Equilibrium Infra Bulletin #34: Fairness in Mysticeti, Memory Efficient ZKPs, Crypto's Airtag Moment, and more...
Equilibrium Labs builds the state-of-the-art of decentralized infrastructure. We are a global team of ~20 people who tackle challenges around security, privacy, and scaling.
🔍 Notions of Fairness in Mysticeti
⚡️ Topic Summary
There’s recently been a lot of discussion around multiple concurrent proposers (MCP) as a solution to reduce harmful MEV and censorship-related issues on Ethereum. DAGs naturally rely on MCPs and inherently delivers many of the benefits discussed. The core idea of DAGs is that all validators contribute to building the DAG through a series of rounds, where each honest validator proposes a single uncertified block (rather than relying on a single entity for data dissemination).
Overview of data dissemination and construction of the DAG The article explores how Mysticeti - a DAG-based BFT consensus model developed by Mysten Labs and integrated into Sui - increases fairness (expanding on previous research by Raikwar et al). However, the notion of fairness depends on from who’s perspective we look at it:
Users (clients) want their transactions to be included in a timely manner without experiencing value leakage or negative extraction.
Mysticeti only requires users to send their transactions to a single validator, which enables potentially lower latency (targeting closest validator geographically) but also increases the risk of censorship (lower fairness).
While users have the option to send transactions to multiple validators, there is no concept of garbage collection in Mysticeti that allows transactions to be re-injected into the DAG.
There is also no public mempool and instead transactions are sent directly to validators. This minimizes exposure to front-running and other (negative) MEV exploits.
Validators want fair participation and equal opportunities to propose blocks to the chain - regardless of differences in networking latency or their competitiveness in hardware specs.
Mysticeti increases fairness for slower validators by requiring a sufficient amount of references to recent blocks (2f+1) as well as enabling references to blocks from all prior rounds (“weak links”).
The move from a certified to an uncertified DAG also reduces the required communication between validators (less communication overhead) and lowers CPU usage (better hardware utilization and potentially opening the door to lower hardware requirements).
🤔 Our Thoughts
We’ve written about DAGs before and been closely following the innovation in the consensus space (DAG-based BFT protocols) along with hands-on work implementing Narwhal-Bullshark in Aleo. With the recent Raptr announcement from Aptos, it doesn’t seem like innovation is slowing down.
While the article focuses on the fairness aspect, it does also cover some fundamentals for Mysticeti. However, for those interested to learn more, we recommend watching this talk or reading the original Mysticeti paper.
One of the benefits of this space is the broad range of different implementations across protocols. As the multi-proposer architecture underpinning Mysticeti and earlier DAG-based protocols is gaining recognition as a viable solution to broader challenges regarding censorship resistance in blockchains, we should learn from previous implementations rather than attempt to re-invent the wheel.
💡 Research, Articles & Other Things of Interest
📚 Minimal Space, Maximum Pace: How memory efficient zero-knowledge proofs work: An overview of different schemes and approaches used in client side proving.
📚 Crypto’s AirTag Moment: Unlocking Mass Adoption with Web Proofs: How web proofs could revolutionize airdrops, incentives, and marketplaces by vastly expanding the addressable market.
📚 zkTLS Canon: A collection of resources for anyone looking to dive deeper into Web proofs (zkTLS).
📚 Overcoming Security Risks in zkSharding: Exploring the security concerns in sharded systems before full ZK finalization, defining the problem of a corrupted state, and proposing a protocol to solve it.
🔥 News From Our Partners
🥳 Aleo launches on mainnet: After working closely with the Aleo team for several years, the day is finally here. Congrats to the whole team and very exciting to see privacy in blockchains take a huge leap forward!
🤌 Personal Recommendations From Our Team
📚 Reading: Freezing Order - Bill Browder: A story of money laundering in Russia and an act to sanction oligarchs that has been implemented in 34 countries so far (Magnitsky act).
🎧 Listening: The Emptiness Machine - Linkin Park: Linkin Park is back after 7 years of silence following the death of Chester Bennington, presenting new music and their new lead singer Emily Armstrong.
💡 Other: Is It Art or Not? Finnish AI entrepreneur turned artist Roope Rainisto comments on the use of AI in art, suggesting it's more about human creativity than machine output.