153 total views
Since the transaction sequence process in Ethereum 2.0 will be the same as the current PoW Ethereum, we have reason to believe that MEV opportunities will still exist in Ethereum 2.0.
Original title: “There will also be MEV in eth2?” How will the verifier’s revenue be affected? 》
Written by: Alex Obadia & Taarush Vemulapalli
Editor: South Wind
Ethereum will soon transition from PoW to PoS consensus protocol. Developers have been committed to achieving this transition for several years, and in multiple steps. The first step is to launch the beacon chain in December 2020. The current beacon chain is now online. At the time of writing, there are already more than 160,000 validators on the beacon chain, which is equivalent to about 500 pledges. Ten thousand ETH.
The second step of “big merger” may happen in early 2022. Although there are still many details to be resolved beyond this step, there are enough details about PoS Ethereum (ie eth2) that have been resolved, which allows us to infer the maximum extractable value (MEV, or Maximal Extractable Value, Formerly known as “miner extractable value”) in eth2.
In this article, we will study the transaction sorting in eth2 and analyze the increase in verifier revenue brought about by MEV value extraction. We found that MEV will significantly increase the rewards obtained by validators, but it may also exacerbate the inequality (revenue) among eth2 participants . We will also discuss the qualitative aspects of MEV in eth2, such as the potential dynamics between the largest stakeholders such as exchanges and validator pools (ie, staking pools).
This article was co-authored by Alex Obadia and Taarush Vemulapalli. For complete analysis documents, see:
01. Overview of eth2
At present, the consensus of Ethereum is achieved by miners who run mining hardware, which is optimized to better solve PoW challenges. The transition from PoW consensus to PoS consensus means that the Ethereum network will be protected by verifiers (not miners). Each verifier node needs to pledge a deposit of 32 ETH and vote to reach a consensus on the state of the beacon chain. The validator has economic incentives to do so, that is, the validator’s good behavior will be rewarded, and offline or malicious behavior will be punished (penalty).
Currently, the beacon chain runs in parallel with the eth1 chain, and the beacon chain has been successfully running since December 2020. The “big merger” will merge the beacon chain with the current eth1 chain. In this article, we will use ” eth1 ” to refer to the Ethereum execution engine containing blocks and transactions; use ” beacon chain ” to refer to the new underlying PoS consensus mechanism of eth2 ; use ” eth2 ” to refer to the merged Ethereum Authority chain, this chain includes the eth1 execution engine and the beacon chain used to achieve consensus .
eth2 reached consensus in 6.4 minutes (called an epoch) increments. Each epoch contains 32 slots, and each slot has a duration of 12 seconds. Each slot represents the opportunity for a block to be added to the beacon chain. Under normal operation, each slot will generate a block, but reasons such as the verifier offline may cause some slots to generate 0 blocks.
For each epoch, all verifiers are to be pseudo-randomly assigned to block proposals (propose block) or on other proposed by the verifier blocks proofs (attest to blocks), the proposed block verifier called ” Proposer “, the verifier who proves the block is called ” prover “. There will be only one proposer and multiple provers during each slot, and these provers will be responsible for proving all the information in the block, including data from eth1 and data from the beacon chain. The prover obtains rewards by correctly voting on the current values of the “three aspects” of the beacon chain. The three aspects are: the chain head of the blockchain (the top block) and the proven Checkpoints and finalized checkpoints.
Note: The last slot of each epoch is called a checkpoint. When two consecutive epochs are justified, then the previous epoch can be regarded as finalized. For details, please refer to 👉 “Ethereum 2.0: How to achieve finality?” “.
02. MEV in eth2
MEV (Maximum Extractable Value) is all possible value obtained by block proposers by reordering, reviewing, or blocking transactions in their proposed block . In order to understand the ordering of transactions in eth2, let’s first understand the internal workings of the software used to order transactions (that is, the eth2 client).
