The collaborative project is being operated as part of a provably.fast workshop, with its core infrastructure and codebase hosted under the starknet-innovation GitHub organization. By opening the challenge to the global research community and advanced AI systems alike, the organizers are leveraging modern automated theorem proving alongside traditional human intellect to push the boundaries of cryptographic efficiency.
At the heart of the challenge is the optimization of STARKs, which are a prominent form of zero-knowledge proofs heavily utilized in modern cryptography and scaling solutions. Unlike other cryptographic proof systems that may rely on complex assumptions, STARKs rely exclusively on a hash function, making them particularly attractive for long-term security and transparency. However, optimizing their performance remains a critical area of ongoing computer science research.
The newly introduced challenge specifically targets the number of what cryptographers refer to as "bad challenges" within a Reed-Solomon proximity test. This fundamental mathematical test determines the precise number of queries a verifier needs to run in order to be confident in the integrity of a proof. By lowering the required query count, systems can achieve substantial efficiency gains. Currently, the baseline bounds utilized in the field stem from research conducted by Dao, Kominers, and Thaler in 2026. The new open mathematics challenge aims to improve upon these established thresholds through tighter mathematical bounds.
Three targets, verified in Lean
To ensure absolute mathematical rigor, the initiative has outlined three specific targets for participants to solve. Among these objectives are replacing a currently known quadratic bound with a more efficient linear one, as well as proving a quadratic bound across a substantially wider range of parameters than what is currently possible.
Rather than relying on standard academic paper submissions or traditional peer review processes, every single target in the challenge is strictly pinned in Lean, an interactive theorem prover developed for formalizing mathematics. Under this framework, a proposed solution counts as officially solved only when it comes with an independently reviewed, machine-checked proof. Furthermore, full credit for the breakthrough is credited directly to the individual human researcher or the specific AI agent responsible for generating the correct proof.
The organizers of the initiative have emphasized an important nuance regarding the relationship between theoretical mathematics and practical implementation. As noted on the project’s website, finding a tighter mathematical bound in isolation does not automatically shrink a proof on its own. Realizing tangible cryptographic savings in practice also requires a comprehensive, end-to-end protocol analysis to ensure that theoretical improvements translate smoothly into operational efficiency gains within actual zero-knowledge systems.
As the challenge gets underway, the intersection of formal verification tools like Lean, cutting-edge cryptographic protocols like STARKs, and the rapid advancement of artificial intelligence agents marks a fascinating frontier in collaborative scientific discovery. By demanding absolute formal correctness through machine-verified proofs, the workshop aims to establish a new standard of reliability for cryptographic optimization research.

Comments (0)