resilience against time-bandit attacks.
In an efficient, competitive market where participants have perfect foresight, the price of renting an Eden Network priority slot would be marginally inferior to the MEV that could be extracted by having ownership over it. In other words, no rational entity would be willing to pay more for reserved priority block space than what it would be able to extract or protect as a result.
As per Eden Network’s tokenomics design, this value transfer between slot tenants and miners takes place indirectly by burning tenants’ locked funds and awarding some EDEN to miners. Assuming that the protocol’s value accrual mechanism successfully transferred all value from the former to the latter group via EDEN, then the maximum revenue miners could expect would be marginally inferior to the value of the MEV captured or protected.
On Flashbots, this equilibrium state has already been reached. Searchers’ profit margins on the platform are very low, since to ensure inclusion over their competitors, they need to incentivize miners with higher fees. This has sparked a race to the bottom between searchers, which lose most of the MEV they extract to miners. Hence, given that currently most MEV captured on Flashbots is already being directly accrued to miners, it seems unlikely that Eden Network could improve miners’ earnings in any way.
It is worth noting that the analysis above disregards any revenue associated with bundles. However, because all bundles in Eden blocks come from the Flashbots relay, the value accrued to Eden miners per unit of gas spent on these should be equal or less than for those who only participate in Flashbots. Whether it is the same or less ultimately depends on if the searchers submitting the most profitable bundles up to 4M gas (i.e. Eden Network’s block space reserved for bundles) are slot tenants too or not.
Additionally, the gas limit structure of Eden blocks is likely to be sub-optimal in most situations. For example, during market corrections, significant MEV opportunities become available in the form of liquidations. Searchers identify these and rush to submit bundles to capture as much MEV as possible. In this situation, miners looking to optimize their revenue would include as many bundles as possible. Evidently, this is not possible for Eden Network’s block producers, since the protocol imposes a hard cap on how much gas they can spend on bundles per block.
Moreover, some of the best-case assumptions denoted above are not necessarily true. For example, it is clearly not the case that the protocol transfers all value from slot tenants’ to Eden miners. This is because EDEN’s issuance as per its current inflation schedule is not only distributed to miners but also to liquidity providers (LPs) and to Eden Network’s treasury. This essentially makes EDEN a value extracting token and prevents miners from accruing all the value users disburse to access the protocol’s services.
In addition, it is also unclear if tenants are willing to spend as much on slots as the value of the MEV exposed in them. In the case of protocols, these may not deem worth spending significant amounts of capital on a priority slot, since that cost would ultimately have to be passed to their end customers and potentially affect their traction. For searchers, acquiring and staking EDEN upfront represents an additional risk, since future MEV revenue is uncertain. This, together with EDEN’s token volatility, will most likely result in a significant imbalance between slot prices and MEV captured. In turn, this should disincentivize miners to join the network long-term, as options like Flashbots present higher profitability for block producers.
Tokenize access to MEV via the protocol’s native token and re-distribute benefits to the wider network.
Given the relevance and increasing magnitude of MEV on Ethereum, MEV tokenization presents a compelling proposition, as it can enable the formation of several multimillion markets and provide individuals with a tool to speculate on and hedge against MEV extraction.
As a result of Eden Network’s design, EDEN – the protocol’s native token – effectively captures demand for reliable MEV extraction and tokenizes it, albeit in a loose manner as shown in the previous section. However, it is unclear how the current protocol’s implementation would facilitate any MEV captured to be redistributed to the wider network.
Moreover, if the protocol achieves a significant share of Ethereum’s hashpower, there are few incentives in place for Eden DAO members to decrease the amount of MEV they capture. On the contrary, governance attacks by miners holding a majority stake in the network could attempt to centralize MEV revenue even further by limiting the list of whitelisted miners or exert their power in different ways.
Eden Network is a novel MEV protection as a service protocol aiming to align incentives across different stakeholders via EDEN – the protocol’s native token. As per the protocol’s whitepaper, Eden Network has the following three main objectives: protect users from MEV extraction, improve earnings for miners and tokenize access to MEV via EDEN.
Marketing “MEV extraction protection” as one of the core objectives of the protocol seems somewhat deceptive when it comes to slots. This is because, among the different types of participants considered, searchers (entities that capture MEV for profit) seem to be the ones best suited to own a slot, making Eden Network more akin to a centralizing MEV extraction tool rather than a protective one. Other entities such as oracles may find reserved block space useful as long as preceding slots are not owned by searchers. In the case of DEXes, the current Eden Network’s implementation is clearly broken as it allows attackers to cheaply capture MEV from those very same DEXes’ end-users.
Smaller accounts staking EDEN for MEV protection may find Eden’s proposition compelling if they are savvy enough to directly communicate with the protocol’s relayer. However, other alternatives such as Flashbots may be more attractive in cases where Eden’s upfront costs are too large.
Given that searchers on Flashbots already lose most of the MEV they capture to miners, it seems very unlikely that Eden Network will improve profits for block producers once speculators fade out. In addition, current protocol mechanics and EDEN’s inflation schedule make EDEN a value extracting token, preventing miners from accruing all the value users disburse to access the protocol’s services. An unexplored path that could result in increased miner profitability would be if Eden Network effectively enabled multi-block MEV capture.
The protocol does indeed facilitate MEV tokenization through its native token. However, how this may help to re-distribute MEV benefits to the wider network as emphasized in the whitepaper is unclear. If anything, the fact that Eden DAO will be controlled by the largest set of EDEN holders seems problematic. This is because miners with the largest stake on the network will have the power to blacklist other miners, thus further centralizing MEV extraction.