What happened
The Ethereum Foundation's Formal Verification team, in partnership with Yukon and zkSecurity, has officially launched better.codes. This initiative is an open autoresearch challenge designed to bridge the gap between high-level cryptographic research and automated security verification. The project focuses on a specific problem from the Proximity Prize research, formalized in the Lean programming language, and invites participants to push the boundaries of soundness for hash-based SNARKs through a public leaderboard.
Technology context
At the heart of this project lie SNARKs (Succinct Non-interactive Arguments of Knowledge) and Formal Verification. SNARKs are cryptographic tools that allow one party to prove they know something without revealing the actual data, which is vital for Ethereum's scalability (via ZK-Rollups). However, the mathematical proofs underlying these systems are incredibly complex.
Lean is a functional programming language and theorem prover. It allows researchers to write mathematical proofs that a computer can verify with absolute certainty. By moving these proofs into Lean, the Ethereum Foundation is enabling "agentic collaboration"—the use of AI agents and Large Language Models (LLMs) to assist or lead the discovery of formal proofs, ensuring that the cryptographic foundations of the network are mathematically unbreakable.
Why it matters
The security of Zero-Knowledge (ZK) systems is paramount for the long-term stability of the blockchain industry. Traditional manual audits are slow, expensive, and subject to human error. better.codes changes the paradigm by gamifying formal verification. By providing a structured, machine-checkable environment, it allows AI developers to test their models on real-world cryptographic problems. This accelerates the creation of secure protocols and lowers the barrier to entry for contributing to Ethereum's core infrastructure security.
Key terms explained
- SNARK: A cryptographic proof that is small in size and easy to verify, used to prove the validity of computations.
- Formal Verification: The act of proving or disproving the correctness of intended algorithms underlying a system with respect to a certain formal specification or property, using mathematical methods.
- Lean: A theorem prover and a programming language developed primarily by Microsoft Research, widely used in formalized mathematics.
- Agentic Collaboration: A workflow where human researchers and autonomous AI agents work together to solve complex problems, such as generating mathematical proofs.
Impact
In the short term, better.codes will produce more robust security bounds for hash-based SNARKs, directly benefiting projects building on Ethereum's Layer 2. In the medium term, it fosters a new ecosystem of "AI for Security," where automated tools can audit smart contracts and cryptographic primitives with a level of rigor that was previously impossible. This could significantly reduce the number of exploits and hacks in the Web3 space.
What's next
We are likely to see an explosion in "formalized research" where new cryptographic papers are accompanied by Lean code from day one. As AI models become better at symbolic reasoning, the role of the human researcher will shift toward setting the constraints and goals, while AI agents handle the tedious work of proof generation. better.codes is just the first step toward a self-securing blockchain ecosystem.
Sources
- Ethereum Foundation Blog (blog.ethereum.org)
- zkSecurity Research Archives
- Lean Prover Community Hub
Educational analysis generated with AI and editorially reviewed.