Introduction to Lean Ethereum with Justin Drake

Anna Rose introduces a six-part mini-series on Lean Ethereum, hosted by Nico and featuring Justin Drake from the Ethereum Foundation. They discuss Lean Ethereum's structure, which includes the consensus, data, and execution layers, emphasizing the role of ZKEVMs in enhancing throughput and gas limits. Justin highlights the limitations of current systems using BLS signatures and KZG, which are not post-quantum, proposing a complete rewrite of the consensus layer to eliminate technical debt and ensure compatibility with SNARKs for real-time processing. This aims to facilitate trustless wallets and bridges, enhancing security and validation efficiency for low-resource devices.

The conversation delves into the challenges of post-quantum signatures, noting that while BLS signatures are compact, post-quantum alternatives are larger and less flexible for aggregation. Justin introduces SNOX to tackle the segregation problem in signature aggregation, ensuring compatibility across layers. LeanVM is presented as a minimal ZKEVM designed for hash-based cryptography, with applications across all layers and potential use in other blockchains. LeanVM features a custom instruction set architecture optimized for hashing, addressing consensus challenges related to large signature aggregation.

The discussion also covers the use of off-chain proofs for flexibility and the importance of formal verification in ensuring the security and correctness of the consensus layer. Alex Hicks leads a $20 million initiative for formal verification, focusing on soundness and analyzing known attacks. The conversation centers on the transition from conjectured to provable proximity gaps in the Ethereum ecosystem, highlighting a $1 million bounty aimed at bridging this gap, alongside a separate $20 million formal verification bounty. Justin emphasizes the significance of proving both theorems and code correctness in developing a Zero-Knowledge Virtual Machine (ZKVM), stressing the need for accurate arithmetization corresponding to the Instruction Set Architecture (ISA) and the concept of end-to-end formal verification across multiple layers of the ZKVM stack.

The introduction of LeanVM, which will undergo formal verification, is discussed, along with the candidate hash function Poseidon2, noted for its snark-friendly design and over a 10x performance improvement compared to traditional hash functions like SHA256. The community's historical baking time for new hash functions is estimated at eight years, but efforts are underway to expedite Poseidon's maturation through workshops and a natural bounty program linked to ZK roll-ups. A $1 million Poseidon prize is announced for anyone who can break the hash function, although challenges in formally verifying its security are acknowledged.

The conversation also addresses throughput and network capacity, aiming to maintain current throughput levels despite larger signatures, with a target of 32,000 validators casting votes per slot. A new aggregation strategy is proposed, involving the division of attesters into subnets for initial aggregation and exploring recursive aggregation to enhance participation and mitigate censorship. The goal is to enable a million validators in a proof of stake system with a low minimum deposit.

Current aggregation methods using BLS are discussed, revealing constraints and inefficiencies in post-quantum contexts. The upgrade to a data layer utilizing post-quantum blobs is proposed, incorporating erasure coding and Merkle trees to prove correctness within LeanVM. Efficiency concerns regarding snarkifying erasure coding and Merkle trees for large blobs are raised, with aspirations for a teragas per second era supporting 10 million transactions per second across Layer 2s.

Future topics for the series include signatures, security in post-quantum migration, LeanVM implementation, networking, and formal verification. Justin notes that no major aspects have been overlooked and highlights the challenges faced in the LeanVM project, including team composition. He emphasizes the lean structure of the Ethereum Foundation team, with significant contributions from AI and the primary developer Emil, who is capable of handling most R&D independently. Looking ahead, weekly post-quantum interrupt calls for collaboration on Lean spec implementations are planned, along with upcoming events such as a post-quantum day at EFCC in Cannes and a three-day workshop in October. The previous workshop's success in fostering collaboration is noted, with anticipation for the next workshop to focus on standardizing solutions applicable across various blockchains. Justin expresses gratitude for the opportunity to discuss Lean Ethereum and acknowledges the contributions of the podcast team and audience.