The arXiv paper
One paper carries the part of this programme that was held to a publication standard. It is a 98-page selection of the eight papers of the series — not a summary of them, but the subset that survived the process described here. The ninth paper on this site was written afterwards and is outside it.
What "refereed" means here
The text went through thirteen blind referee rounds, each read by a fresh model instance with no memory of the previous round, each returning a written report with blockers — plus one cross-model control round, in which a different model family read the locked text. Every printed constant was re-derived centrally, at the site, rather than accepted from the report. The last round returned no mathematical error in any proof and no printed constant that failed to reproduce.
One reporting error was found after the rounds. A label on a fit model in §5 had been carried over wrongly from the source paper — a defect the sealed rounds could not see, because it lived in the draft they were sealed to. It was found by a non-blind cross-read against the source record and corrected at v1.6; no measured quantity moved.
Everything on the rest of this site is outside that process.
The entry point is the abstract, the one statement claimed as a theorem, and the table that tests it — copied from the paper's source, nothing added. The reading form and the PDF are generated from the same source file, so they carry the same text; the PDF is the typeset version of record.
Status
The paper is built and frozen, but it is not yet posted. Submission to math.NT requires an endorsement, and I do not have one yet. The arXiv identifier will appear here on posting — until then the copy above is a pre-posting draft, and the paper says so in its own front matter: a copy with no persistent identifier is a draft.
A request for endorsement
arXiv endorsement — math.NT
arXiv asks first-time submitters in a subject class to be endorsed by an established author in it. What arXiv asks of an endorser is less than a refereeing: not that the paper is publication-ready, but that the submitter is a legitimate member of the community and the work is a legitimate contribution to math.NT, appropriate for the class. If you are eligible to endorse for math.NT and, having looked at the two-page entry point or the paper, judge it that, I would be grateful for your endorsement. If you look and do not think so, declining at the same link is just as useful to me — and a line saying why would be worth more than the endorsement.
Ondřej Dvořák requests your endorsement to submit an article to the math.NT section of arXiv. To tell arXiv that you would (or would not) like to endorse, visit:
https://arxiv.org/auth/endorse?x=P7NVV3
If that link does not work, go to arxiv.org/auth/endorse.php and enter the six-character endorsement code:
P7NVV3
The shortest honest route in is the two-page entry point. The full text is linked above — the PDF is the version of record, and the referee history it went through is described at the top of this page.
The other 34 claims
The programme records 62 substantive claims. The paper carries 28. The rest are either measured under a pre-registered gate but never searched against the literature, or exploratory observations that were recorded and never gated. They are all in the papers on this site, which is the reason this site exists.