Keynote: The REAL state of L2s by Bartek Kiepuszewski | Devcon SEA

Bartek Kiepuszewski, founder of L2B, discusses the advancements in the Layer 2 (L2) ecosystem, noting the launch of nearly 50 roll-ups and over 50 LDA L2s, along with a significant total value locked (TVL). He acknowledges the foundational contributions of early pioneers in zero-knowledge technology, Plasma, and Arbitrum.

Kiepuszewski emphasizes the importance of security in L2s, highlighting that users often mistakenly believe they inherit Ethereum's security. He raises concerns about the lack of proof systems in many current L2s, particularly those launched with minimal security measures like multisig. He predicts that many upcoming L2s, especially OP stack chains, will also lack essential security features.

The discussion includes the role of security councils, which require a minimum of eight participants and a 75% consensus threshold. Kiepuszewski points out the need for geographic decentralization among council members, noting that only a few projects, such as Arbitrum, Optimism, Polygon, and ZK Sync, have established security councils. He critiques the poor data availability trust assumptions in many decentralized applications, using Arbitrum and Immutable X as examples.

Kiepuszewski outlines three reasons to prioritize proof systems: they allow permissionless exits for users, enhance trust by making malicious proposals harder, and reduce reliance on intermediaries. He mentions the Nethermine team's plans for a roll-up designed as a stage two system from the outset. He distinguishes between security councils and simple multisigs, emphasizing that multisigs lack the necessary social layer for true decentralization.

The vulnerabilities of multisig systems are discussed, citing recent attacks and the risks of relying on them without proper verification. Kiepuszewski notes the security implications of different roll-up stages, indicating that stage zero roll-ups without security councils are particularly risky, while stage two roll-ups offer improved security.

He proposes a new classification system for roll-ups based on proof systems, data availability guarantees, and security measures, aiming to transparently represent the security status of various L2s. This system would promote stage one and stage two projects while making information about less secure systems more accessible.

L2B's objective is to ensure users are informed and accountable for the systems being built, with a commitment to analyzing proof systems for soundness and completeness over the next year. This analysis will focus on both Zero-Knowledge (ZK) and optimistic proof systems, reviewing source code, trusted setups, and verification processes on-chain, while emphasizing transparency.

Kiepuszewski warns against creating insecure side chains and stresses the importance of launching systems that inherit security from Ethereum. While decentralizing sequencers may not be critical for user fund security, it is vital for real-time censorship resistance, leading to potential new designs, including centralized sequencers with escape hatches.

Current TEE-based proof systems, such as those using SGX, are not classified as stage two due to trust issues with hardware manufacturers. Efforts are underway to develop more trustless and permissionless systems. Optimism is classified as stage one, meeting criteria related to fund control and exit strategies, with centralized proposals still allowing for permissionless exits if they fail.

The criteria for stage designations are dynamic and may evolve as systems develop, requiring ongoing evaluation of leading projects like Optimism and Arbitrum. Challenges in advancing to stage one include complexity, time, cost, and talent shortages. Pioneers of ZK roll-ups, such as DYDX, were early builders of application-specific solutions, but the governance process for stage two is currently lacking. Although the stage two system was essentially built four years ago, most roll-ups today are launched by service providers rather than original development teams, highlighting the need to incentivize these providers to improve their performance. Acknowledgment is given to Mr. Badek and his research team's contributions.