1. eth2 client
Since eth2 is essentially two chains merged together (that is, the eth1 chain and the beacon chain), it is not surprising that the eth2 client consists of two “sub-clients”: one of them is the execution The engine client, and the other is the consensus client. It is worth noting that the current PoW Ethereum client (ie eth1 client ) will continue to exist in eth2 and run with the beacon client , with different divisions of labor.
Among them, the eth1 client in eth2 strips off its consensus responsibility from the current PoW Ethereum client , and only focuses on the transaction pool, eth1 execution and EVM of the eth1 chain ; while the beacon client is responsible for consensus and assigning verifiers Responsibilities (such as proof and proposal for beacon blocks). These two clients run in parallel , each maintaining its own p2p network stack (the beacon client maintains libp2p, and the eth1 client maintains devp2p).
The eth2 client may look like the modified diagram below (from the article written by Danny Ryan )
2. Block proposal for eth1
Just like in the current PoW Ethereum, the eth1 client in eth2 will maintain a local transaction pool (mempool) that contains transactions received from its p2p network. As described in the Rayonism specification , the beacon client will interact with the eth1 client to form an eth1 block . Although the details of the communication path (between two clients) in the specification may change in production, the general approach is likely to remain the same:
After many rounds, the beacon client queries the eth1 client for a certain eth1 block formed by the eth1 transaction pool, and checks whether the block meets various validity conditions;
Once this eth1 block is received by the beacon client and satisfies various validity checks, the eth1 block will be packaged by proposers into the current beacon block and become attesters Part of the data to vote.
Then the beacon client will ask the eth1 client to update the chain head (the top block) of the eth1 chain to this latest packaged eth1 block;
After a while, the epoch containing the beacon block will be finalized , and then the beacon client will inform the eth1 client that this eth2 block has been finalized at the consensus layer.
Although the way of reaching consensus in eth2 has changed, the ordering of transactions in each eth1 block in eth2 is the same as today . They are all in the software for ordering transactions (such as the PoW Ethereum client Geth) and the p2p transaction network Achieved.
3. Does MEV exist in eth2?
Since the transaction ordering process in eth2 will be the same as the current PoW Ethereum, we have reason to believe that MEV opportunities will still exist in eth2 , as we have seen in PoW Ethereum today. The difference lies in who has the ultimate control over the sorting, that is, in eth2, the verifier (not the miner) will have control over the transaction sorting . The verifier is selected to propose a beacon block, and the beacon block will It will contain a new eth1 block that was queried from the eth1 client.
This means that something like Flashbots’ MEV-geth (a modified eth1 client software designed to optimize the extraction of MEV) allows eth1 transaction senders to tip block proposers (and transaction sequencers) The technology to achieve the transaction sorting you want will still exist. After clarifying this proposition, we can now think about how much money the verifier can make by running software like Flashbots?
03. Reward mechanism for validators
Although MEV is notoriously difficult to measure, we use Flashbots data as the lower limit of the minimum additional income that the proposer of the eth2 block can obtain through MEV. This is a lower limit of revenue, as only a small portion of MEV activity occurs on Flashbots.
One caveat of the analysis in this article is that this article considers MEV based on the staking income specified in the eth2 protocol, but does not include the transaction fee rewards that block proposers can obtain. The main reason for not including these transaction fees is that it is difficult to predict how much the proposer will earn from transaction fees after EIP-1559 (EIP-1559 will introduce a basic transaction fee basefee destruction mechanism).
1. Ideal situation
Let us first consider an ideal situation where all validators participate perfectly and get the largest agreement reward (ie there is no large-scale penalty), and all staking rewards are evenly distributed, because all validators are on an infinite time scale The same number of blocks are proposed above.
Figure 4: The Y-axis represents the rate of return, and the X-axis represents the number of validators. The blue line represents the annual return rate of the validator in an ideal situation without considering the MEV; the yellow line represents the annual return rate of the validator in the ideal situation with the MEV considered. The blue vertical dotted line indicates the number of verifiers at the time of writing (approximately 160,000 verifiers).
