We checked every paper we've covered. Here's what we found.

Yes, Lucent actually runs the check it advertises, and no, it has not yet flagged a retraction. As of 28 July 2026, Lucent has published 13 issues covering 151 papers. Of those, 47 distinct papers carry a DOI published openly in the archive, checkable by anyone, not just paying subscribers. We ran all 47 through the same live Crossref lookup our own /check tool uses: does the DOI resolve, and does Crossref's metadata show a retraction, correction, or erratum. The result: 47 of 47 resolved. Zero of 47 carry a retraction flag. That is not a disappointing result to explain away. It is what a working check on a young, small archive of recently published papers is supposed to find, and it is the reason a future flag, if one ever comes, would be worth believing.

Why we're doing this in public

Lucent's whole pitch is a disclosure layer: before you read a paper we've translated into plain language, you can see exactly what we checked about it, not just take our word that we checked anything. Every issue page prints an "Integrity screen" line under each paper, with a checked date, a DOI link, and a plain retraction status. That is a promise made in public, issue after issue. A promise made in public should be auditable in public too, so this piece turns the same check back on Lucent's own coverage, using the same public data source anyone else could use, and states the real numbers rather than a rounded or estimated one.

The method

The full methodology lives on /integrity and we won't repeat it here. In short: before a paper is covered, Lucent confirms its DOI resolves and reads Crossref's metadata for the paper, checking for an update-type field that would indicate a retraction, correction, or erratum. For this audit we did the same thing again, independently, on 28 July 2026, using live queries against api.crossref.org/works/<doi>, the identical public endpoint /check uses. Nothing here relies on Lucent's own past record of what it found; every result below comes from a fresh Crossref query run today.

The tally

Real numbers, pulled directly from the live /issues/ archive, issue-00 through issue-12, the latest published as of this audit:

A word on why 151 and 47 are different numbers, because blurring them would be dishonest. Each issue's table of contents lists every paper in that issue, free and subscriber-only alike. But the DOI link and the printed integrity screen, the part anyone outside Lucent can independently verify, is only published in the static page for the free portion of each issue: 3 papers for issue-00 (Lucent's 7-paper launch issue) and 5 papers for every issue since (issues carry 9 to 15 papers total). That free portion is the only part of the archive with a public DOI to check, so it is the only part this audit, or any outside reader, can actually re-verify today. We are stating that boundary plainly rather than implying we screened all 151.

Those free-tier slots add up to 63 paper mentions across 13 issues, but not all 63 are different papers. Lucent revisited five papers a second time: issue-02 published the identical five-paper free selection as issue-01, and four other stretches of adjacent issues (04/05, 06/07, 08/09, and 09 through 12) each share one to three repeated papers between neighbouring free sets. Once duplicates are removed, 47 distinct DOIs remain, and that is the set we checked.

What we found

Zero of 47 checked DOIs carry a retraction, correction, or erratum flag in Crossref's metadata, and all 47 resolve. Every paper came back with a matching journal record: Nature, Science, Science Advances, Nature Communications, Nature Medicine, Nature Genetics, Nature Astronomy, Nature Climate Change, Nature Human Behaviour, PNAS, Geophysical Research Letters, and BMJ all returned clean.

Stated plainly, per how we handle every other "we looked and found nothing" result on this site: a clean record here is expected, not remarkable. Lucent's archive is six weeks old (issue-00 published 3 July 2026, issue-12 on 28 July 2026) and every paper in it was, by definition, not retracted at the moment Lucent covered it, or the DOI-resolution and retraction check built into the publishing process would have caught it before it went out. Retraction Watch's own database put the total historical count of retraction records at 63,000 as of a post dated 30 December 2025, built from millions of papers published over decades. A newsletter's first 151 papers, most of them weeks old, landing at zero is the base rate you would expect, not a surprising absence. What this result actually establishes is narrower and more useful: the check runs, it queries the real source, and it returns a real answer instead of a rubber stamp. A rubber-stamp check would also say "no problems found" on every paper, which is exactly why the number alone can't prove anything: what would prove something is what happens the day a checked paper's Crossref record does change, and whether that shows up here.

What this check cannot tell you

Read the archive's own line, printed on every issue: "This screen checks process signals. It does not tell you the findings are true." A clean DOI-resolution and retraction check confirms three narrow things: the paper's identifier still points somewhere, Crossref has a metadata record for it, and that record carries no retraction-type flag today. It says nothing about whether the paper's methods were sound, whether its images were manipulated, whether its statistics hold up, or whether it will ever be retracted in the future. It also is not a standing subscription: each check is a snapshot on the date it ran, not a monitor that watches a paper forever after. And it depends on Crossref's own metadata being current. Crossref hosts and rebuilds the Retraction Watch dataset regularly, but a very recent retraction notice can take time to be indexed anywhere, Lucent's check included. None of that makes this audit meaningless. It means what it proves is narrow, and we would rather understate it than round up.

For base-rate context on retractions across the wider literature, not Lucent's own archive, see how many papers get retracted, with sourced numbers. For the methodology this audit re-tested, see how Lucent's screen actually works. The raw material for everything above is public: every issue, with every printed screen result, is in the archive.

How the screen works  ·  How many papers get retracted  ·  Check a DOI  ·  Full archive  ·  Latest issue  ·  Privacy