An average citizen with just a few hundred Twitter followers fires off a 140-character zinger during the presidential debate. The message seems to spread rapidly, apparently leaping across personal networks and ultimately being shared around the country. A celebrity backs a human rights campaign by sharing a video on Facebook, setting off what appears to be a contagious chain of peer-to-peer sharing that finally commands a spot on the evening news.
Stories of communications pandemics abound. We live in the age of the viral phenomenon, where the potential for virality is ubiquitous. Every story, image, meme, and sentiment contains the ember of this hanging fire. The metaphor is meant to imply, of course, that a bit of information might ricochet and expand across a population as a virus would, branching out from a single, original node—a patient zero.
But do we have our mental models, and terminology, right? Does information on social media networks typically behave like a disease? No doubt, one can see what looks like dramatic proof—millions of pageviews, shares, Likes, or retweets. But what is typically going on beneath, within the network graph that connects all the dots?
Getting eyeballs online is a lot of work, and sweat. Pulling attention away from big media sources remains difficult in the extreme—perhaps more so than anyone in earlier Internet days expected.
It may seem counterintuitive, but very little of what we are seeing online these days is actually viral in a technical sense, according to recent research. When data scientists peer into the gearbox of our social networks, they do not usually see long chains of peer-to-peer sharing. More often, a big broadcast of information to many people simultaneously explains the reach of information. This is being confirmed by those doing the biggest, most comprehensive analyses of network data, such as the Microsoft Research team with Twitter data and the Facebook data science team with its own network data. We can speculate and hyperventilate about a democratizing information ecosystem, but scientific network analysis reveals the true pattern. And frequently it does not look like a grassroots phenomenon.
A spectacular, recent example of this potential confusion might be illustrated by the group celebrity selfie photo taken by 2014 Academy Awards host Ellen DeGeneres, which generated more than three million retweets, a record, and crashed Twitter’s technical platform. Across the Web, commentators were quick to deem this a viral phenomenon. But some 43 million people were watching the Oscars. It very well may be that the majority of the effect was caused by the broadcast, and the ensuing social media cascades were short and shallow, not extending beyond a couple of people through most social networks. In other words, this organic viral phenomenon may just be a traditional broadcasting event—a particularly compelling one, but not one driven by peer-to-peer networks.
A FEW THINGS DO go naturally and spontaneously viral, but these are the exceedingly uncommon exceptions. Given that millions of pieces of content are being uploaded every minute on platforms such as Facebook, the true viral success rate is vanishingly small, researchers say.
For the most part, behind any mass online phenomenon there is a powerful group with a megaphone—still usually a big media outlet—doing most of the yelling. We often think of social media platforms as somehow more democratic, but they often function as yet another television channel. Most content is consumed passively.
“We are not seeing the end of hierarchy,” says Duncan Watts, principal researcher at Microsoft Research who has been involved in some of the biggest studies to date on social networks and information diffusion. “We may be seeing the replacement of one hierarchy with another hierarchy. We may be seeing the replacement of one set of gatekeepers with another set of gatekeepers…. But we’re certainly not seeing an egalitarian world where everything has the same chance to become known or accessible.”
So we may have our language and our metaphors wrong.
This matters because we are increasingly invested in seeing the online world as the primary mover in democracy, in creating a level playing field for news, information, and ideas that can allow more diverse activism and awareness. In Going Viral, Karine Nahon and Jeff Hemsley of the University of Washington argue that viral phenomena create a “self-organized interest network” outside the control of traditional power structures. The gatekeepers, in other words, cannot control the grassroots in these situations.
We have seen the power of networks during situations of extreme political unrest around the world, when attention is funneled and social networks are wired for transmission around a single topic. But big events are no guarantee of a more democratic communications system.
A May 2014 study published in PLoS One analyzed activity from nearly 194,000 Twitter users during major media events in the 2012 presidential election season. The authors explored whether these moments of intense shared public attention created a “rising tide” that helped lift up more social media voices, or whether these events simply created “rising stars.” Their conclusion was that democracy actually diminishes at such moments. “The beneficiaries of this newfound attention were not distributed throughout users with different numbers of followers,” they write, “but concentrated among users with the largest pre-existing audiences.”
