Who answers for Pith, and how a review is put right.

The Recognition Physics Research Institute runs Pith and answers for it. Who runs it, and the conflict of interest that comes with that, are on the operator page.

The institute's founders are Jonathan Washburn and Emma Tully. Their public identifiers are on the colophon.

Which person decides a challenge, signs a correction or answers a complaint has not been published yet.

No adviser from outside the institute has been named.

Every review is written by a machine and can be wrong. Anyone with a free Pith account can challenge a factual error or a specific failed rule. The challenge is signed and public, and any response or revision stays attached to the paper's record. Challenges and their outcomes are listed at challenges.

When a challenge is upheld, the passage is struck through on the review, or the review is run again and the original stays linked from the page. The stored review is signed; a correction is shown on the page and the stored text is kept.

An author can reply beside the review after claiming the paper with an ORCID iD. What an author can do about a review of their paper, and what Pith does not do, is on the page for authors.

How often reviews are right is measured and published at accuracy.

A paper posted on Pith can be withdrawn by its author, from the paper's own page. Its text, its PDF and its review then leave the site, and the log of decisions about it is kept. The author chooses what the paper's address shows afterwards: a short notice with the title, the authors, the DOI, the date and the reason, or nothing, in which case the address answers “not found”. The states a posted paper can be in are set out in the preprint policy.

An author who cannot sign in can ask an editor to withdraw the paper: write to hello@pith.science.

Review text and scores are written by Pith's machines and may be reused with attribution to pith.science. Paper titles and abstracts stay under the terms of their sources.

No named licence has been attached to review text yet.

Each review is signed, and the signing key is published, so a copy can be checked without Pith. A paper posted on Pith and put on the record carries a proof of the exact file anchored in Bitcoin and, when it is not already on arXiv, a Crossref DOI.

Every current verdict is written to a file once a month. The list of files is public; the files themselves are sent on request today (bulk data).

Pith names no outside archive that holds a copy of the record.

Write to hello@pith.science. Every message goes to that address and a person reads it. Say which page or paper you are writing about and what is wrong.

If the complaint is that a review is wrong about a paper, a challenge is the faster route: it is public and its outcome is recorded with the paper.

Pith has not published times for an answer or a route of appeal.