MEV and EIP-1559

The inclusion of EIP-1559 in the London hard fork will bring about a major adjustment to the Ethereum fee mechanism, aiming to make it easier for users to estimate fees, and to consolidate ETH’s status as the network’s basic currency by burning some transaction fees.

This article will analyze some of the consequences of this EIP under the phenomenon of MEV (Maximum Extractable Value) (that is, by reordering transactions, adding transactions, or reviewing transactions to achieve value extraction without permission).

Under the new charging mechanism, users do not choose gas prices for their transactions, but set a “priority fee” for miners to motivate them to pack, and a “max fee” to indicate that they are willing to pay The absolute highest price. The protocol will determine the “basefee” for each block, which is calculated by programming the amount of gas consumed by the previous block. The calculated equation is a negative feedback loop, which aims to stabilize the block size at one The target value s 0 (this value is initially equal to the maximum value of the current block capacity).

The gas price to be paid for a valid transaction is equal to the sum of the basic fee and the tip (only if the basic fee increases suddenly, the sum of the two will rise to the upper limit of the fee); the tip belongs to the miner, and the key is that the basic fee will be burn.

These changes have some significant direct impacts on the MEV-related infrastructure. For example, there will be no more transactions with a transaction fee of 0 gwei. DEXs like MistX are now taking advantage of this to protect against runaway transactions, because in runaway transactions Miner fees are directly withdrawn from the tokens in the transfer. On the other hand, we expect that the new fee mechanism will not bring about a new source of MEV.

In this article, we will focus on three areas where EIP-1559 may have interesting effects on MEV-higher MEV extraction incentives for miners, different bidding mechanisms coexisting in the Ethereum protocol and their impact on EIP design. There is also the role of Flashbots as a collaborative organization for miners and its moral implications.

Miner economics and higher incentives to extract MEV

Whether due to considerable economic incentives or loyalty to the network, miners have been producing blocks since the creation of Ethereum and have not maliciously violated the agreement. However, the Flash Boys 2.0 paper warns that indiscriminately extracting MEV may bring harmful results to the agreement, such as transaction review or chain reorganization, which will ultimately threaten consensus stability.

However, it is worth noting that only recently, when Flashbots introduced MEV-geth, MEV extraction became a standard operation for miners. MEV-geth is forked from the geth client, miners can run it to start receiving “MEV transaction bundles”-a combined transaction set that ensures that miners will be paid when packaged into a block.

Part of the reason is that the threat of EIP-1559 reducing the income of miners is approaching, and most miners will soon adopt MEV-geth to partially alleviate this impact. Therefore, we have reason to ask whether miners will further invest more in extracting MEV when EIP-1559 is launched on the mainnet, especially in ways that damage the network.

Although the loyalty part is difficult to quantify, and therefore we cannot make quantitative predictions, we found that MEV extraction is regarded as miners must use other strategies to compensate for their reduced revenue (such as using their computing power for other GPUs) Chain) can help us to quantify.

To this end, we provide a basic model of miner economics to estimate how much the Realized Extractable Value (REV [1]) needs to increase in order to offset the profit growth of transferring to other blockchains after the London upgrade .

Translator’s Note: The definition of REV comes from Flashbots’ previous article “Quantifying Realized Extractable Value”, which states: MEV is the theoretical amount that we can only keep close to, and REV is the actual value that can be extracted from the MEV opportunity of the blockchain , So REV≤MEV.

We first defined a set of GPU-based PoW blockchain set X except for Ethereum. Miners on these chains can transfer computing power/hash rate to each other (we assume that there is no cost). Then, we assume that in an equilibrium state, the total GPU computing power H will be distributed between Ethereum (before and after the London upgrade) and X with the maximum computing power profit (otherwise, there will be more before the average Miners transfer to other chains). Finally, assuming that the cost of computing power is constant, and that all chains are the same, we can derive a computing power fraction that remains unchanged after the London hard fork of Ethereum. This fraction can be succinctly expressed as:

MEV and EIP-1559

Wherein, H 1559  is a hash of the Ethernet upgrade Square London, H E  is the hash of the former Ethernet Square London upgrade, R & lt gamma] = 1559 / R & lt E  represents the start EIP-1559 miners left in total in an Ethernet Square Revenue, and δ = R X /R E  is the ratio of the total X revenue to the Ethereum revenue before the London upgrade (we have omitted the algebraic process here, and readers can view it here).