SPECIAL CASES ASIDE, THERE’S little evidence that peer-to-peer will be the primary, regular pathway that information travels among citizens. What we tend to forget is that open networks may allow for equal opportunity in principle, but attention is finite and some people and institutions start with greater visibility and recognition. A “power law” tends to characterize how attention is distributed: A few big winners, millions of small losers.
“Attention is an arms race,” New York University’s Clay Shirky says. Videos and memes can reach millions over time through informal media channels and networks—think Psy’s “Gangnam Style.” Large-scale social media-driven phenomena—funny videos, quirky memes—now regularly constitute a kind of asynchronous, creeping background in our media world. But “breakthrough virality will be rare and rapidly decaying,” Shirky notes.
These dynamics affect people and organizations of all kinds across the Web. A 2014 study, for example, analyzed the efforts of 257 human rights organizations to get wider media and general online attention. Because of the “zero-sum nature of public attention,” most human rights groups get little media publicity; instead, news coverage heavily favors a few larger organizations with greater resources. And on the Web, the same thing happens: Organizations compete for audience attention through social media. Among the groups studied, the top 10 percent of human rights organizations accounted for 90 percent of YouTube views, 81 percent of Facebook Likes, and 92 percent of Twitter followers.
To see viral capacity as a reliable new instrument for mobilizing the masses is to neglect how random a phenomenon is; when truly viral phenomena do break out, they are like freakish, uncontrollable weather patterns.
Where, then, does most of our widely known, commonly accessed information come from? And why is it that certain content has the appearance of virality, coming at us from multiple friends and media streams?
For the most part, behind any mass online phenomenon there is a powerful group with a megaphone—still usually a big media outlet—doing most of the yelling.
Researchers have consistently found that most content on social media derives from traditional media sources. And despite social Web triumphalism and declarations that the media industry is dead, most Americans still receive most of their news and information directly from a mainstream media outlet. Old habits die hard, and most people, it turns out, circle back to the same few information sources. Even now, 70 percent of Americans generally get their news from television, radio, and print outlets, according to a 2014 American Press Institute survey and report. Although 21 percent of respondents cite the Internet as their main source, just two percent say social media is their go-to way of getting news. Whether it’s on television, a website, or a Twitter feed, we still mostly see a “one-to-many” broadcast of information, not a “many-to-many,” democratic pattern.
Audience data confirm this, too. Seth Flaxman of Carnegie Mellon University and Microsoft researchers Sharad Goel and Justin M. Rao tracked 1.2 million Internet users over several months. They found that 94 percent of people get their news from “at most two outlets,” and almost everyone went straight to news sites. Relatively few received news passed along by their peers through social media channels. The Pew Research Center has noted that news is found through Facebook relatively infrequently—and mostly “incidentally.”
The difficult economics of media production also continue to mean that certain outlets have much higher quality and remain attractive for social reasons, as James G. Webster of Northwestern University and Thomas B. Ksiazek of Villanova University have noted. We read the New Yorker or the Weekly Standard or BuzzFeed not just for their interesting content, but as a point of connection with peers. A 2012 study by Webster and Ksiazek published in the Journal of Communication analyzed television- and Internet-related patterns across a representative sample of American households. The data suggest the “persistence of popularity”—most people are still accessing the same content from the same outlets.
For online writers and activists who spend each day vying for the public’s attention, none of this is exactly news. Getting eyeballs online is a lot of work, and sweat. Pulling attention away from big media sources remains difficult in the extreme—perhaps more so than anyone in earlier Internet days expected.
It all suggests that we often judge the Internet based on the relatively few stories of success—where democratization seems to operate—rather than the millions of failures. Viral is the exception, big broadcasts—and lonely voices whistling in the digital hurricane—are the norm. But popular perceptions have not caught up with the emerging research, which is now taking a careful second look.
“One of the reasons that we are so interested in trying to figure out what makes something go viral is because it’s so rare,” says James Fowler, a social scientist and network researcher at the University of California-San Diego. “I think the general critique that we have been studying this question by studying the successes, rather than studying both the successes and the failures, is correct.”
This article is adapted from a new Shorenstein Center discussion paper, “The Challenges of Democratizing News and Information: Examining Data on Social Media, Viral Patterns and Digital Influence.”
