The paper was retracted. The internet kept the other version.
A retraction is supposed to work like this: the journal flags the paper, the databases follow, and anyone who looks it up learns it's dead. The middle step is where it breaks, and it breaks like a software bug. One boolean, a lot of trust OpenAlex is the spine of a lot of scholarly search, and it stored retraction status as a single true/false field: is_retracted . A 2024 analysis,…
Retraction procedures often function smoothly – journals flag the paper, databases update, and subsequent searches reveal its removal. However, the process can malfunction, much like a software glitch. OpenAlex, a key player in scholarly searches, stored retraction status as a simple true/false flag: is_retracted. In 2024, an analysis revealed that this flag was inaccurate in both directions.
A single update signal could misclassify a live paper as retracted, while genuine retractions could appear as valid. The affected period was from December 22, 2023, to March 19, 2024, and the journal advised users to check and replace their data if they had used OpenAlex during this time. This issue goes beyond database problems; it propagates into every tool built on the API.
In fact, a separate analysis found that around 7.5% of papers listed as retracted on Retraction Watch are not marked as such in OpenAlex at all. This discrepancy between two sources that are supposed to agree represents a significant gap. Meanwhile, the retracted paper continues to circulate, appearing in search results, citation managers, and even mirrored PDFs on lab wikis.
The Wakefield paper, for instance, has been cited 1,577 times in Scopus, with 53% of those citations coming after its retraction. Brazilian researchers have documented similar patterns with other retracted heart-risk studies, which continue to be cited as valid years later. The retracted paper is often more accessible than the notice that caused its removal.
The fix to this problem is ongoing, as Crossref now incorporates Retraction Watch's list of retractions into its records, and tools like Zotero's retraction plugin and RetractoBot attempt to inform authors directly instead of relying on users to manually verify database corrections. Ultimately, every correction that exists only within a database is one step away from the person who needs it.
The correction must reach the reader's specific path (the citation, the browser tab, the reference manager) before it can be considered complete.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.