The relationship between these quantities can be visualized as the following figure:

MEV and EIP-1559

Ethereum’s Gas price has experienced two orders of magnitude changes, making it difficult for us to accurately estimate the parameters of the model. Based on the recent gas price, we can still use the mining reward data on the Flashbots dashboard and Etherchain to select a value for the total miner income fraction γ=0.86 available on Ethereum after 1559.

Finally, using CoinMetrics data to find out the daily mining revenue (in USD) RX of GPU-based PoW chains other than Ethereum, we get an approximate value of δ=0.15 [2]. Inserting these values, we get the calculation power fraction on Ethereum after the London fork [3]

MEV and EIP-1559

This constitutes an extremely simplified economic model for miners, especially ignoring the cost of transferring to other chains. However, it does provide a framework for analyzing MEV earnings in the broader miner economics. Also based on the fluctuating Flashbots data, we found that under the same other conditions, the fractional result of miner income from the actual extractable value will rise from 2.9% before the London upgrade to 3.4% after the fork[ 4].

If the computing power is not transferred to other chains until the average value is reached, miners decide to withdraw more MEV to make up for the reduced revenue. They need to withdraw an additional 22% of MEV to offset the increase in revenue transferred to other chains (this It still means that there is a substantial reduction in revenue compared to before the London upgrade).

A more detailed model with a clear formula for comparing the cost of extra MEV extraction and transfer to other chains is beyond the scope of this article. The values ​​involved have huge differences in nature, which will make people question the utility of highly refined models. In addition, the subjective factor of miner loyalty plays a very large role in this matter.

Integrate multiple bidding mechanisms in one

The fee mechanism proposed by EIP-1559 is only designed for transaction packaging (and later only analyzed) a bidding mechanism. However, in fact, most activities on Ethereum are not only related to packaging, but also related to the ordering of transactions in a block. Most of the MEV withdrawal opportunities depend on the relative position of the exchange, and it is not enough to earn MEV just to pack the transaction.

The current first-price mechanism is limited because users can express their desire to pack their transactions in a position earlier than a certain transaction, or at most by selecting its specific gas price In order to be packaged in a position behind a certain transaction (tail-chasing transaction). This limitation has led to the birth of systems like Flashbots, which provide a richer language for expressing preferences (users can bid for the precise relative ranking of transaction sets, which refer to the “MEV transaction bundles” mentioned above. ).

Another ideal attribute that the current pricing mechanism cannot provide is “privacy”. Searchers extracting MEV particularly hope that their strategy can be kept secret—at least until the block is mined and inevitably made public—to prevent other actors from stealing their opportunities. Flashbots and other privacy transaction pool providers provide this guarantee (except for uncle block risk or misconduct by miners).

Recently, some user-facing applications, such as MistX and 1inch, have turned to Flashbots in their transactions to provide users with preemptive transaction protection. If more and more participants in the network care about bidding for transaction ranking and privacy rather than simply packaging, this raises a question: Does EIP-1559 have an advantage in dealing with this problem, or will it be used by similar Flashbots? , A more expressive mechanism is completely replaced.

In an extreme case, these different and coexisting auctions may have negative interactions, that is, users participating in one of the auctions will destroy users in the other auctions.

The more users pay attention to transaction sequencing and privacy rather than just packaging, the wider the field will expand, which will cause practical applicability issues like these rigorous formal analyses by Tim Roughgarden (quoted above).

In his paper, he showed that the EIP-1559 mechanism is compatible with short-sighted miner incentives (Myopic Miner Incentive Compatible, abbreviated as MMIC, that is, miners are motivated to act in accordance with EIP-1559 at the scale of a single block) and can prevent The off-chain agreement (Off-Chain Agreement-proof, abbreviated as OCA-proof, which means that users and miners cannot break through the system regulations and use off-chain communication to reach an agreement), and are compatible with user incentives (User Incentive Compatible) except for the high-demand phase. , Abbreviated as UIC, that is, it achieves an ideal user experience improvement-users can express their true preferences for packaging without speculating on the behavior of other users).

