Contagion for the first time. For those who haven't seen it, the movie follows several people as they try to cope with a deadly global pandemic of an unknown virus.
While following the spread of the virus, the movie also draws parallels between the disease epidemic and the spread of information (or misinformation) through blogs and Twitter.
Contagion briefly touches on the models that scientists use to track epidemics, and real-life scientists have been applying these models to information spreading online. New research, however, suggests that the parallels between information exchange and disease spreading are shallow.
In other words, 'viral' snippets of information, such as those that helped propel the Occupy movement into the national spotlight, may not spread like a virus. Instead, there's subtle distinctions that reveal 'viral' in this context can be a bit of a misnomer, according to the research.
The Susceptible, Infectious, Recovered (SIR) model has become the most commonly used tool to model the progress of an epidemic. In this type of model, one virtual person belongs to one of the three aforementioned categories. An infectious person, for instance, has the ability to spread the disease to his neighbors, and the likelihood of spreading depends on certain variables that relate to a specific disease.
For this new research, physicist Javier Holthoefer and his colleagues from the University of Zaragoza in Spain decided to apply these models to online "rumor" spreading. The team used data linking various Twitter users and separate data that connected different political blogs. On top of this foundation of data, the researchers applied a traditional epidemic model to see if it would match up with actual interactions found online.
Holthoefer found that regardless of which set of data they used (e.g. blogs or Twitter), the epidemic model was missing a key aspect of online interactions: influential spreaders. Influential spreaders serve as gate keepers of information, and they tend to have the power to quash a budding rumor before it spreads widely.
"On the experimental side, we're seeing that these [epidemic] dynamics are lacking some genuine human ingredient," said Holthoefer.
In addition to the role of influential rumor spreaders, Holthoefer's research revealed another important distinction between epidemics and information spreading. Under epidemic models, infectious people recover after a period of time that depends on probability. Infectious people can spread a disease as long as they remain contagious, but rumor spreaders stop sending out their message once all of their followers have been reached.
"When the neighbors know the rumor, there's no reason to keep propagating it," said Holthoefer. "It's lost value."
Now, Holthoefer and his team hope to analyze Twitter exchanges to create a better model specifically tailored to online information spreading. Although their next research step remains preliminary, they have already gleaned new information from the online exchanges.
For instance, Holthoefer notes that how frequently a user posts information can be an important factor overlooked by epidemic models: "We're introducing this ingredient, and this changes the output of the dynamics," he said.