Based on the current number of validators (160,000 validators), we found that MEV can increase the rewards of validators by 75.3% , or provide an APR (annualized interest rate) of 12.83% , which is higher than if the MEV is considered. The 7.35% APR income from staking ETH. One conclusion that can be drawn from this is that a higher validator reward means that more ETH holders will be attracted to become validators, which in turn means that Ethereum has a larger set of validators and becomes more secure .
As more verifiers go online in the near future, the increase in verifier revenue based on MEV will be less significant. For example, the reward for 250,000 verifiers (ie 8 million ETH pledged) will only increase by 60% . As mentioned above, this analysis does not consider how much transaction fees the verifier will receive, as this will reduce the relative impact of MEV on revenue. However, compared with the additional MEV rewards currently earned by PoW miners through Flashbots (currently about 5.6%), these data are still useful. This obvious difference stems from the significant drop in the issuance rate of PoS. This shows that in eth2, the withdrawal of MEV will be more worthwhile than in eth1, and stakers may vigorously promote the staking income realized through MEV.
2. Take into account the time factor & REV allocation
On any limited time scale, the rewards of validators are variable, because the proposed block has a specific protocol reward, and because some validators will be lucky enough to have the opportunity to propose more blocks than the average number, and Some less fortunate validators will propose fewer blocks .
For example, if there are 100,000 validators in the network, the average number of blocks proposed by each validator per year is 26 blocks, and the most unfortunate 1% of validators have the opportunity to propose 15 blocks at most , the luckiest 1% of validators propose at least 39 blocks. See below:
Based on this logic, we can draw the validator staking reward based on 3 different levels of block proposal “luck” (ie the luckiest 1% validator, the most unfortunate 1% validator, and the average validator) The variability (not considering the influence of MEV):
Figure 6: The Y-axis represents the rate of return, and the X-axis represents the number of validators. The green line represents the staking annual return rate that the luckiest 1% validator can get without considering MEV; the red line represents the staking annual return rate that the most unfortunate 1% validator can get without considering MEV; the yellow line Indicates the average annual return rate of staking that a validator can obtain without considering MEV. The blue vertical dotted line indicates the number of verifiers at the time of writing (approximately 160,000 verifiers).
Now, we add the average extracted value ( REV , Realized Extractable Value) of each block recorded on Flashbots , we can compare these 3 different levels of block proposal “luck” considering MEV Value and the rate of return of the validator without considering the value of MEV:
Figure 7: The green line indicates the annual rate of return that the luckiest 1% validator can obtain with MEV value extraction; the red line indicates the annual rate of return that the most unfortunate 1% validator can obtain with MEV value extraction ; The yellow line indicates the average annual rate of return that the verifier can obtain if the MEV value extraction is included. The bottom thick line is the annual staking rate of return that the validator can obtain in the three block proposals “luck” without considering MEV, but because the difference between the three lines in the above figure is too small, so the three lines Overlapped.
The three curves in the above figure (Figure 7) are used to represent the annual staking yield curves of validators brought about by the three “luck” levels that do not consider the value of MEV almost overlap and are difficult to distinguish. This shows that the extraction of MEV value expands the inequality of benefits between validators brought about by the “luck” of the block proposal .
In addition, the distribution of REV is uneven and can be regarded as the second dimension of “luck”, that is , some blocks have larger MEV rewards than others . For example, the following is the (long tail) distribution of REV rewards received by miners using Flashbots’ MEV-Geth mining in the recent 100,000 consecutive blocks of Ethereum (starting from block height of 11600000):
In the above figure (Figure 8), we cut the X-axis (the REV value actually obtained by the miners in each block) to 3 ETH, but in fact, in our sample, the miners can get a maximum of 101 ETH REV value. Using this distribution of Flashbots miner rewards to represent the distribution of REVs, we can define and plot 3 based on the income of the most unfortunate 1% of validators, average validators, and the luckiest 1% of validators from MEV rewards. Annual yield curve for each luck level:
Figure 9: The green line represents the annual rate of return that the most fortunate 1% validator can obtain with MEV value extraction; the red line represents the annual return that the most unfortunate 1% of validators can obtain with MEV value extraction Rate; the yellow line represents the average annual rate of return that the verifier can obtain with the MEV value extraction included. The blue vertical dotted line indicates the number of verifiers at the time of writing (approximately 160,000 verifiers).
