Research on meme tracking online, published recently as part of the ACM Knowledge Discovery conference, may interest any marketers who think about using the web to spread a message. Online press and social media outlets provide a technological base for mass diffusion of memes, but setting out to spread one’s message is no easy task, as anyone who’s tried to do so knows. With the complexity of the web of interactions that must take place to spread a message online, how much can we really know about the cycle by which a message is spread?

According to computer science researchers Leskovec, Backstrom, and Kleinberg, there is some truth to our conception of the ‘cycle’ that occurs from the time a news story is first published online, by a site like the New York Times, and the mass conversation on blogs and other social media outlets that often follows. To find this out, they identified short distinctive phrases that remain relatively in-tact as they travel online. For example, ‘put lipstick on a pig’ was popular during the presidential election, as shown in the visualization below:

news-cycle-image2

These distinctive phrases can be thought of as the ‘genetic signatures’ for the various messages. Naturally, they ‘mutate’ over time, so algorithms were devised to capture all the minor variations that make up a single message. Their analysis includes a global and a local level.

At a global level, memes related to the last presidential election were found to compete with one another over short time periods, causing some variation in the trends observed. Also, patterns were found to arise naturally, as news sources imitate each others’ decisions on what to cover, while still seeking the more recent content over the older content.

At a local level, they found that at its peak, a meme tends to ’spike’ with an explosive amount of activity. The peak of news media attention of a phrase typically comes 2.5 hours before the peak in the blogging communities. They also saw a ‘heartbeat’ type shape that happens when a meme bounces between mainstream media and blogs. Together the findings from the study suggest that there really is a ‘news cycle’ that occurs, one that can actually be measured and analyzed.

Check out the memetracker visualization for yourself!