Following our discussion on viral content, I looked into a more scientific study on why things go viral and how users can predict that potential. A group in California, with the help of other researchers across the nation, designed a study in order to observe photos on Facebook and measure their potential to go viral, which they call “sharing cascades.” They measure the number of shares when the content is first introduced on the site, claiming the number of shares must double in order for the photo to spread quickly. If the number of shares doubles, the content has the potential to go viral. At a later stage, they begin paying attention to the number of shares the Facebook photo garnered over time, because “the greater the number of observed reshares, the better the prediction.” The researchers have even developed an algorithm and trained a machine to look for certain features in the images.
“These features include the type of image, whether a close-up or outdoors or having a caption and so on; the number of followers the original poster has; the shape of the cascade that forms, whether a simple star graph or more complex structures; and finally how quickly the cascade takes place, its speed.”
As it turns out, the algorithm is accurate almost 80% of the time. In other words, researchers can accurately predict what content will go viral 8 out of 10 times, but I’m skeptical. Although this is one of the first studies of its kind, it leaves a lot of questions unanswered and fails to address two major factors related to viral content, the first of which is the ambiguity of elements in viral content. The researchers claim their algorithm takes into account the features of the image, but it is likely impossible to encapsulate all the random features of every image. Cat videos have the potential to go viral, but no two are just alike. To truly measure the viral potential of the content, researchers may have to label the video, sorting it into a category. So, do all cat videos get the same label?
The second factor the study fails to address in detail is time. The video below was posted on January 8, 2010, and has almost 40 million views on Youtube.
Theoretically, the algorithm could be applied to the video when it was posted in order to measure its potential to reach viral status (which it did). However, take a look at the graphs below taken from the “stats” section of the video information.
By looking at the cumulative and daily viewing totals, we can see the video didn’t reach viral status until 6 months after it was posted. How would the algorithm be applied in this situation? This study measures viral content based on potential at the beginning, so it may not be applicable to videos such as this one. Regardless, if this formula is perfected it could unleash an entirely new wave of advertising and disseminating messages online, specifically on social media. If organizations are able to predict what content will go viral, they could include features that would increase viral potential.