The language of contagious disease has long infected computer science. Decades ago, information security pioneer Len Adleman of the USC Viterbi School of Engineering applied the term "virus" to malicious code that could take over computers.
More recently, the word "viral" has been used to describe the way ideas or "memes" circulate through the Internet, with unusually popular contagious ones attaining epidemic proportions. Marketers and opinion makers of all kinds are eager to understand and control this process in order to get words or products into wide circulation.
Kristina Lerman
But what disease does the spread of an idea on Digg or another social network resemble? How contagious are memes, compared to say, HIV? How do idea epidemics happen, how far do they spread, and what stops them?
Kristina Lerman, a colleague of Adleman’s at the USC Viterbi School Department of Computer Science and a Project Leader at the School’s Information Sciences Institute, has long been studying patterns of information spread on Digg, Twitter and other social networks. In a paper she co-authored with two colleagues entitled, “What Stops Social Epidemics?” presented July 18 at the 5th International Conference on Weblogs and Social Media (ICWSM), she delves into the similarities and differences between disease and Internet epidemics.
Social epidemics, Lerman notes, are much easier to trace than outbreaks of disease. Working out the mathematics of disease epidemiology involved intense and unrelenting house-by-house work by public health officials and doctors tracing back all the contacts of each patient. Checking spread of "like it" votes on Digg or retweets on Twitter is more automatic and much easier.
Lerman defines “infected” as a social media user who posts or recommends some content to his or her followers. Then by retweeting the content or voting for it on Digg, the followers themselves become infected and go on to infect others.
The mechanism allowed Lerman and colleagues to trace infection/information flow.
“We found that social epidemics look and spread very differently from diseases,” Lerman said. "Contrary to the expectations raised by the disease analogy, the vast majority of information cascades failed to reach "epidemic" proportions. Rather than propagating to thousands or tens of thousands of people as would be expected of a viral outbreak, most of the information cascades on Digg reached just hundreds of people.”
Why? The answer, according to the USC group, appears to lie in the fundamental difference between exposure to disease and exposure to, say, jokes.
In disease, repeated exposure to multiple carriers increases the likelihood of infection, often drastically. In social networking, it doesn’t seem to. Social networkers are not more likely to repeat a tweet or like a Digg if eight or ten of their contacts like it than if only one or two do. In fact, the reverse is true.
“The fundamental difference between the spread of information and disease is: despite multiple opportunities for infection within a social group, people are less likely to become spreaders of information with repeated exposure,” Lerman wrote.
“In the end, there may be a good reason for this difference,” Lerman said. “Imagine if ideas and information did indeed spread like viruses. We would all be drowning in information.”
Lerman's collaborators on the paper were Greg Ver Steeg, postdoctoral researcher at the USC VIterbi Information Sciences Institute and USC Viterbi School computer science graduate student Rumi Ghosh.
Lerman has posted a non-technical exposition of the research at:
http://thebigblogtheory.wordpress.com/2011/04/05/s04e20-the-herb-garden-germination/
The National Science Foundation and the Air Force funded the work.
