Layla Zulfa
better.codes, an innovative open autoresearch challenge developed by the Ethereum Foundation Formal Verification team in collaboration with Yukon and zkSecurity, has officially gone live, marking a significant stride in the pursuit of provably secure cryptographic systems. This initiative directly addresses a critical gap in the formal verification of security benchmarks for Succinct Non-interactive Arguments of Knowledge (SNARKs), aiming to elevate the trustworthiness of foundational technologies underpinning the decentralized web.
At its core, better.codes presents a meticulously crafted, self-contained problem derived from the broader Proximity Prize research. This problem, formalized within the Lean proof assistant, focuses on determining the soundness bound of koalaIRS12, a specific Reed-Solomon proximity problem. The challenge transforms this complex cryptographic research into a public, competitive, yet collaborative endeavor, where participants—leveraging their own AI agents and sophisticated tools—vie to incrementally push forward the machine-checked soundness bound on a live leaderboard. Each successful submission, verified rigorously by the Lean kernel, not only advances the bound towards a fixed 128-bit target but also contributes new lemmas, proof techniques, and impossibility results back to the collective knowledge base, accelerating progress for the entire community of solvers.
The Imperative for Provable Security in Modern Cryptography
The launch of better.codes is not merely an academic exercise; it responds to a pressing need within the rapidly evolving landscape of blockchain and cryptographic technology. SNARKs are fundamental to the architecture of numerous production-grade systems, including zero-knowledge rollups (zkRollups), zero-knowledge Virtual Machines (zkVMs), and are integral to Ethereum’s long-term post-quantum roadmap. The security of these systems heavily relies on the theoretical underpinnings of proximity gaps and correlated agreement for Reed-Solomon codes.
While these deployed systems are designed to achieve a robust 128-bit security level—a benchmark widely accepted in classical cryptography as resistant to brute-force attacks from even the most powerful supercomputers—the formal, mathematical proof for this exact security guarantee often hinges on unproven conjectures. Researchers in the field acknowledge that current provable results fall short of what is intuitively believed to be the true security benchmarks. This discrepancy creates a "security gap": the assurance of 128-bit security holds fully only if these underlying conjectures are indeed true. better.codes is specifically engineered to bridge this gap through a mechanism of open, incremental, verifiable, and public research, transforming conjectures into formally proven facts.
Earlier this year, the Ethereum Foundation demonstrated its commitment to this foundational research by launching the Proximity Prize initiative. This broader program was established with the ambitious goal of either proving or disproving the critical Reed-Solomon proximity gaps conjectures. The intellectual framework for these grand challenges was comprehensively laid out in the influential paper "Open Problems in List Decoding and Correlated Agreement" by prominent cryptographers Gal Arnon, Dan Boneh, and Giacomo Fenzi. The koalaIRS12 challenge problem, chosen for better.codes, is a direct offshoot of this seminal paper, designed to bridge directly to these grand challenges and is fully formalized end-to-end within ArkLib, the Lean 4 library specifically developed for formally verified arguments of knowledge. The seamless integration with ArkLib ensures that all mathematical statements and proofs are expressed in a language that can be machine-checked for correctness, eliminating ambiguity and potential human error.
Understanding Reed-Solomon Codes and Their Role in SNARKs
To appreciate the significance of koalaIRS12, it is essential to understand Reed-Solomon codes. These are a class of error-correcting codes widely used in digital communications and storage (e.g., CDs, DVDs, QR codes) to detect and correct burst errors. In the context of SNARKs, Reed-Solomon codes are employed to encode data in a way that allows a prover to demonstrate knowledge of a secret computation without revealing the computation itself, and a verifier to check this proof with minimal effort. The concept of "proximity gaps" relates to how "far" a given codeword is from a valid Reed-Solomon codeword. The larger the proximity gap, the more robust the security guarantee against malicious provers attempting to forge proofs. Proving the exact bounds of these gaps is paramount for quantifying the cryptographic strength of SNARKs. A 128-bit security level implies that breaking the system would require an astronomical 2^128 operations, a feat currently considered computationally infeasible.
The Power of Always-On Autoresearch
better.codes introduces a novel paradigm for open collaboration known as an "autoresearch challenge." This model diverges from traditional academic research by inviting participants to deploy and run their own AI models, specialized software harnesses, and computational tools in parallel against a common, rigorously verified benchmark. The distinguishing feature is that every promoted submission, validated by the system, automatically raises the "floor" for future progress. This means that successful contributions are not just recorded but are integrated, becoming the new baseline for subsequent research and development.
