DLT Interoperability and More ⛓️#28 — SoK: Cross-Domain MEV⛓️
In this series, we analyze papers on blockchain and interoperability.
This edition covers a survey paper on cross-chain MEV.
➡️ Title: SoK: Cross-Domain MEV
➡️ Authors: Conor McMenamin
➡️ Paper source: https://arxiv.org/pdf/2305.16468.pdf
➡️ Motivation:
As interoperability trends progress and mature, cross-chain MEV may be an additional source of revenue for miners and validators alike. As token bridges and interoperability mechanisms (including arbitrary message passing bridges) develop and gather interest from the community, we are laying out the foundations (messaging and coordination protocols) for building a variety of infrastructure and services, such as bridge aggregators and cross-boundary MEV solutions (and the respective searchers, builders, new wallets, validator software implementations).
“MEV is a complex systemic challenge for blockchains with various potential solutions, implementation paths, and externalities” [1], which brings a lot of complexity for one blockchain alone. Studying the cross-model of decentralized applications running in different blockchains could provide a map of exploitable dependencies by a set of colluding validators operating on different blockchains. As sophisticated actors will invariably try to exploit these dependencies, it is important to attempt to mitigate the negative impacts of MEV (e.g., reducing the likelihood of a winner-takes-all scenario).
➡️ Background:
We introduced a background on proposer build separation (PBS) and maximum extractable value (MEV) in a previous article of this series. Essentially, MEV is extracted by ordering transactions such that inserting, removing, or reordering on-chain transactions yields a positive sum for the extractor. The authors also refer to signaling, where off-chain information informs possible profitable transactions (e.g., positive sentiment on project X inspired by news headlines).
➡️ Contributions:
- Systematize prominent cross-domain protocols, focusing on the types of MEV that occur within each protocol.
💪 Strong points:
- the author focuses on how MEV is extracted and where the extractable value originates, providing practical insights. This is a nice complement to the 2021 paper that introduced cross-domain MEV.
- The paper does a great job of systematizing different mechanisms for mitigating negative MEV externalities / democratizing MEV. The main contribution of the paper is this table:
🤞 Suggestions for improvement:
- As a SoK paper, its main goal is not to give novel technical contributions (necessarily) but to systematically analyze the corpus of knowledge in a certain area. I would like to see practical searcher and bundler algorithms for the cross-domain scenario.
- It would be interesting to understand what is the current prevalence and the expected prevalence of each of the methods, broken down by ecosystem. Are there interesting insights we could collect from this data?
- Terminology is confusing: MEV terminology (bundler, searcher, validator) and rollup terminology (sequencer, proposer) are intertwined in the paper and might confuse the reader.
🔥 Points of interest:
- “Cross-domain MEV is an existential risk to the decentralization of blockchains. Cross-domain MEV is the value captured from arbitrage transactions executed in a specified order across multiple domains (blockchains L1/L2).” [1]. For exploiting opportunities that include the atomic processing of transactions in different domains, cross-chain MEV will be slightly different from one-domain MEV.
- searchers: will have to create cross-chain models for an MEV use case that spawns across multiple domains; the possible MEV revenue stems from optimizing a certain parameter of the cross-chain model with n domains (e.g., calculate the spread from one DEX to the next, and then to the next). The bigger the n, the more unlikely (and potentially profitable) the searchers operations are. Note that, for some blockchains such as Ethereum it could be possible to atomically sequence transactions across several blocks (aka MMEV, a term coined multi-block MEV, as pointed out by Jensen, von Wachter, and Ross). The implication is that now we can use a transaction in a block to, for example, swap A to B and bridge B, and the transaction on the following block to do the reverse (bridge B back and swap back to A). Of course, the time window is crucial, and modern bridges typically have high latencies. This particular type of exploration looks difficult to implement in practice but seems to be easier when collusive MMEV is in place — this is, there is a pool of proposers elected for n consecutive blocks. Simulations from Jensen et al.’s work show that 70% of the validators could have control of 8 consecutive blocks. For Ethereum to leverage bridging results, we would need bridging to another chain, performing operations, and bridging back to happen in around 1.5 minutes, which does not seem realistic. Interestingly, as observed in practice, Flashbots has allocated ten consecutive blocks. The mechanism captured in this paper that would allow extractors to propose blocks in different chains, at the same time, is the Slot Auction (no implementations yet).
- bundlers: Bundlers will keep receiving bids from searchers and pass to the validators the highest fee bundles. Bundlers that also control validators can capitalize on having control over n slots, and thus reorganize transactions.
- Miners and validators: “a large operator could validate on multiple chains and maintain a large inventory of tokens at any given time allowing for cross-chain atomic arbitrage. This is especially troubling if these cross-chain operators are consensus validators who, due to their sophistication, will earn more rewards than their peers and eventually dominate the MEV market as their stake weight in proof” [1] — a major infrastructure provider could be in a good position to develop searcher algorithms, due to their presence on the block validation market; and thus increase their competitive advantage against smaller players. This is the rationale for democratizing cross-chain MEV. A possible direction is to provide extensions to MEV-boost for the cross-chain scenario.
- The figure above is particularly interesting, and showcases what we discussed before. The longer the time interval where we are exploring MEV opportunities (i.e., the longer we can keep transactions atomic, i.e., the more consecutive blocks we can control), the higher the profit potential because we will be able to sequence more transactions, and therefore have a higher reordering capability. Nonetheless, the complexity of creating profitable transactions grows exponentially with the number of available blocks and domains. As such, these actions are very time-sensitive, and a market for optimizations arises.
- The authors point out two different types of extractable value. First, the Intrinsic-extractable value “is the expected value to the extractor at the time when the blockchain state must be acted on”. For example, if performing a sandwich attack on a DEX, the intrinsic-extractable value would be the profit that can be generated within the re-ordering time frame (including the time to create and send to the mempool a frontrun and a backrun transaction). On the other hand, Time-extractable value is a weighted expected value of all the possible different extractable value routes on a particular timeframe. Note that the extractor has no obligation to conduct any re-ordering (profit = loss = 0), thus, extracting value is an option that typically incurs a cost (at least gas fees).
- Common techniques for mitigating MEV (see Table above, and see the paper) are using a shared sequencer for multiple domains (e.g., Espresso, Astria), a single domain sequencer (e.g., Arbitrum), order-flow auctions (e.g., SUAVE), app-chains (e.g., Osmosis on Cosmos), and others.
🚀 What are the implications for our work?
- As a node provider, Blockdaemon supports Ethereum’s open-source client software and promotes decentralized, democratized MEV.
References:
