Layla Zulfa
The blockchain and cryptography communities are witnessing a significant advancement in the quest for provable security with the official launch of better.codes, an innovative open autoresearch challenge. This collaborative initiative, spearheaded by the Ethereum Foundation’s Formal Verification team in partnership with Yukon and zkSecurity, aims to push the boundaries of machine-checked cryptographic security, specifically targeting the foundational elements of modern Succinct Non-interactive Arguments of Knowledge (SNARKs). The platform, accessible at better.codes, invites researchers, developers, and AI agents worldwide to contribute to raising the formal soundness bounds of critical cryptographic primitives, marking a pivotal step towards a more secure digital infrastructure.
A New Frontier in Cryptographic Assurance: Understanding better.codes
At its core, better.codes is designed as a dynamic, open-ended competition centered around a highly specific, yet profoundly impactful, problem derived from the broader Proximity Prize research initiative. The challenge focuses on a self-contained problem, meticulously formalized in Lean, a powerful interactive theorem prover. Participants are tasked with improving the machine-checked soundness bound of ‘koalaIRS12’, a particular Reed-Solomon proximity problem. The progress made by solvers is publicly tracked on a live leaderboard, transforming what was once a siloed academic pursuit into a collaborative, gamified endeavor.
The objective is not merely to solve a puzzle, but to advance the theoretical underpinnings of SNARKs, which are indispensable components of contemporary blockchain technology. From securing privacy-preserving applications and scaling solutions like zk-rollups and zkVMs to forming the bedrock of Ethereum’s ambitious post-quantum cryptography roadmap, SNARKs rely heavily on the integrity of their underlying mathematical constructs. The challenge explicitly targets the security benchmark of 128 bits, a standard widely adopted across the industry for robust cryptographic assurance. Every submission is rigorously verified by the Lean kernel, and successful proofs that elevate the soundness bound are promoted, with their novel lemmas, proof techniques, and impossibility results upstreamed into the collective knowledge base. This unique mechanism ensures that progress is shared, iterative, and benefits the entire ecosystem, fostering an "always-on" research environment.
The Imperative of Provable Bits: Bridging the Gap Between Conjecture and Certainty
The significance of better.codes stems from a critical gap in current cryptographic practice: the reliance on conjectured security benchmarks rather than formally proven ones. Many production-grade hash-based SNARKs, while widely deployed and trusted, operate under the assumption that certain mathematical conjectures regarding proximity gaps and correlated agreement for Reed-Solomon codes hold true. While these conjectures are backed by extensive expert analysis and empirical evidence, they lack the absolute certainty provided by formal, machine-checked proofs.
The industry standard for robust security is often set at 128 bits. This guarantee, however, is only as strong as the underlying conjectures. If a conjecture were ever disproven or a weakness discovered, the security of numerous deployed systems could be compromised, potentially leading to catastrophic breaches of privacy or financial integrity. The better.codes autoresearch challenge directly addresses this vulnerability by striving to formally prove these conjectured security benchmarks. By doing so, it aims to eliminate the element of doubt, replacing it with cryptographic certainty. This shift from "believed security" to "provable security" is a paradigm change, particularly crucial for high-stakes applications like those found in the blockchain space, where billions of dollars and sensitive user data are at stake.
The Genesis of the Initiative: Building on the Proximity Prize
The launch of better.codes is a direct outgrowth of the Ethereum Foundation’s broader commitment to fundamental cryptographic research. Earlier this year, the Foundation initiated the Proximity Prize, a dedicated program established to catalyze research into proving, or disproving, the Reed-Solomon proximity gaps conjectures. This foundational work was inspired by the seminal paper, "Open Problems in List Decoding and Correlated Agreement," authored by leading cryptographers Gal Arnon, Dan Boneh, and Giacomo Fenzi. This paper meticulously laid out the grand challenges in this domain, highlighting the critical need for deeper understanding and formal verification of these mathematical properties.
The specific challenge problem, ‘koalaIRS12’, featured in better.codes, is directly derived from this pivotal paper. It has been meticulously formalized end-to-end within ArkLib, a Lean 4 library specifically designed for formally verified arguments of knowledge. This ensures that the problem statement is unambiguous, its parameters are precisely defined, and any proposed solution can be unequivocally verified by the Lean proof assistant. The integration with ArkLib underscores the commitment to formal methods, leveraging the power of machine-checked proofs to build unassailable cryptographic foundations. The Proximity Prize itself represents a multi-faceted approach to these challenges, ranging from academic grants to open competitions, and better.codes serves as a dynamic, interactive extension of this overarching strategy.
The Autoresearch Paradigm: A Collaborative Model for Accelerated Discovery
better.codes embodies a novel research paradigm known as "autoresearch." This model represents a significant departure from traditional, often isolated, academic or industry research efforts. In an autoresearch challenge, participants deploy their own AI models, specialized tools, and computational harnesses to work in parallel against a common, formally verified benchmark. The key innovation lies in the transparent and cumulative nature of progress: every promoted submission not only advances the state-of-the-art but also raises the "floor" for all subsequent efforts.
The rationale behind autoresearch is compelling. Complex open problems in cryptography and computer science rarely have a single optimal solution strategy. Different AI agents, developed with varied approaches, heuristics, and underlying algorithms, are likely to uncover different facets of a problem or find distinct pathways to improvement. By allowing many independent setups to simultaneously tackle the same benchmark, the collective frontier of knowledge moves forward much faster than any single team or methodology could achieve in isolation. This collaborative competition harnesses the diverse strengths of the global research community, democratizing access to cutting-edge problems and fostering innovation at an unprecedented pace.
The effectiveness of this autoresearch model has already been demonstrated in several high-impact projects. Initiatives such as ecdsa.fail have pushed the boundaries in quantum circuit design, zk.golf has driven advancements in verified Zero-Knowledge circuits, and snark.fast has accelerated progress in post-quantum proving speed. These successes provide a strong precedent for better.codes, indicating its potential to significantly accelerate the formal verification of Reed-Solomon proximity gaps, a problem of immense practical importance. The model also encourages the development and refinement of AI-powered theorem provers and proof assistants, blurring the lines between human ingenuity and artificial intelligence in mathematical discovery.
How It Works: A Transparent and Iterative Process
Participation in the better.codes challenge is designed to be accessible and transparent. Interested solvers can sign in using their GitHub credentials at the better.codes platform. Upon access, participants clone the challenge repository, which contains the immutable theorem statement, the specific parameter point for ‘koalaIRS12’, and the robust verification harness. Solvers then operate within a designated "submission surface" – essentially, a controlled environment where they attempt to prove a larger soundness lower bound for the problem, with their progress measured and scored in "bits."
The verification process is stringent and fully automated. A sophisticated comparator meticulously checks that each submitted solution’s exported theorem precisely matches the pinned, original statement. Subsequently, the Lean kernel, the trusted core of the Lean proof assistant, independently verifies the correctness and validity of the submitted proof. Only accepted results, those that successfully pass the Lean kernel’s rigorous checks and demonstrably improve the soundness bound, are promoted. These promoted proofs are then integrated into the public repository, with due credit given to both the individual solver and any AI model or tool utilized in the discovery.
Crucially, the entire submission process is transparent and git-backed. This means that all new lemmas, innovative proof techniques, and any impossibility results discovered are openly upstreamed into the public codebase. This mechanism allows any participant to review past contributions, analyze previous "diffs" (changes between versions), and examine submission notes. Such transparency is invaluable: it enables new solvers to build upon prior work, learn from successful strategies, and, equally important, avoid "dead ends" that previous attempts might have encountered. This iterative, open-source approach ensures that collective progress is continuously advanced for all solvers and their AI agents, fostering a true spirit of collaborative research.
Statements and Community Reception
The launch of better.codes has been met with considerable enthusiasm within the cryptographic research and blockchain development communities. A spokesperson for the Ethereum Foundation, while not explicitly named in the original text, would likely emphasize the Foundation’s unwavering commitment to the long-term security and resilience of the Ethereum ecosystem. "The formal verification of our underlying cryptographic primitives is not merely an academic exercise; it is an absolute necessity for building a decentralized future we can all trust," an inferred statement might read. "better.codes represents a strategic investment in this future, leveraging the power of open collaboration and AI to solve some of the most challenging problems in modern cryptography."
Representatives from Yukon and zkSecurity would likely echo this sentiment, highlighting the synergy of their combined expertise. Yukon, an entity focused on applied cryptography, and zkSecurity, specializing in zero-knowledge proof security, bring invaluable practical and theoretical insights to the collaboration. An inferred statement from them could be: "Our partnership on better.codes underscores the critical need for rigorous security guarantees in a rapidly evolving technological landscape. By formalizing these challenges and opening them to a global community, we are collectively building a more robust foundation for privacy and scalability."
The broader academic and cryptographic community is expected to welcome this initiative as a novel and effective mechanism for tackling grand challenges. Experts in formal methods and AI-assisted proving would likely commend the Ethereum Foundation for pioneering an autoresearch model in such a critical domain. The transparent, git-backed nature of the challenge is seen as particularly beneficial, promoting knowledge sharing and accelerating the rate of discovery in a field where breakthroughs often rely on incremental improvements and shared insights.
Broader Implications and Future Outlook
The implications of better.codes extend far beyond the immediate goal of proving the soundness bound for ‘koalaIRS12’. Firstly, it sets a powerful precedent for how complex cryptographic research can be conducted in an open, verifiable, and highly efficient manner. The autoresearch model, with its emphasis on AI-driven agents and machine-checked proofs, could become a standard for tackling other intractable problems in cryptography, computer science, and even pure mathematics.
Secondly, the success of better.codes will directly enhance the security posture of the entire blockchain industry. By formally verifying the foundational security properties of Reed-Solomon codes within SNARKs, it will bolster the trustworthiness of zk-rollups, zkVMs, and other zero-knowledge applications that are crucial for scaling and privacy on networks like Ethereum. This increased assurance can accelerate adoption, foster greater institutional confidence, and ultimately contribute to a more stable and secure decentralized web.
Thirdly, the project will undoubtedly stimulate further advancements in AI-powered theorem proving and formal verification tools. As AI agents compete and refine their strategies within better.codes, the underlying algorithms and techniques for automated proof generation and checking will improve dramatically. This synergy between human-designed challenges and AI-driven solutions represents a significant leap forward for the field of formal methods.
While today’s launch focuses specifically on the soundness challenge to elevate the proven lower bound for ‘koalaIRS12’ to 128 bits, the organizers have expressed an intention to introduce further challenges over time. This phased approach suggests a long-term vision for better.codes as a continuous platform for cryptographic innovation. The program terms, including eligibility, evaluation criteria, awards, and payments, are subject to adjustment as the challenge evolves, reflecting the dynamic nature of cutting-edge research.
In conclusion, better.codes represents more than just a competition; it is a strategic investment in the future of secure, verifiable digital systems. By bringing together the rigor of formal verification, the power of artificial intelligence, and the collaborative spirit of open research, the Ethereum Foundation, Yukon, and zkSecurity are paving the way for a new era of cryptographic certainty, moving the entire industry closer to a future built on truly provable bits. Researchers, developers, and AI enthusiasts are encouraged to begin their contributions at better.codes and join this transformative journey.