The previous figure (Figure 7) shows us that MEV enlarges the inequality of benefits between validators brought by the block proposal “luck”; this figure (Figure 9) shows the difference of REV. Uniform distribution is a greater source of income inequality between validators , especially considering that the Y-axis in this graph (Figure 9) has grown to 600%, while the Y-axis in Figure 7 is only 80%.
However, in reality, the verifier will be eliminated from block proposals luck and REV the uneven distribution of the differences brought about by the convergence of income verifier in the verifier resource pool (validator pools, also known as the pledge pool) . But this means that the impact of MEV on verifiers’ income may inhibit people from independently running verifier nodes, making joining a validator pool more attractive in terms of financial incentives, which may lead to the centralization of network verification .
Ultimately, we worry that MEV may aggravate the oligopoly dynamics in eth2, because entities with the most ETH staking will grow faster than those with less ETH staking (validator pools). This will make the democratization of MEV extraction particularly important in eth2, so as to maintain the decentralization of consensus voting rights.
04. New consensus participants
Although the above quantitative analysis is important to start thinking about MEV in eth2, this article is incomplete without a qualitative analysis of the participants in the eth2 consensus. As mentioned earlier, in eth2, miners and mine pool will be in control of a large number of ETH entity (such as Exchange, library funding agreements, investment funds and verifier pool) replaced. This can already be seen from the distribution of the current eth2 validator’s eth1 deposit address displayed on the beacon chain browser beaconcha.in:
Above: The distribution of eth1 deposit addresses of all validators on eth2. It can be seen that a large number of eth2 validators are pledged through the addresses of entities such as several exchanges and pledge pools, which means that these entities control a large proportion of eth2 voting rights .
It is worth noting that this pie chart does not distinguish between the final entity that controls the voting rights of the consensus and the infrastructure it runs on. Although the centralization of eth2 consensus voting rights is worrying, the centralization of infrastructure may not be the case. PoS economic incentives encourage the decentralization of infrastructure to minimize related slashing risks.
Specifically, this means that an exchange like Kraken that controls a large number of (users) eth may spread the (user’s) deposit to many infrastructure providers, and run eth2 nodes in different regions and on different hardware. Rather than undertake this huge infrastructure operation task internally, thereby reducing the risk of substantial confiscation.
The most noticeable change in eth2 is that the exchange has become the largest ETH holder and therefore the largest validator . Centralized companies such as Coinbase, Binance, and Kraken may control the largest number of validator slots . These participants are subject to different rules than mining pools, which can affect their reputation in many ways. Compared with the pattern of miners, this difference may have a new impact on the pattern of validators, and may affect the activities that validators participate in, such as the type of MEV in which they earn revenue.
Interestingly, in addition to participating in the eth2 pledge, these entities also participate in a number of activities, which may bring new opportunities for synergy between the existing services provided by these exchanges and the extraction of MEV value. These activities include speeding up transactions, providing private withdrawals before the withdrawals are packaged on the chain, and reducing on-chain transaction fees through encrypted native payments for the order flow, and so on.
Such services may be the most cutting-edge at first, and their benefits may mean that users will migrate to exchanges that provide these services, which may harm those exchanges that do not or cannot provide these services due to regulatory reasons. In addition, the potential vertical integration of exchanges in the MEV game (for example, exchanges running their own robots to submit transactions to their own validator nodes) is a problem worthy of attention, and we think it should be further studied.