However, the establishment of these conclusions all rely on the “simple packaged” auction model, and it is not clear whether they are still valid in the current more complex environment.

In particular, the paper discussed another proposal-“no tipping mechanism”, that is, no tipping for miners. The paper stated that this mechanism is compatible with short-sighted miner incentives, always compatible with user incentives, and can prevent off-chain Agreement (except during the high-demand phase). In short, it sacrifices features that prevent off-chain transactions to maintain the overall user experience.

In fact, unlike what is said in the paper, off-chain protocols are already very common today through Flashbots and other systems. This indicates that the no-tip mechanism may be a better option than the 1559 “basic fee + tip” model. The selection optimizes the standard bidding mechanism for ordinary users, because they only care about packaging, and direct users with more complex needs to the bidding mechanism that focuses on sorting.

Unfortunately, it is extremely difficult to perform rigorous modeling of sorted auctions (Imagine that the allocation rules cannot be expressed simply by binary “package” and “non-package”, or the feasibility conditions need to consider the interaction between transactions, so it cannot Expressed as a single inequality). On the contrary, it may be more useful to separate the different auction modes as mentioned above and then analyze the interaction between them.

We can use the marginal cost of packing a transaction as an example of this interaction μ. In environments with few MEVs, this can be regarded as a constant and has actually been estimated. However, in an environment with a large number of MEVs, there are huge MEV rewards in the trading pool, which may make it meaningless to run regular auctions for packaging in a block, because the value of μ is greatly increased, as a result It is impossible for users to converge on a reasonable tip value.

In summary, we now have several auction models coexisting in Ethereum:

MEV and EIP-1559

At Flashbots, we are considering expanding the expressiveness of the bidding mechanism to include use cases for users who don’t care about ranking and just want to be protected from running away. Although the current pricing formula may be biased towards certain types of withdrawals (especially arbitrage rather than liquidation).

As pointed out above, from a design point of view, this de facto layered system may not be optimal because it is pieced together by actors who try to solve part of the problem individually. In traditional finance, different instruments (stocks, options, etc.) have independent markets to adapt to their different characteristics.

This is an active research field and long-term goal of Flashbots-to create a “Turing complete” auction mechanism, so that users can effectively express their arbitrary preferences [6]. Ultimately, this requires more discussion around multiple coexisting auction mechanisms, so that a more holistic approach can be directly adopted at the protocol layer. This is especially important considering the changes that will be brought about by the upcoming Ethereum 2.0.

Ethical principles of Flashbots

As mentioned above, Flashbots has introduced a way for searchers to express transaction ranking preferences to miners, bringing a more efficient market that all Ethereum users can ideally benefit from. To achieve this, Flashbots provides a large number of miners with customized mining software (MEV-geth), and together they control the vast majority of Ethereum’s computing power (85% at the time of writing).

This is equivalent to Flashbots establishing an effective off-chain miner collaboration mechanism, which will naturally lead to exploration of its potentially harmful effects. In particular, Flashbots has created a new Shelling point (Shelling point), miners using this software and its default settings can automatically collaborate-different from the previous use of geth.

The miner collaboration discussed here, we only mean intra-protocol collaboration, that is to say, the joint action of miners may harm the network, but it is still acceptable at the consensus level (that is, it will not be double spend, or other disregards). The behavior of the Ethereum protocol). A true 51% attack will seriously damage the network, and in general is contrary to the interests of miners. Instead, we focus on more “benign” situations, where miners follow the rules of the agreement while coordinating their own interests with other actors.

There is no analysis on how the centralized point of miner collaboration affects the network (because this can already be achieved through the geth upgrade), but we have to ask a more specific question-whether the introduction of EIP-1559 is possible to increase Flashbots as a miner collaboration mechanism The vicious consequences.

We have introduced how EIP-1559 can change the pattern of this field by increasing the relative weight of MEV rewards in the total share of miners’ income, which is independent of Flashbots. One way Flashbots affect this area is to change the block space share used for transaction bundles and ordinary transactions, or the relative weight of the different auction mechanisms described in the previous section.

MEV-geth can set the maximum number of transaction bundles that are packaged into the block. Based on this, someone builds a template block for each number of transaction bundles (from 0 to the maximum configurable value), and then all of them The template blocks were later used for profitability comparison. The larger the value set, the more profitable the block, and the less block space left for standard user transactions.

