A retraction is the formal withdrawal of a published scientific paper. The paper stays online, but it is marked retracted, and its findings should no longer be trusted. Here is what that means in plain terms, why it happens, and how to check any paper yourself in about two minutes.
When a journal or the authors of a study conclude that a published paper cannot stand, they issue a retraction. The paper is not deleted (that would erase the record), it is flagged. A retraction notice explains, sometimes briefly, why. From that point the paper's conclusions are considered unreliable, even though thousands of other papers may already have cited it.
Two very different reasons lead to the same label. The first is honest error: a mistake in the data, a coding bug, a sample that turned out to be contaminated, found after publication. The second is misconduct: fabricated data, plagiarism, or manipulated images. Both end in a retraction, but they are not the same, and a careful reader keeps the distinction in mind.
If a study you are about to rely on has been retracted, treat its headline claim as withdrawn. This matters most when a finding is dramatic or widely shared, because retracted papers can keep circulating in news and social media long after the retraction. The single most damaging mistake is to build a decision, or a story, on a paper that the field has already pulled back.
Find the paper's DOI (a string like 10.1038/s41586-000-00000-0). Open it at doi.org: if it resolves to the paper, the identifier is live. Then look at the paper's page for a retraction notice, or check its Crossref metadata, which records retraction status across three fields. It takes a couple of minutes and it is the check almost no one does.
Lucent runs exactly this check on every paper before it reaches you. Each paper is screened for a retraction and a live DOI at the point of publication, and we show you what we checked. We check integrity signals, not whether a paper's conclusions are correct, and we say so.
Every two days: brand-new papers in plain language, each screened for retractions and a live DOI. Three free every issue.
How to tell if a study is trustworthy · Latest issue · Privacy