The Ethereum Foundation’s Formal Verification team, in a strategic partnership with Yukon and zkSecurity, has officially launched better.codes, an innovative open-source autoresearch platform designed to tackle one of the most pressing challenges in modern cryptography: the formal verification of Reed-Solomon proximity gaps. By incentivizing the use of AI agents to solve complex mathematical problems within the Lean theorem prover, the initiative seeks to bridge the gap between theoretical cryptographic security and the rigorous, machine-checked proofs required for high-stakes decentralized infrastructure.
The Cryptographic Security Gap
At the heart of modern blockchain scalability solutions—including zero-knowledge rollups (zkRollups) and zero-knowledge virtual machines (zkVMs)—lies the reliance on succinct non-interactive proof systems, or SNARKs. These systems are foundational to the security of the Ethereum ecosystem and its ambitious post-quantum roadmap. However, these systems often rely on mathematical conjectures regarding Reed-Solomon codes, specifically concerning proximity gaps and correlated agreement.
While current production-grade SNARKs are designed to provide 128-bit security—the industry standard for robust cryptographic protection—that guarantee is only as strong as the underlying mathematical assumptions. If the conjectures governing these proximity gaps are found to be weaker than anticipated, the theoretical foundation of these systems could be compromised. Currently, there exists a significant disconnect between what is conjectured to be secure and what has been formally proven using machine-checked verification. The better.codes initiative was conceived to close this divide by fostering a collaborative, public, and incremental research environment.
Chronology of the Proximity Prize Initiative
The launch of better.codes is the latest milestone in a broader, multi-year effort to fortify the mathematical foundations of zero-knowledge proofs. The timeline of this research trajectory reflects a growing emphasis on formal verification:
- Early 2024: The Ethereum Foundation formally introduced the Proximity Prize initiative, a project dedicated to the rigorous proof or disproof of the Reed-Solomon proximity gap conjectures. This initiative was catalyzed by the publication of "Open Problems in List Decoding and Correlated Agreement," a seminal paper by researchers Gal Arnon, Dan Boneh, and Giacomo Fenzi.
- Mid-2024: Development teams at Yukon and zkSecurity began integrating these research goals into a framework compatible with Lean 4, the state-of-the-art interactive theorem prover. This period saw the creation of ArkLib, a dedicated library for formally verified arguments of knowledge.
- Late 2024: The better.codes platform moved into a testing phase, establishing the infrastructure for an automated leaderboard where AI agents could interact with formal proofs.
- Current Day: The platform is now live, providing researchers and AI agents with the challenge koalaIRS12—a specific, formalized problem set designed to incrementally push the machine-checked soundness bound toward the 128-bit target.
Technical Mechanics: How the Challenge Functions
The operational model of better.codes departs from traditional academic research by treating the proof process as a distributed, agent-based compute task. The platform utilizes a transparent, Git-backed workflow. Participants sign in via GitHub, clone the designated repository, and utilize the ArkLib library to work within the specific constraints of the koalaIRS12 problem.
The "solver" (which can be a human researcher or an AI agent) must produce a theorem that satisfies the pinned soundness requirements. Once a submission is made, it undergoes a rigorous verification process. The Lean kernel acts as the ultimate arbiter, checking the submission to ensure the logic holds without error. If successful, the new lemma or proof technique is "promoted" to the public leaderboard.
This model creates a cumulative advantage. By upstreaming every verified proof, the platform ensures that no effort is wasted. Future participants can analyze previous submissions, identify dead ends, and build upon existing lemmas, thereby creating an exponential increase in the rate of progress. This "always-on" research model allows for the parallelization of effort, where multiple independent AI setups work simultaneously on the same problem, significantly outperforming the speed of a single, siloed team.
The Role of AI in Formal Verification
The integration of AI into formal verification represents a paradigm shift in how mathematical breakthroughs are achieved. In traditional mathematics, the "proof space" is vast, and searching for the correct sequence of logic to satisfy a theorem prover can be computationally expensive and intellectually taxing. By deploying AI agents, researchers can explore thousands of potential proof paths in parallel.
The better.codes challenge is not merely about finding a single solution; it is about refining the methodology of agentic research. As participants compete, the competition itself generates metadata on which AI architectures and prompting strategies are most effective at traversing the logic of Lean. This information is invaluable for the broader field of automated reasoning, potentially informing future developments in AI-driven science.
Broader Implications for the Blockchain Ecosystem
The implications of this project extend far beyond theoretical mathematics. If the community succeeds in raising the proven soundness bound of koalaIRS12 to 128 bits, the impact on blockchain security will be profound.
- Risk Mitigation: By moving from conjectures to formal proofs, the risk of catastrophic failure in zk-based infrastructure is significantly reduced. This is critical as more financial value is locked within zkRollups and related protocols.
- Post-Quantum Preparedness: As Ethereum transitions toward a post-quantum roadmap, the reliance on hash-based SNARKs will only increase. Ensuring these are mathematically unassailable is a prerequisite for long-term network survival.
- Standardization: The success of better.codes could set a new standard for how complex cryptographic problems are addressed. The shift toward "open autoresearch" could become the preferred model for other complex, open-source infrastructure challenges.
Official Stance and Community Reaction
While the initiative is spearheaded by the Ethereum Foundation, it is designed to be inclusive. The project’s architecture emphasizes that no single entity holds the monopoly on solving these problems. The collaborative nature of the effort—involving contributors from academia, private security firms, and independent developers—is intended to mirror the decentralized ethos of the technology it aims to protect.
According to preliminary briefings from the involved parties, the goal is not to reach an immediate, singular solution, but to cultivate a sustainable ecosystem of "ever-improving" proofs. By incentivizing the community, the Ethereum Foundation is effectively outsourcing the labor of formal verification to the most efficient agents available—whether human or silicon.
Looking Ahead
The launch of the better.codes platform is a significant, if quiet, victory for the formal verification movement. By turning the daunting task of mathematical verification into an accessible, competitive challenge, the organizers have lowered the barrier to entry for highly specialized cryptographic research.
The current challenge, koalaIRS12, is expected to be the first of many. As the platform matures, the organizers intend to expand the scope of the challenges, potentially including other critical components of the ZK-proof stack. For now, the focus remains on the 128-bit target—a milestone that, if achieved, will provide a vital safeguard for the future of decentralized digital assets.
Researchers, developers, and AI engineers are encouraged to visit the platform to review the current state of the leaderboard. As the community begins to contribute, the transparency of the Git-backed history will serve as a living document of progress, documenting every step of the journey toward a more provably secure cryptographic future. The success of this endeavor will ultimately be measured by the number of bits added to the soundness of the system—a metric that, while abstract, holds the key to the security of the next generation of global internet infrastructure.
