This story is over 5 years old.


Researchers Retract Study That Said Fake News Is Just as Likely to Go Viral as the Truth

A paper that claimed the quality of information doesn't factor into how viral it becomes under conditions of "information overload" has been retracted.
Person holding a phone and using social media
Image: Shutterstock

In June of 2017, seven months into the Trump presidency, an academic study published in a respected journal offered a dire proclamation on the newly urgent problem of falsities spreading on the internet—so-called “fake news.”

Fake news is as likely to go viral on social media as truthful information, the study published in Nature Human Behaviour claimed, when human beings are crunched between the tandem forces of shorter attention spans and a glut of information filling our brains. It was an alarming conclusion, but it was ultimately unsupported by the evidence.


This week, the journal published a retraction notice at the request of the authors, Xiaoyan Qiu, a researcher at Shanghai Institute of Technology, and several colleagues at US institutions.

According to the retraction notice, the paper’s issues were twofold. Firstly, the researchers claim a “software bug” led to an incorrect value, and secondly, the authors report that they used “erroneous data” to model their results. Using correct data, the retraction notice states, “the original conclusion, that the model predicts that low-quality information is just as likely to go viral as high-quality information, is not supported.”

“At the highest level, this is a story about science working the way it should over time,” Ivan Oransky, co-founder of academic watchdog Retraction Watch, said over the phone. “It’s not necessarily a story about science working as we hope it would in the immediate sense, when the paper was submitted and reviewed, but it’s now been retracted, and the authors realized they made an error and came forward about it.”

“It’s always hard to admit a mistake, especially when that mistake was as widely covered and got you as many plaudits as this original study did,” Ivan continued.

The paper’s perfect storm of a buzzy phrase—fake news—and the stamp of a respected academic journal meant that it was widely covered by the media, including Motherboard. It’s not ideal that false information was left uncorrected in public for more than a year, but it’s still ahead of the average retraction time of three years, Oransky said.

When reached for comment, Nature referred Motherboard to the retraction notice. Emails sent to corresponding author Diego Fregolente Mendes de Oliveira’s personal email and that of his current academic affiliation were not immediately answered.

For journalists and readers, Oransky said, the best thing to do when confronted with a bombastic new study from a respected outlet is to put it in context with previous work on the topic. If it’s the twelfth paper confirming a certain finding, then it might have more legitimacy than a paper making a bold claim for the first time.

Science is hard, but the important thing is that the truth comes out.

Get six of our favorite Motherboard stories every day by signing up for our newsletter.