Skip to main content

The Nature of Social Media Information Cascades

There is usually a large disparity between the levels of popularity that different posts on Facebook garner. Whether a photo, text, or a viral video. The idea that there is a specific timeline of progression that accounts for a post’s popularity, is one that has lead researchers to probe deeper into the science of social media information cascades. The idea of an information cascade is that, it a set of people making choices in sequence rather than all at once. The choices that people make are formed from observing those that preceded them regardless of what novel information is presented to them. This set of choices, in turn can lead to a cascade effect where the larger something grows (Facebook post in this case), the more traction it gains.

Researchers at Stanford, Cornell, and MIT together have attempted to explain/predict this information cascade eventuality in Facebook posts. The researchers over a 28-day period analyzed over 150,000 photos which were altogether shared over 9 million times. The data indicated which people(nodes) were responsible for sharing the photos, and when they shared them. This allowed them to construct a network through which this information cascade took place.

Through this and machine learning, they were able know which traits of a particular post would indicate a possible information cascade. Some of these features include how fast it initially spreads, topics mentioned, whether the caption is associated with a picture, whether it is newsworthy, or whether it is associated with a present meme. With studies continuing and prediction techniques getting more advanced, researchers say that the understanding of cascades on social networks is one that is “more gold them thar hills”.




Leave a Reply

Blogging Calendar

November 2017
« Oct   Dec »