What happened
The Ethereum Foundation's Formal Verification team, in collaboration with Yukon and zkSecurity, has officially launched better.codes, an open autoresearch challenge platform. The core objective of this project is to enhance the mathematical security of hash-based SNARK (Succinct Non-interactive Arguments of Knowledge) systems. The project formalizes a complex research problem from the "Proximity Prize" in the Lean programming language and places its soundness bounds on a public leaderboard, inviting researchers and AI agents to push the boundaries of cryptographic safety.
Technology context
SNARK systems are vital for Ethereum's scalability and privacy, allowing for the verification of transactions without revealing sensitive underlying data. However, their security relies on "soundness bounds"—the mathematical probability that a malicious actor could generate a fake proof. Traditionally, these bounds are proven manually on paper, which is prone to human error. Formal verification uses specialized software (like Lean) to create machine-checked proofs, ensuring that the mathematical logic behind the protocols is 10 to 100% accurate and immune to reasoning flaws.
Why it matters
This initiative marks a shift from isolated theoretical research to "agentic collaboration," where both humans and Artificial Intelligence models can contribute to securing Web3 infrastructure. By establishing a public leaderboard based on machine-checked proofs, the Ethereum Foundation is setting a gold standard for code trust. If SNARK soundness bounds are optimized and formally verified, Layer 2 protocols and privacy systems become significantly more robust, reducing the risk of catastrophic exploits that could lead to the loss of user funds.
Key terms explained
- SNARK: A type of cryptographic proof that allows one party to prove they possess certain information without revealing it, in a way that is very fast to verify.
- Lean: A functional programming language and theorem prover used to write code whose mathematical correctness can be automatically verified.
- Soundness Bound: A numerical value indicating the maximum probability that an attacker could trick the system with an invalid proof.
- Formal Verification: The act of proving or disproving the correctness of algorithms underlying a system with respect to a certain formal specification, using mathematical methods.
Impact
In the short term, better.codes will accelerate the discovery of new optimization methods for proximity protocols (such as FRI, used in STARKs). In the medium term, the success of this "open autoresearch" model could lead to the integration of AI agents into the smart contract auditing process, transforming security from a reactive process (bug hunting) into a proactive, mathematically guaranteed one. This will likely increase institutional investor confidence in the Ethereum ecosystem.
What's next
We can expect more blockchain protocols to adopt formal verification as a mandatory standard before deployment. Furthermore, collaboration between human researchers and AI agents (Large Language Models specialized in mathematics) will become the norm for solving complex cryptographic problems. The "safety leaderboard" inaugurated by better.codes could serve as a blueprint for other critical fields, such as aviation software or autonomous medical systems.
Educational analysis generated by AI and editorially reviewed.