TL;DR
This paper provides a comprehensive longitudinal study of Ethereum's block construction market post-Proof-of-Stake (PoS) transition, focusing on the Proposer-Builder Separation (PBS) mechanism. It reveals that while PBS has become dominant, the market exhibits significant MEV payment disparities and potential conflicts of interest, with crises like the FTX collapse dramatically amplifying MEV revenues.
What Problem Does This Paper Address?
The transition of Ethereum from Proof-of-Work (PoW) to Proof-of-Stake (PoS) introduced the Proposer-Builder Separation (PBS) model, aiming to decentralize block production and reduce the power of individual block producers in extracting Miner Extractable Value (MEV). MEV refers to the profit gained by strategically ordering, including, or excluding transactions within a block. While PBS, particularly through MEV-Boost, has become the standard, its actual impact on market dynamics, decentralization, and potential for new forms of value extraction remains unclear. This paper investigates the structure of this new market, the flow of value, and the implications of its design, particularly concerning potential conflicts of interest and the concentration of power.
Methods
The authors conducted a longitudinal study of the Ethereum block construction market from September 2022 to May 2023. They collected data from Ethereum execution and consensus layer nodes, as well as from the API endpoints of various MEV-Boost relays. This allowed them to analyze:
- Market Share Analysis: Tracking the market share of builders, relays, and proposers over time.
- Financial Flow Analysis: Quantifying the ETH transferred between builders, proposers, and relays, often referred to as "bribes."
- MEV-Time Law: Developing a model to quantify expected MEV revenue based on the time elapsed since the last proposed block.
- Crisis Event Analysis: Examining how significant market events (e.g., FTX collapse, USDC de-peg) correlate with spikes in MEV payments.
- Network Topology Visualization: Mapping the connections between builders, relays, and proposers.
Datasets
The study utilized data collected from Ethereum execution and consensus layer nodes. Additionally, they accessed data from the "builder_blocks_received" and "proposer_payload_delivered" endpoints of active MEV-Boost relays. The final dataset comprises 1,624,885 blocks from the launch of PBS until May 1st, 2023. The authors also make their raw dataset available at https://mevboost.pics/data.html.
Key Insights and Ideas
- MEV-Boost Dominance: MEV-Boost quickly became the dominant PBS implementation, capturing around 90% of block proposers' activity within months of its introduction.
- Market Diversity: Despite MEV-Boost's dominance, the builder market remains relatively diverse, with no single builder consistently holding more than 20% market share.
- Relay Concentration and Shift: Flashbots initially dominated the relay market but saw its share decrease significantly by May 2023, with other relays' market shares becoming more aligned with builders.
- Extreme MEV Payment Inequality: A small fraction of MEV transactions account for a disproportionately large share of the total revenue. Specifically, 20% of payments make up 72% of the total value, and 1% of transactions exceed 1.36 ETH. This suggests a "lottery" effect for proposers.
- Crisis Amplifies MEV: Major market shocks, like the FTX collapse and the USDC de-peg, led to dramatic, multi-day spikes in MEV revenues (400% and 1000% respectively), highlighting MEV's sensitivity to market instability.
- MEV-Time Law: The study identifies an optimal waiting time of approximately 2.78 seconds within a slot for proposers to maximize MEV, yielding a 107% higher bid value compared to an earlier baseline. However, this introduces risks of block reorganization.
- Proposer Timing Strategies: Different proposers exhibit distinct timing strategies for submitting bids, with some (like Bitcoin Suisse) being very consistent and others (like Kraken) more flexible.
- Potential for Vertical Integration and Conflicts: The paper raises concerns about the potential for vertical integration between builders and relays, and the possibility of builders and relays extracting MEV from searchers, which could undermine the decentralized ethos of PBS.
Key Results
- MEV-Boost Market Share: Approximately 90% of block proposers adopted MEV-Boost.
- Total ETH Transferred: 160,000 ETH (approximately $320 million USD) was transferred to block proposers via MEV-Boost during the study period.
- MEV Payment Distribution: 80% of lower MEV payments account for only 28% of the total value, while 20% of payments capture 72% of the revenue.
- Crisis Impact: FTX collapse led to a 400% increase in MEV revenue for several days; USDC de-peg led to a 1000% increase.
- Optimal MEV Timing: An optimal waiting time of 2.78 seconds in a slot can yield a 107% higher bid value.
- Builder Diversity: No single builder consistently held over 20% of the market share.
- Relay Market Shift: Flashbots' relay market share dropped from over 60% to around 25% by May 2023.
Limitations and Future Work
- Implicit Trust: The PBS model relies on implicit trust between proposers, builders, and relays, which contradicts the decentralized ethos of blockchain. The paper questions the justification for this trust, especially given the potential for builders/relays to operate their own searchers.
- Regulatory Concerns: The ability of MEV-Boost to enable "risk-free sandwich extraction" raises concerns for regulators.
- Attestation Data: The analysis of MEV-timing did not include attestation data, which is necessary to fully understand the trade-off between optimizing bid value and ensuring block validity.
- Future Work: The authors suggest further investigation into the optimality of the PBS design and its regulatory implications, particularly concerning private order flow. They also highlight the need for a comprehensive understanding of MEV payment mechanisms for the evolution of the Ethereum blockchain. The paper also notes that while PBS has seen widespread acceptance, it hasn't been universally adopted across other blockchain networks, suggesting potential for alternative MEV auction designs.
Summary updated