We re-checked all 209 papers we have published. Here is the result.

What was actually run

One request per DOI to https://api.crossref.org/works/, the public Crossref works endpoint. For each record we read the update-to, updated-by and relation fields for a retraction, correction or erratum marker. That is the same screen the pipeline runs before an issue goes out and the same one the free checker runs for any reader, on the same endpoint, with no account and no key.

None of the 209 came back with a retraction marker, and none of the DOIs failed to resolve. That is the result we expected and it is worth being plain about why: these are papers screened once before publication and then re-screened weeks or months later, in venues where retraction is rare and slow. A clean sweep here means the plumbing works. It is not a claim that Lucent caught something.

The command is npx tsx src/cli/self-audit.ts in the pipeline. It writes a dated artifact, and this page renders from that artifact rather than from prose, which is why it cannot quietly fall out of date the way its first version did.

What a clean result does not prove

Crossref records what publishers tell it. This screen reads that record. So it can tell you a paper has been formally withdrawn, and it cannot tell you a paper is right.

• It does not read the paper, check the statistics, or evaluate the claim.
• It does not detect problems nobody has reported yet. Most retractions arrive months or years after publication, and until the notice exists there is nothing in the metadata to find.
• It is a snapshot with a date on it, not a subscription that watches a paper forever. That is why this page is re-run rather than written once.
• It says nothing about peer review quality, sample size, funding, or conflicts of interest. Those checks are a separate guide.

The full method, including its limits, is written out here.

Check any of it yourself

Every row below links to the paper at doi.org. Paste any of those DOIs into Lucent's checker and you will get the same answer this run got, because it is the same request. The screen result is also printed permanently next to each paper in the public archive, so if one of these ever fails, it fails in public and not only on this page.

150 of these DOIs also have a permanent result page of their own, linked from the DOI itself. The rest are being added in weekly batches. Each of those pages re-runs the check live when it is opened, so they are released slowly rather than all at once: Crossref's API is free and public, and a crawler finding 209 of them in one afternoon is a burst somebody else pays for.