The landscape of zero-knowledge cryptography has reached a significant milestone with the official launch of better.codes, an open autoresearch challenge developed by the Ethereum Foundation Formal Verification team in collaboration with Yukon and zkSecurity. The platform is now live, inviting researchers, developers, and autonomous systems from around the globe to push the boundaries of modern cryptographic proof systems.
Designed to tackle complex theoretical limitations in decentralized infrastructure, the initiative combines formal verification methods with artificial intelligence. By leveraging machine-checked proofs, better.codes aims to bridge a critical gap in the mathematical foundations that secure modern zero-knowledge rollups, virtual machines, and Ethereum’s long-term post-quantum roadmap.
Why Provable Bits Matter for Modern Cryptography
At the heart of nearly all production-grade, hash-based succinct non-interactive argument of knowledge (SNARK) systems lie intricate mathematical concepts known as proximity gaps and correlated agreement for Reed–Solomon codes. These cryptographic frameworks are essential for scaling blockchain networks, ensuring privacy, and securing Layer 2 execution environments. However, a persistent challenge has plagued cryptographers: the security guarantees that can be formally proven today often stop short of the security margins that researchers believe are achievable.
While deployed systems routinely target a robust 128-bit security level, that theoretical guarantee only holds in full if the underlying mathematical conjectures are entirely correct. The better.codes autoresearch challenge has been established specifically to close this gap between conjectured security benchmarks and rigorously proven security benchmarks. By fostering open, incremental, verifiable, and public research, the initiative seeks to replace assumptions with absolute mathematical certainty.
This launch directly connects to broader academic and development initiatives. Earlier this year, the Ethereum Foundation introduced the Proximity Prize initiative to definitively prove or disprove long-standing conjectures regarding Reed–Solomon proximity gaps. The core challenges in this domain were laid out comprehensively in the research paper Open Problems in List Decoding and Correlated Agreement, authored by Gal Arnon, Dan Boneh, and Giacomo Fenzi.
The primary problem featured on better.codes—designated as koalaIRS12—is derived directly from this foundational research. It bridges the gap to the broader grand challenges outlined by the cryptographic community and is formalized end-to-end within ArkLib, the specialized Lean 4 library designed for formally verified arguments of knowledge.
An Always-On Autoresearch Model
The architecture of better.codes embraces a relatively new and powerful paradigm in scientific and technological advancement: the always-on autoresearch challenge. Traditional mathematical and cryptographic research often relies on isolated teams working sequentially or competitively behind closed doors. In contrast, this collaborative model encourages participants to point their own independent AI agents, custom harnesses, and analytical tools in parallel against a single, common, and rigorously verified benchmark.
In an open problem of this complexity, no single agentic setup or automated methodology is universally optimal. By allowing dozens of independent setups to attack the same benchmark simultaneously, the cryptographic community can move the research frontier forward at a pace that no single team could achieve alone.
This collaborative, distributed approach builds upon the success of prior decentralized scientific challenges. Platforms such as ecdsa.fail, zk.golf, and snark.fast have previously demonstrated how open, gamified, and verifiable problem-solving can rapidly accelerate progress across quantum circuit design, verified zero-knowledge circuits, and post-quantum proving speeds. Through better.codes, this philosophy is now being applied to the rigorous domain of formal verification and error-correcting codes.
How the Platform Operates
Participating in the challenge is designed to be accessible to researchers and developers equipped with the right technical tooling. Participants sign in securely using GitHub credentials at the better.codes platform and clone the dedicated challenge repository. Within this environment, the core theorem statement, parameter points, and verification harness are permanently pinned to ensure consistency.
Solvers work within a strictly designated submission surface, focusing their AI agents and human ingenuity on proving a larger soundness lower bound, which is meticulously scored in bits. To maintain the integrity of the leaderboard, an automated comparator checks that every submitted and exported theorem matches the pinned statement word-for-word and parameter-for-parameter. Following this initial check, the Lean kernel rigorously verifies the mathematical proof from the ground up.
Once a submission passes this rigorous verification process, the accepted result is promoted to the public repository, with full credit awarded to both the human solver and the specific artificial intelligence model utilized in generating the proof.
Transparency is a core tenet of the platform. All submissions are fully transparent and git-backed. As solvers make progress, new lemmas, advanced proof techniques, and even valuable impossibility results are automatically upstreamed. This ensures that any researcher or autonomous agent can review past code diffs and detailed submission notes, build directly upon prior work, and avoid documented dead ends. Consequently, every incremental breakthrough raises the floor for the entire ecosystem, accelerating progress for all competing solvers and agents simultaneously.
Looking Ahead at the Cryptographic Frontier
The initial launch of better.codes focuses squarely on the primary soundness challenge: raising the proven lower bound for the koalaIRS12 Reed–Solomon proximity problem toward the vital 128-bit target. As the initiative matures, the organizers hope to introduce additional cryptographic challenges to the platform, further expanding the scope of formally verified security proofs.
Eligibility, evaluation criteria, awards, and any associated disbursements are governed by the established program terms and may be adjusted dynamically as the challenge progresses and new insights emerge from the research community. For researchers, developers, and autonomous agent builders eager to contribute to the mathematical foundations of zero-knowledge cryptography, the challenge is now open at better.codes.

Comments (0)