The Internet Loves Bad News. And That's Bad.
Derek Thompson
Illustration by Matt Chase / The Atlantic
This is Work in Progress, a newsletter by Derek Thompson about work, technology, and how to solve some of America’s biggest problems. Sign up here to get it every week.
Last week, I saw a new paper in the journal Nature Human Behavior called “Negativity Drives Online News Consumption.” That seems bad, I thought. Naturally, I clicked.
In a randomized study of 105,000 headlines and 370 million impressions from a data set of articles published by the online news dispensary Upworthy, researchers concluded that each negative word increased the click-through rate by more than 2 percent. “The presence of positive words in a news headline significantly decreases the likelihood of a headline being clicked on,” they said.
Are you even remotely surprised by any of this? Probably not. Neither was New York University’s Claire E. Robertson, a co-author of the paper. “People have been saying ‘If it bleeds, it leads’ for decades,” she told me. But what does that actually mean? Maybe substantively bad news naturally gets more attention, as it probably should. Or, maybe, even humdrum and unimportant stories can be juiced to attract eyes and ears if editors inject their headlines with a dose of sadness and catastrophe.
Upworthy might seem like an unusual choice for studying the properties of hard news, given that the site is typically associated with frivolous curiosity-gap bait: “This Baby Panda Learned to Breakdance—What Happened Next Will ASTONISH You,” and so on. But its database offers an unusually perfect opportunity to test the effect of headlines on audience behavior, because the site has made public the headline tests it ran for many news stories. This way, Robertson and her co-authors could control for the substance of each article, because some stories (a Harry Styles breakup, for example) will always get more clicks than others (a new law for Vermont pension accounting, say). “Even controlling for the same news story, framing more negatively increases engagement,” Robertson said.
Although blaming journalists and editors for this bias is easy, it’s also too simple. After all, it’s audiences who are reading—and watching, clicking, and subscribing to—all this stuff. (An alternate media maxim might be “If it bleeds, she reads.”) Even public-service-minded editors and journalists may feel they need to shape their coverage to match the decisions and emotional dispositions of their consumers. Negativity is not, strictly speaking, a news-maker problem; it’s a human problem—or, more to the point, a collective-action problem, in a dual-sided marketplace.
The internet is not best understood as a big room full of people screaming at one another about breaking news and policy debates. In fact, the room for political yelling is one of the smaller antechambers of the house of online content. One study of internet users in Poland found that news accounts for barely 3 percent of people’s digital-information diet. Much of the rest of the online world is populated by joyful gossip and animals doing stuff. In fact, a 2021 analysis of 126,301 Twitter posts found that rumors with positive emotions were much more likely to go viral, overall.
But although news makes up a small fraction of online content, this is where negativity seems to have the biggest lift for traffic. Robertson said her research validated several other studies showing that people are “especially likely” to consume political and economic news “when it is negative.” Surprisingly, to both me and the researchers, the study did not find that anger increased clicks; instead, sadness seemed to drive traffic in the Upworthy data set. But other research has found that high-arousal emotions, such as outrage, are most likely to be shared by users.
“There’s evidence that the people who post and retweet are both in the minority of online users and tend to be more extreme than the average user,” Robertson said. “When taking this into account, it’s logical that high-arousal content is most often shared or posted, even when it’s not what people are most interested in.”
Derek Thompson
Illustration by Matt Chase / The Atlantic
This is Work in Progress, a newsletter by Derek Thompson about work, technology, and how to solve some of America’s biggest problems. Sign up here to get it every week.
Last week, I saw a new paper in the journal Nature Human Behavior called “Negativity Drives Online News Consumption.” That seems bad, I thought. Naturally, I clicked.
In a randomized study of 105,000 headlines and 370 million impressions from a data set of articles published by the online news dispensary Upworthy, researchers concluded that each negative word increased the click-through rate by more than 2 percent. “The presence of positive words in a news headline significantly decreases the likelihood of a headline being clicked on,” they said.
Are you even remotely surprised by any of this? Probably not. Neither was New York University’s Claire E. Robertson, a co-author of the paper. “People have been saying ‘If it bleeds, it leads’ for decades,” she told me. But what does that actually mean? Maybe substantively bad news naturally gets more attention, as it probably should. Or, maybe, even humdrum and unimportant stories can be juiced to attract eyes and ears if editors inject their headlines with a dose of sadness and catastrophe.
Upworthy might seem like an unusual choice for studying the properties of hard news, given that the site is typically associated with frivolous curiosity-gap bait: “This Baby Panda Learned to Breakdance—What Happened Next Will ASTONISH You,” and so on. But its database offers an unusually perfect opportunity to test the effect of headlines on audience behavior, because the site has made public the headline tests it ran for many news stories. This way, Robertson and her co-authors could control for the substance of each article, because some stories (a Harry Styles breakup, for example) will always get more clicks than others (a new law for Vermont pension accounting, say). “Even controlling for the same news story, framing more negatively increases engagement,” Robertson said.
Although blaming journalists and editors for this bias is easy, it’s also too simple. After all, it’s audiences who are reading—and watching, clicking, and subscribing to—all this stuff. (An alternate media maxim might be “If it bleeds, she reads.”) Even public-service-minded editors and journalists may feel they need to shape their coverage to match the decisions and emotional dispositions of their consumers. Negativity is not, strictly speaking, a news-maker problem; it’s a human problem—or, more to the point, a collective-action problem, in a dual-sided marketplace.
The internet is not best understood as a big room full of people screaming at one another about breaking news and policy debates. In fact, the room for political yelling is one of the smaller antechambers of the house of online content. One study of internet users in Poland found that news accounts for barely 3 percent of people’s digital-information diet. Much of the rest of the online world is populated by joyful gossip and animals doing stuff. In fact, a 2021 analysis of 126,301 Twitter posts found that rumors with positive emotions were much more likely to go viral, overall.
But although news makes up a small fraction of online content, this is where negativity seems to have the biggest lift for traffic. Robertson said her research validated several other studies showing that people are “especially likely” to consume political and economic news “when it is negative.” Surprisingly, to both me and the researchers, the study did not find that anger increased clicks; instead, sadness seemed to drive traffic in the Upworthy data set. But other research has found that high-arousal emotions, such as outrage, are most likely to be shared by users.
“There’s evidence that the people who post and retweet are both in the minority of online users and tend to be more extreme than the average user,” Robertson said. “When taking this into account, it’s logical that high-arousal content is most often shared or posted, even when it’s not what people are most interested in.”
Similar Readings (5 items)
Talk, Talk: How Many Words Do We Use in a Day?
Summary: Wikipedia says traffic is falling due to AI search summaries and social video
Over 4,500 WordPress Sites Hacked to Redirect Visitors to Sketchy Ad Pages
Conversation: Wikipedia says traffic is falling due to AI search summaries and social video
New York Times subscribers top 10 million mark
Summary
Title: "Negativity Drives Online News Consumption - Study"
Researchers from New York University found that negative words in news headlines increase click-through rates by over 2%. The study, published in Nature Human Behavior, analyzed 105,000 headlines and 370 million impressions from
Researchers from New York University found that negative words in news headlines increase click-through rates by over 2%. The study, published in Nature Human Behavior, analyzed 105,000 headlines and 370 million impressions from