2. Validator pool
Another important change is eth2 verifier pool (validator pools, also known as “collateral pool”) of the rise, these pools provide minimum requirements such as reducing the number of ETH participation eth2 pledge (to verify the user’s own running a single node needs Stake 32 ETH, and joining the validator pool can provide less than 32 ETH, because the validator pool will pool all users’ ETH for pledge), build validator nodes for customers, and eliminate “luck” due to block proposals (Which will affect the MEV+ transaction fee income) and the benefits of providing additional services such as staking derivatives (thanks to the fund base they manage).
An interesting phenomenon is the emergence of meta-pools, such as Rocketpool and Lido. These entities are connected to many validator pools and are likely to become a major source of the staking amount of these validator pools. Therefore, they can exert influence on the behavior of the validator pools, such as the types of MEV withdrawals that the validator pool participates in and their pledge Profit sharing provided by the provider.
These meta-pools usually provide pledge derivatives . An example of this is to provide users with a liquidity tokenized version of their locked ETH pledge deposits, which users can use in the (Ethereum) network. By allowing users to reuse the locked ETH in DeFi in the form of derivatives, this will further increase the verifier’s income beyond the value of MEV.
05. Open-ended questions
Our exploration of MEV in eth2 found many unresolved issues, and we plan to conduct research in the next few months. Here are four of them:
1. eth1 block proposer market
Since eth2 actually has two clients to run (eth1 client + beacon client), it is likely that independent verifiers will choose to default their eth1 node to a service provider, such as Infura, because the eth1 node is running The cost itself is very large. This may imply that the eth1 and eth2 node operators are beginning to separate. Assuming such a dynamic emergence, we can imagine the emergence of an eth1 node operator market, which runs high-performance hardware and MEV simulation software to meet the needs of eth2 block proposers.
2. New limitations when optimizing MEV search
MEV opportunities such as price arbitrage and liquidation still exist in eth2 , but the system for extracting the value of these MEVs has new parameters, which may modify or introduce constraints on the extraction of MEV.
For example, the block generation time of eth2 is fixed at 12 seconds, instead of the current block generation time of eth1, and the slots of the block proposer are allocated at the beginning of each epoch, which means that the proposer will be able to There is 6.4 minutes to calculate their tasks (of course, proposers of slots assigned at the beginning of the epoch do not have that long time). This not only provides potentially more time for validators to run calculations on the eth1 client transaction pool to obtain the best MEV extraction, but also makes simulation and execution easier due to the predictability of the block time.
This means that there are longer and more predictable time intervals to calculate and execute MEV extraction strategies, thereby enabling more complex and computationally intensive MEV extraction.
3. Changes in the leader selection mechanism
Verifiers will know in advance whether they need to propose a block (unless it is the first slot of a new epoch). They can even (although the probability is very low) propose multiple blocks in an epoch. How will the determination of the identity of the block proposer change the dynamics of MEV value extraction? And if it is determined that multiple blocks will be proposed in one block, how will this affect the dynamics of MEV extraction? In particular, large validator pools/exchanges are most likely to have (assigned to) multiple consecutive slots within the same epoch.
4. L2s & Sharding
Most of the content in this article assumes that the block content of eth1 will remain as it is today. However, in reality, many transaction flows will be transferred to L2s, Ethereum L1 will be used as the data availability layer, and zk-rollups and optimistic rollups will be responsible for submitting batches of packaged transaction data to L1.
This will intuitively reduce the verifier’s revenue from MEV . However, this is difficult to predict, because the world of multiple L2s brings additional complexity and may open up new forms of MEV (that is, cross-L2 transactions, cross-L1-L2 transactions). Similarly, with the continuous development of eth2 and the production of shards, the ordering of shards in the beacon block may be of great significance. MEV may become the realization of Vitalik’s proposal to “stagger the shards to achieve faster Block generation time”  incentive mechanism.
Thanks to Terence Tsao, Raul Jordan, Alejo Salles, Luke Youngblood, Tomasz Stanczak, Lakshman Sankar, Barnabe Monnot, Caspar S, and Viktor Bunin for their valuable contributions and editing of this article. Thanks also to other members of the Flashbots team for discussions.
Links involved in the main text: