Predicting What Will Go Viral
WHY YOU SHOULD CARE
Knowing what’s going to go big on the Twitterverse isn’t a guessing game anymore.
Oh, the randomness of the Internet, that fickle beast. Could anyone have really predicted the rise of the BuzzFeed quiz, the Upworthy headline (“You’ve Got To See This!”), the popularity of cat photos with misspelled captions? The answer, surprisingly, is maybe — if you were paying attention to the right people on Twitter.
Researchers from the U.S. and Madrid figured they’d be doing well if they could map viral spreads a few hours before they exploded across the Twitterverse. Stunningly, they found they could predict them — as early as 2 months in advance.
The scholars achieved their online sleuthing by checking thousands of Twitter accounts and identifying followers who seem to be more plugged in than others — people known as “sensor-friends.” The scientists tracked those folks who appeared at the center of their friends’ circles to see if they might be harbingers of online virality.
They were. To a stunning effect.
The team predicted the rise of #Obamacare two months before its peak on Twitter and three months before the Google search peak, according to the universities.
“We were really surprised,” says James Fowler, professor of medical genetics and political science at the University of California-San Diego, in a Yale University story. “We thought the method would give us a few hours early warning, but instead it gave us several days, and sometimes even weeks or months.”
Caveats exist. Viral info isn’t created equal, and so not all tracking succeeds. Certain daily news items, like explosive sports results, proved tricky, the team reported. But slow-boiling movements, or trends on the verge, the scientists could predict when they’d go big. And it’s stunning how few accounts they tracked in order to guess the next big movers.
Beyond the gee-whiz factor, experts say the study has some very concrete real-world potential.
Of some 255 million active Twitter accounts, the viral study could make accurate predictions based on 50,000 accounts — a fraction of a percent of all users.
For businesses, this means it really is possible to track what’s going to break big — from the stock market to the box office. Could the next Gangnam Style international video sensation be predicted? More seriously, the next real-world virus outbreak?
“Public health officials around the world could use our sensor method to mount a quicker and more focused response to health epidemics,” says Dr. Nicholas Christakis, of Yale’s School of Medicine. “It would give them more lead time to save lives.”