Submission and publication
A contributor uses the common submission form to provide the problem ID, format, title, authors, contact details, a paper or video link, and a plain-text overview. The form sends a private submission record to the organisers. An organiser checks that the entry concerns a problem listed in the AIM collection of open applied problems on GitHub. Entries remain in scope as their problems progress to partially resolved, claimed, solved or Lean verified. Proposals for new problems or corrections follow the source repository procedure. The mathematical statement, status, attribution and linked material are reviewed separately. Questions go back to the author before the public version is agreed.
Each accepted library page credits the explanation’s authors and presents their account. It cites the AIM problem record and exact source revision, records the problem’s status when reviewed, describes the intended audience and review scope, and links to the paper or video. Authors approve the public text and attribution before publication.
Review criteria
Understanding: does the explanation help its intended reader grasp the problem, an application, the evidence or any proof it discusses? Faithfulness: does it preserve the original claim, assumptions, scope and essential steps? Context: do historical attribution or worked examples add useful understanding where included? Care: are limitations, sources and the nature of any verification accurately described? The sophistication of the mathematics is not itself a measure of quality.
A paper may focus entirely on the mathematics behind an application. Reviewers assess the explanation of the application, the role of the mathematics, and the treatment of assumptions and limitations. A new proof, a new theorem or a solved conjecture is not required.
AI assistance and author responsibility
Contributions with or without AI assistance are equally welcome and are reviewed on the same explanatory merits. Participation does not require using or endorsing AI. Developments in AI raise serious questions about learning, attribution, careers and the culture of mathematics. Respectful disagreement about these questions is welcome.
Authors should check their statements, references and examples, explain what they have verified, and identify uncertainty or gaps. This does not require proving every result they cite: distinguish established results and their sources from new claims or unverified steps. If AI assisted the work, briefly describe what it helped with and how its output was checked. Private chat histories are not required. The authors remain responsible for the submitted account, and reviewers may request clarification or revision.
Authorship and attribution
Clearly credit both the authors of the explanation and the people behind its source mathematics. As relevant, acknowledge problem formulation, earlier partial results, proofs, expository sources, examples, checking and formalisation, as well as reused code or illustrations. Cite the sources you rely on and make clear what your explanation adds. A new account should help readers trace the work on which it builds.
Multiple explanations of a problem
Several papers or videos may address the same problem or conjecture. An existing explanation does not reserve the topic or prevent another account from being accepted. Different perspectives, examples and intended audiences can each add understanding. Each submission is reviewed on its own explanatory merits; being first to explain or prove a conjecture gives no priority in the library or competition.
Private review material
The form sends submissions to a private review repository, where invited organisers and competition judges discuss entries and record their review. Entrants do not need a GitHub account or access to that repository.
Contact details, unpublished submissions and review correspondence remain within the review process. The organisers agree the public version with the authors before publication.
Checking code and formal proofs
Reviewers assess the supplied text and verification evidence. Any reproduction of submitted code or Lean projects takes place in an isolated environment. Published review notes describe the checks actually performed and their scope.
Corrections and concerns
Contact the website maintainers about mathematical corrections or concerns about attribution, privacy, accessibility, review or the treatment of contributors. Where relevant, include the page or submission and a description of the concern. Respectful disagreement about AI and its role in mathematics is welcome. A review label records the work actually done; it is not a promise that errors are impossible.