Brands will be clamoring to get this research into action.
The mechanics of viral sharing have proved a complicated subject for researchers. They’re just keeping pace with the continual flux in content-sharing tools and mechanisms, but now a new set of variables have allowed researchers at Stanford to predict viral events up to 80% of the time, a statistic that’s catnip to brands trying to keep their products on the minds of customers.
The term that the researchers focused on was a ‘cascade’ – an event in which photos or videos are shared multiple times. Their research will be presented at the International World Wide Web Conference, where a team of scientists, including Jure Leskovec, assistant professor of computer science, Stanford doctoral student Justin Cheng, Facebook researchers Lada Adamic and P. Alex Dow, and Cornell University computer scientist Jon Kleinberg, will present their formula for predicting with 80% accuracy when such a ‘cascade’ will double in shares. Their discoveries are based on their initial finding, using over 15,000 Facebook photos (stripped of names and identifiers), that at any point during a ‘cascade,’ there was a 50-50 chance that the number of shares for a given item would double.
For a long time, it was uncertain whether cascades were even worth measuring. According to research recently provided by Facebook and university researchers, only 1 in 20 photos posted on the site gets shared even once, and just 1 in 4,000 gets over 500 shares. “It wasn’t clear whether information cascades could be predicted because they happen so rarely,” Jure Leskovec, assistant professor of computer science, told Stanford’s news website. Events larger than 500 shares can reach all the way up to the range of Gangnam Style’s 2 billion shares, and was difficult to determine what was worth investigating for cascades. As it turns out, though, these events are essential to how culture spreads on the internet.
The more easily you can think of an example of viral content, the more likely their rules are to have applied to its history. The accuracy of their predictions goes up according to the number of shares a given photo gets, going as high as 88%.
[h/t] Stanford News