Despite networked connections to millions of people and ideas, internet users tend to silo themselves from opposing points of view.

Mapping the Online Life of Science Stories

Sometimes it seems like information is shared on two rivaling internets: One, a utopian palace where digital democracy lifts the most valuable ideas from the fray. The other, a darker place where the cacophony is overwhelming and people read and share only things that exist inside a networked bubble.

Despite networked connections to millions of people and ideas, internet users tend to silo themselves from opposing points of view.  (Visual by

Despite networked connections to millions of people and ideas, internet users tend to silo themselves from opposing points of view. (Visual by

Since 2011, researchers at the Berkman Center for Internet and Society at Harvard University and the Center for Civic Media at MIT have been studying how ideas actually spread online using an innovative tool called Media Cloud. Media Cloud is an open-source program that allows researchers to capture articles, press releases, and social media — and track how postings influence individuals or groups online through linking, shared language, or social media exchanges. Using the program, the research team has generated striking reports like this map of what sources pushed Trayvon Martin’s name into headlines after his death.

Recently, a group of researchers used the tool to examine how conversation about health infiltrates social groups on the web, using Ebola and vaccination as case studies.

The team used the program to pull media published at the height of the Ebola epidemic, between July 1 and Dec. 1, 2014. When they studied what stories were linked to and what information was shared, they found a notable discrepancy: Within media reports, the most linked-to stories were primary sources from scientific organizations, like the U.S. Centers for Disease Control and the World Health Organization. But when they searched for what was being shared on Twitter, it was mostly political stories about Ebola policy.

“That’s where we felt like the breakdown started to happen,” says Brittany Seymour, a professor at the Harvard School of Dental Medicine and one of the researchers on the project.

The Harvard and MIT researchers also found that words like “quarantine” and “ban” were used frequently by the readers sharing political stories. Yet those words were largely absent from the scientific sources, since health officials did not recommend domestic quarantines or immigration bans during the Ebola outbreak.

“This leads us to believe that what’s driving coverage of Ebola in the U.S. is fear of domestic infection,” says Hal Roberts, a fellow at the Berkman Center, “not understanding of the disease.”

While both Roberts and Seymour stressed that their research on vaccinations is still preliminary, they have found some interesting results here as well.

The groups sharing information about vaccines online fell into four fairly distinct categories: a vaccine-hesitant group; a ‘pro-vaccine’ group; members of the science and medicine community; and the mainstream media. Typically we think of media organizations as a bridge between people with opposing viewpoints. But the researchers found that, for the most part, people in the anti-vaccination community weren’t linking to or acknowledging mainstream media reports.

Roberts thinks that’s a significant insight that could change how we communicate vaccination information. “If you’re the CDC and your goal is to get more people to change sides in a debate, one strategy might be to work hard to get The New York Times or The Washington Post to write more stories to be more pro-vaccination,” he says. But that wouldn’t be a helpful approach, “because none of the sources in the anti-vaccine community are paying much attention to the mainstream media anyway.”

Similarly, rather than promoting the most recent science on vaccination, the pro-vaccination group worked mostly as debunkers of the anti-vaccination group. “It was like, ‘They’re dumb, look at what they’re sharing now,’” says Seymour. “They weren’t necessarily promoting science as much as they were de-promoting what was happening within the anti-vaccine community.”

Eventually, the research team hopes that analyzing how information spreads online will help them be able to create a data-driven road map for how scientific institutions can get their message across to disparate groups—something that’s especially hard to do on the internet.

“When someone tweets something they’re really saying, ‘this is my identity online,’” says Seymour. “It’s less about the validity of the information, and more about the sociology of who you think you are.”