Is fake news a fake problem?

Fake news became the defining media concept of our day. But does it really warrant the attention? The data says no.

For the better part of a decade, fake news has been a defining media concept. The debate (or, really, battle) of what's considered fake news regularly fills out headlines, and is part of the daily churn of social media. Academic interest bears this out, with over 5,000 papers on the topic between 2017 and 2021—compared with "just 100 in all the years leading up to 2016," according to an important study by Duncan Watts, David Rothschild and Markus Mobius.

What Watts, Rothschild and Mobius show about the impact of fake news is even more important. Far from "shaking democracy to its foundation" (or other similar metaphors), its impact is not just negligible but barely measurable. Instead, what we really should be concerned about is the impact on what I call accredited news which delivers false or misleading information, or skews an issue through deliberate or unconscious bias.

But before we dive into that, let's do what's rarely done in this conversation and define what we mean by fake news. We'll take the definition of Duncan et. al. of fake news as:

"The deliberate spread of online misinformation"

This concept lies at the core of claims that fake news is being used to achieve political or geopolitical goals. It's not just wrong, but it's deliberately wrong in a way calibrated to produce a behavioral response by a large group of people.

Over the past six years, the effects of fake news have been cast in the most grave terms. For example, in an article titled "Social Media is rotting democracy from within," Vox made a sweeping claim about the impact of fake news: "When fake news is flying around, no matter who’s spreading it, people lose faith in the trustworthiness of their social institutions or perhaps even in the very idea of truth in politics."

Interestingly, to back up this quite bombastic, claim Vox provides a link that has since been removed. But the study that article seems to refer to says little about the impact of fake news, and it limited only to claims about the intentions of governments using it for political purposes.

Beyond this single article, the authors of Watts et. al. study, "Measuring the news and its impact on democracy," similarly take note of the "breathless" claims of "tsunamis" or "epidemics" of fake news.

The data, however, don't bear this out. Watts et. al. reference another study, this one by Hunt Allcott and Matthew Gentzkow published in the Journal of Economic Perspectives, which found that the fake news they studied related to the 2016 presidential election would have changed the total vote by two-hundredths of a percentage point (.02%).

This is in keeping with larger patterns of consumption of fake news. According to Watts et. al., fake news accounts for "less than 1% of regular news consumption and less than 1/10 of 1% of overall media consumption..." For the heaviest consumers of fake news, this translates to "1 min of fake news per day on average, compared with 106 min of regular news...and over 500 min of total media consumption."

Watts et. al. are careful to acknowledge that this doesn't mean there isn't a significant effect of fake news on society; but they do strongly suggest it's not the tsunami/epidemic of the popular media's imagination.

What, then, can account for the effect of biased—and biasing—information? The study's authors are pretty clear on that. "The mainstream media can also promulgate falsehoods by reporting on them." The researchers cite instances like widespread reporting by the media that Sadaam Hussein definitively had weapons of mass destruction, and the claim made by Sarah Palin in 2009 that the Affordable Care Act would create "death panels"—a claim reported by over 700 news outlets.

Finally, the study's authors make what I consider to be the most important point of all: that even when the mainstream press reports on true or real events, it can skew perception by employing a broad range of tactics, including:

"[P]artial or biased data, quoting sources selectively, omitting alternative explanations, conflating correlation with causation, using loaded language, insinuating a claim without actually making it (e.g. by quoting someone else making it), strategically ordering the presentation of facts, and even simply changing the headline..."

It’s one of the many ironies of our media environment that “fake news” was a bit of a fake concept—at least in the way it was deployed by the media, i.e. as a game-changing threat. The reality is subtler, as it so often is. What this highlights is the urgent need for objective, verified and verifiable data about media. That’s the source from which real change will come.

Editor’s Note: This post was first posted at the Alitheum blog, VERIFIED. At Alithem, we’re building technology to measure media bias, error and inaccuracy. Learn more at www.Alitheum.com.