The Record Nobody Wanted to Build
Retraction Watch began as a side project by two journalists, Ivan Oransky and Adam Marcus, who noticed that journals were quietly pulling flawed papers without anyone writing down why. Speaking with CBS News in 2015, Oransky said the blog was then five years old, and backing from a MacArthur Foundation grant of $400,000 funded the leap from blog to database. The formal record that emerged, the Retraction Watch Database, is published by The Center for Scientific Integrity under a 2018 citation line, and its own documentation now calls it the fullest single record of pulled papers in existence.
How Big It Got
The growth became an operational problem. As of October 23, 2024, the database’s user guide explains that searches must now include criteria in at least one field, author, title, journal, publisher, affiliation, country, article type, date range or identifier, because the archive’s rapid expansion had swamped the team, at times stalling entry work altogether. The interface caps any single result view at 50 entries, with the full count shown in a banner above. Casual users can keep using the search page, but anyone wanting the whole dataset is pointed elsewhere: the database was acquired by Crossref, which makes the complete file openly available for bulk download. The guide is candid about limits, too. Coverage of corrections and expressions of concern is deliberately incomplete, skewed toward cases tied to existing retractions or high-profile studies, and any figures drawn from those categories should not be read as representative of publishing at large.
Why the Numbers Matter
The scale of the underlying problem is what made the project necessary. Oransky estimated at least 500 retractions a year by 2015, a rise of more than tenfold since 2001, against the 2 to 3 million papers that ship every year, a share of just 0.02 percent. About two-thirds of retractions trace to researcher misconduct, faked data and plagiarism among them, which averages out to roughly one retraction per day for dishonesty alone. And detection is the bottleneck. “The number of people who are convicted of a crime does not necessarily reflect the number of people who have committed a crime,” Oransky told CBS News. In a New York Times opinion piece, he and Marcus made the counterintuitive case that visible corrections are healthy: “Retractions can be good things, since even scientists often fail to acknowledge their mistakes, preferring instead to allow erroneous findings simply to wither away in the back alleys of unreproducible literature. But they don’t surprise those of us who are familiar with how science works; we’re surprised only that retractions aren’t even more frequent.”
The Cautionary Tale
The case that put the site on the map in 2015 was the Science paper on gay canvassers by graduate student Michael LaCour. When replication attempts unearthed inconsistencies, his coauthor, Columbia professor Donald P. Green, asked for retraction. Green wrote: “I am deeply embarrassed by this turn of events and apologize to the editors, reviewers, and readers of Science.” Oransky predicted swift action, “Given all the press, I wouldn’t be surprised if they made a decision on it quite quickly,” and tied the episode to a wider lesson: “We should be concerned about this and it should force us to think about the incentives that scientists are working under and whether or not peer review is equipped to pick up misconduct and fraud.” Retracted claims also outlive their corrections, as the 1998 vaccine-autism study, pulled 12 years later, still demonstrates. Once a notion takes hold online, he said, “it’s difficult to un-ring that bell.”
Sources
- Retraction Watch Database User Guide – Retraction Watch
- How often are scientific studies retracted? – CBS News