This approach is predicated on the understanding that no single agentic setup or research team holds a monopoly on optimal solutions for complex, open-ended problems in cryptography and formal verification. By fostering a diverse ecosystem where numerous independent setups tackle the same benchmark, the collective frontier of knowledge can be advanced significantly faster than any single team could achieve in isolation. This distributed, competitive-yet-collaborative methodology harnesses the collective intelligence and computational power of the global research community.
The effectiveness of this autoresearch model has already been demonstrated in several high-profile initiatives. Challenges such as ecdsa.fail have pushed the research frontiers in quantum circuit design, zk.golf has driven advancements in verified Zero-Knowledge circuits, and snark.fast has significantly improved post-quantum proving speeds. These precedents underscore the potential for better.codes to similarly catalyze breakthroughs in Reed-Solomon proximity gap research, directly impacting the security guarantees of SNARKs. The Ethereum Foundation’s commitment to this model reflects a broader vision for decentralized, open-source scientific discovery, aligning perfectly with the ethos of the blockchain ecosystem itself.
How the Challenge Works: A Transparent and Collaborative Process
Participation in the better.codes challenge is designed to be accessible and transparent, leveraging familiar developer tools and platforms. Prospective solvers begin by signing in with their GitHub account at better.codes, gaining access to the challenge repository. This repository contains the immutable theorem statement, the precise parameter point for the koalaIRS12 problem, and the robust verification harness. Solvers operate within a designated submission surface, focusing their efforts on proving a larger soundness lower bound, with their progress scored and tracked in "bits."
The integrity and correctness of each submission are ensured through a two-tiered verification process. First, a comparator mechanism meticulously checks that the exported theorem from a submission precisely matches the pinned, official statement. Following this, the Lean kernel, the formal proof assistant, independently verifies the mathematical proof provided. Only submissions that pass both checks are accepted and subsequently promoted to the public repository. Crucially, each successful submission is credited to the individual solver and, if applicable, the specific AI model or tools used, fostering recognition and accountability within the community.
A cornerstone of the better.codes challenge is its commitment to transparency and collaborative learning. All submissions are git-backed, meaning every change and every successful proof is recorded and publicly auditable. This infrastructure allows for new lemmas, innovative proof techniques, and even impossibility results to be "upstreamed." This mechanism ensures that anyone participating can readily access past contributions, review previous diffs (changes), and examine submission notes. This open-source approach enables researchers to build upon prior work, learn from successful strategies, and, equally important, avoid "dead ends" that have already been explored and found unfruitful. This continuous feedback loop and shared knowledge incrementally advance progress for all solvers and AI agents, embodying a true spirit of collective scientific endeavor.
Broader Implications and Future Outlook
The launch of better.codes represents more than just a specific cryptographic challenge; it signifies a strategic investment by the Ethereum Foundation and its collaborators in the long-term security and robustness of the decentralized web. By focusing on formally proving the security guarantees of Reed-Solomon proximity gaps, the initiative directly strengthens the cryptographic foundations upon which zkRollups, zkVMs, and future privacy-preserving technologies are built. This enhanced certainty in cryptographic primitives translates into greater confidence for developers, users, and enterprises adopting blockchain solutions.
Furthermore, better.codes contributes significantly to the advancement of formal verification as a discipline. By pushing the boundaries of automated theorem proving and integrating AI into the proof-generation process, the challenge helps mature tools like Lean and expand their applicability to complex real-world cryptographic problems. This could pave the way for formal verification to become an even more indispensable part of the software development lifecycle, particularly in high-stakes environments.
The "autoresearch" model itself carries profound implications for the future of scientific discovery. Its success in cryptography could inspire similar initiatives in other domains of mathematics, computer science, and even natural sciences, where complex problems require distributed intelligence and verifiable, incremental progress. This model fosters a truly global and meritocratic research environment, aligning with the decentralized principles that underpin blockchain technology.
While today’s launch focuses specifically on the soundness challenge to raise the proven lower bound for koalaIRS12 to 128 bits, the Ethereum Foundation has indicated a forward-looking vision. There is an explicit hope and intent to add further challenges over time, expanding the scope of better.codes to tackle other critical open problems in cryptography and formal verification. The eligibility criteria, evaluation metrics, and potential awards and payments associated with the challenge are governed by the program’s terms and may be adjusted dynamically as the initiative progresses, reflecting its adaptive and evolving nature.
The better.codes challenge is now live, inviting cryptographers, AI researchers, and formal verification experts from around the globe to contribute to a shared mission of cryptographic excellence. For those ready to contribute to the foundational security of the decentralized future, the journey begins at better.codes.
