Plectis

Credit ledger

What people told me, what changed because of it, and where to see the change. All human acknowledgements remain anonymous for now.

Thanks here records a particular contribution. Each linked record explains what was used and what was checked.

Advice from people

Showing an inspectable comparison

A researcher advised showing a clear advantage over existing systems when asking people to use the work.

What changed: The communication rules now require an inspectable comparison and evidence for a claimed advantage, while keeping an invitation to look distinct from a completed review.

Name withheld for now. Exact public files and evidence.

Clarifying the role of persistent research records

A researcher challenged the framing of the system as an external improvement to rapidly advancing models.

What changed: The explanation now distinguishes persistent records for inspection and continuation from a model’s problem-solving capability; the distinction is also published in the credit policy.

Name withheld for now. Exact public files and evidence.

Writing for a first-time reader

A mathematician said the #243 note used names that meant nothing on a first or second reading, defined notation it used once, and never said how restrictive its main hypothesis was.

What changed: All sixteen problem papers were rewritten for a first-time reader, naming standard tools and saying how restrictive each hypothesis is, and the public writing rules now require it.

Name withheld for now. Exact public files and evidence.

Showing where methods and ideas come from

A mathematician said the trouble with AI-assisted mathematics is that it rarely shows where its methods and ideas come from.

What changed: A literature review was run for each of the eight problems, and each paper now credits its sources at the point where they are used.

Name withheld for now. Exact public files and evidence.

Leading the #249 paper with its exact theorem

A mathematician advised leading the #249 paper with its exact finite-level theorem, explaining the mechanism behind it, and comparing it precisely with the existing literature on regular sequences.

What changed: The #249 paper now opens with the exact rank and basis, compares the result with work of Allouche and Shallit, Coons and Martin, and links a minimal Lean entry; the irrationality question stays open.

Name withheld for now. Exact public files and evidence.

Listing each checked result, with a cheap way to check it

A mathematician advised listing each selected result with its exact statement, proof source, novelty status, axioms and limits, and giving reviewers a cheap way to check the statements themselves.

What changed: formalization.yaml now lists the selected results this way, and Comparator checks them against separately written statements, including an altered statement that must be rejected.

Name withheld for now. Exact public files and evidence.

Removing an unsupported specialist connection

A mathematician saw no connection between their specialist methods and the question presented to them.

What changed: That connection was removed from future correspondence unless new evidence supports it; the published credit policy records the same rule for correcting scope.

Name withheld for now. Exact public files and evidence.

Earlier work on the #1049 Lambert value

A mathematician suggested comparing the #1049 construction with an earlier irrationality paper.

What changed: The comparison was carried out: a source note and a Lean module recording where the two methods part are public, and the #1049 record cites the paper where they meet.

Name withheld for now. Exact public files and evidence.

Explaining the purpose of the work

A reader could not identify the system’s goal, the object being shown, or why its interface was useful.

What changed: The correspondence rules now require one concrete object, its purpose, a narrow request and a clear reading or time cost; those rules are also published in the credit policy.

Name withheld for now. Exact public files and evidence.

Agent acknowledgements

These contributions were already acknowledged in the linked public repositories. The links distinguish discussion, reported disposable replays and implementation, and record the resulting change and its limits. Dates in the JSON record when the credit was first published or accepted.

Sources and other acknowledgements

Literature, upstream code, tools and general AI-assistance disclosures remain with their existing source owners.

Download the credit record (JSON)