EIP-1559 incentivizes miners to pursue more MEVs, which may cause their acceptable transaction bundle value to become higher and higher, and finally there are only MEV transactions in the block. However, this situation is unlikely to happen, because transactions with high tips are excluded, and there is an opportunity cost for only packaging transaction bundles. In general, we do not believe that EIP-1559 will increase the risk of known malicious behaviors facilitated by Flashbots.

However, EIP-1559 may introduce a bad behavior: miners cooperate to mine blocks below the target capacity to drive the basic fee to 0, which will effectively eliminate the user experience improvement that this EIP wants to provide. Similarly, Flashbots is in a special position that can make this behavior possible by providing version updates.

More specifically, instead of the gas consumed exceeding the target gas limit, all transactions that require a tip of more than 0 for the basic fee are packaged into the block, but the software sets the block size to s 0 ∈( 1-ϵ), where s 0  is the target block size, and ϵ∈[0,1] is a parameter deliberately set to drive and keep the basic cost down, which depends on the percentage of computing power running MEV-geth.

We used the framework of the Robust Incentive team of the Ethereum Foundation to explore the feasibility of this attack (the notebook with the code of these results is here). Take a simple shock demand situation and set the possibility of miners to run MEV-geth as P FB = 0.85. We have simulated system dynamics with different values ​​of ϵ, indicating how big this value needs to be before the basic cost will be driven to 0.

MEV and EIP-1559

As in the previous case, the use of this strategy by miners will cause immediate economic losses, because they will not be able to make some money in each block.

The next question that should be asked is how long it takes for miners to work at a loss before the basic cost drops to make back that much money, making their collusion profitable. In the simulation, we found that this issue is extremely sensitive to the nature of the demand, depending on whether users will continue to offer bids close to their true value, or reduce their bids by adapting to new, artificially low basic fees.

Although we are unable to provide a meaningful answer before seeing EIP-1559 on the mainnet, we will increase our attention to this question, and also note that even if such collusion proves to be profitable in a short period of time, it can It is broken at any time because a selfish miner packs a large number of transactions, so this is unlikely to happen in practice.


We did not find a key way for EIP-1559 to interact with MEV extraction. However, we have discovered several areas where new dynamics may occur, especially around the incentive for miners to extract more MEV or passive collusion through a possible vicious Flashbots software update to defeat the new fee mechanism. There is no doubt that providing such an update is not in the interests of Flashbots, but the fact that 85% of Ethereum’s computing power is running MEV-geth requires us to think more about its impact.

Finally, we point out that although the design of EIP-1559 only includes a transaction package bidding mechanism, there are now many different bidding mechanisms on Ethereum. If we want to achieve the openness of Ethereum, it is important to recognize the above points and design our system to apply these different auction mechanisms.

We would like to thank the Barnabé Monnot and Robust Incentive groups of the Ethereum Foundation, Leo Zhang, Tim Beiko, Tina Zhen, and the entire Flashbots team for participating in the dialogue, which made the ideas presented here.

1. In our previous article on quantifying MEV extraction, the definition of actual extractable value (REV) was given.

2. Dogecoin, Ethereum Classic, ZCash, Dash, Monero, Bitcoin Gold and Verge are all included in the calculation of chain X revenue.

3. At the time of writing, these values ​​are calculated as follows:

MEV and EIP-1559

4. Other conditions remain the same. If we assume that the burned gas fee in the transaction bundle will be equal to the burned gas fee in the tail transaction of the squeezed block, then the basic cost of burning will not reduce the actual extractable value.

5. However, please refer to this note to learn how to limit the risk of uncle blocks in an environment with a large number of MEVs.

6. The field of auction design has already had a wealth of research results on computability preferences/allocations, see Tim Roughgarden’s lecture notes for getting started.

Article author: Kristof Gazso (Nethermind), Alejo Salles (Flashbots)


Posted by:CoinYuppie,Reprinted with attribution to:
Coinyuppie is an open information publishing platform, all information provided is not related to the views and positions of coinyuppie, and does not constitute any investment and financial advice. Users are expected to carefully screen and prevent risks.

Leave a Reply