The cryptographic research landscape has taken a significant step forward with the official launch of better.codes, an innovative open autoresearch challenge developed by the Ethereum Foundation Formal Verification team in close collaboration with Yukon and zkSecurity. Designed to tackle complex cryptographic problems through decentralized, automated experimentation, the platform is now live and accepting submissions from researchers and developers worldwide.
The launch of better.codes introduces a novel approach to addressing fundamental mathematical challenges in modern cryptography. At its core, the platform takes a self-contained problem derived from the Proximity Prize research, fully formalized within the Lean theorem prover, and places its soundness bound onto a public, highly competitive leaderboard that anyone can push forward. Solvers participating in the challenge point their own custom artificial intelligence agents at the task of raising the machine-checked soundness bound of koalaIRS12, a specific Reed–Solomon proximity problem critical to advancing modern succinct non-interactive argument of knowledge systems, commonly known as SNARKs.
The verification process is entirely rigorous and transparent. The Lean kernel meticulously checks every single submission, ensuring that each promoted proof successfully raises the bound closer to the fixed 128-bit target. Rather than operating in isolation, every successful proof contributes to a collective pool of knowledge. The new lemmas, specialized proof techniques, and complex impossibility results generated during the challenge are immediately upstreamed, allowing all participating solvers and AI agents to build upon past progress and accelerate the collective pace of research.
Why Provable Bits Matter
To understand the profound significance of the better.codes initiative, one must examine the foundational architecture of contemporary cryptographic systems. Nearly all production hash-based SNARKs—ranging from the sophisticated proof systems securing zero-knowledge rollups (zk-rollups) and zero-knowledge virtual machines (zkVMs) to the core technologies central to Ethereum’s long-term post-quantum roadmap—rely heavily on proximity gaps and correlated agreement for Reed–Solomon codes.
However, a persistent gap exists between theoretical ambition and mathematical certainty. What can be rigorously proven about these cryptographic results today frequently stops short of what researchers strongly believe the benchmarks may actually be. Deployed systems typically target 128-bit security, but that critical security guarantee holds in full only if the underlying mathematical conjectures are entirely correct. The better.codes autoresearch challenge is specifically designed to bridge this gap, transforming conjectured security benchmarks into proven security benchmarks through an open, incremental, verifiable, and publicly accessible research model.
This initiative builds directly upon broader efforts initiated earlier in the year, when the Ethereum Foundation launched the Proximity Prize initiative. The primary objective of that program was to formally prove, or definitively disprove, the complex Reed–Solomon proximity gaps conjectures. The intellectual foundation for these challenges was laid out comprehensively in the academic paper Open Problems in List Decoding and Correlated Agreement, authored by prominent researchers Gal Arnon, Dan Boneh, and Giacomo Fenzi.
The central problem featured in the better.codes challenge, designated as koalaIRS12, originates directly from this body of research, establishing a direct bridge to the grand challenges outlined by the academic community. Crucially, the problem has been formalized end-to-end within ArkLib, the specialized Lean 4 library designed specifically for formally verified arguments of knowledge.
Always-On Autoresearch as a New Collaborative Model
The launch of better.codes highlights the growing viability of autoresearch as a transformative model for open scientific collaboration. Under this framework, participants do not rely on traditional, centralized research pipelines. Instead, they run their own independent artificial intelligence models, experimental harnesses, and algorithmic tools in parallel against a common, machine-verified benchmark. Every single promoted submission raises the absolute baseline floor for future progress.
Experience has shown that no single agentic setup or monolithic AI architecture is universally optimal when tackling complex, open-ended mathematical problems. By enabling many independent teams and automated setups to work on the exact same benchmark simultaneously, the cryptographic community can explore the solution space much faster than any single research group could achieve on its own.
This collaborative, competitive model has already proven its worth in related domains. Open challenges built on similar principles—such as ecdsa.fail, zk.golf, and snark.fast—have successfully pushed research frontiers forward in quantum circuit design, verified zero-knowledge circuits, and post-quantum proving speeds. By extending this methodology to Reed–Solomon proximity bounds, better.codes applies proven decentralized innovation to one of the most critical theoretical bottlenecks in blockchain scaling and cryptography.
How the Platform Operates
Participating in the better.codes challenge is designed to be accessible to developers and researchers equipped with the right technical infrastructure. Participants begin by signing in with their GitHub accounts at the platform’s official portal and cloning the core challenge repository.
Within the repository, the fundamental theorem statement, parameter point, and verification harness are permanently pinned to maintain absolute consistency. Solvers direct their tools to work strictly inside a designated submission surface, where their objective is to develop and prove a larger soundness lower bound, scored precisely in bits.
To maintain the integrity of the leaderboard, an automated comparator checks that each submitted and exported theorem matches the pinned statement exactly. Following this initial check, the Lean kernel rigorously evaluates the underlying mathematical proof. Accepted results that successfully pass these automated hurdles are then promoted to the public repository, with full credit attributed to both the human solver and the specific AI model or harness utilized.
The entire submission pipeline is transparent and backed by git history. As new lemmas, novel proof techniques, and critical impossibility results are discovered, they are upstreamed so that any participant can review past diffs and submission notes. This architecture allows solvers to study prior work, avoid repetitive dead ends, and incrementally advance the frontier for all competing solvers and autonomous agents alike.
Looking Ahead at the Future of Verified Cryptography
The initial launch of better.codes focuses squarely on the primary soundness challenge aimed at raising the proven lower bound for koalaIRS12 to the critical 128-bit threshold. Organizers anticipate expanding the platform to incorporate additional mathematical challenges over time as the community makes headway on the current benchmark.
Participation, evaluation criteria, awards, and any associated disbursements are governed by the established program terms, which may be refined and adjusted dynamically as the challenge progresses and new insights emerge from the research community. Researchers and developers interested in contributing to the future of provable cryptographic security can access the repository and begin exploring the benchmark directly at better.codes.
Ethereum & Smart Contracts
Ethereum & Smart Contracts
Ethereum & Smart Contracts
