Information Cascades on Facebook
https://www.technologyreview.com/s/525821/the-curious-nature-of-sharing-cascades-on-facebook/
Facebook is one of the biggest social media platform out right now. Almost everyone I know has a facebook. One of its defining features is the way photos, videos, text, and events can be shared among countless users. This leads to viral news, videos, and photos that lead to cascades where the number of reshares count to the millions. There are teams from Stanford, Cornell, and Facebook that have found ways to forecast and predict the popularity of a post by analyzing its popularity shortly after it is published. It is very difficult to predict a posts popularity because of the myriad of factors that need to be taken into account, such as the number of people the original poster is connected to, the popularity of the poster, the content of the post, and much more.
This team analyzed 150,000 photos, which collectively were shared over 9 million times. They set a time frame of about 28 days after the photos were uploaded. They were able to construct networks of people (nodes) that reshared the viral photos. In the past, researchers have looked at how large cascades begin then attempt to use that information to spot large cascades in the future. However, this new team had a different approach. They started with a photo that was shared k amount of times then determined the likelihood that the photo will double its cascade. They then used machine learning algorithm to search for features of cascades that make them predictable. It takes features that I have mentioned before as well as the speed of how quickly the cascade forms, which is usually a certain kind of power law. The best indicator of the cascade are temporal performance of the cascade, how fast it spreads. Another important factor is the content of the photo (caption). The more information, the better, because it will be exposed to more than one facet of the content. There are limitations on their methods, but overall this algorithm will help with future research of popularity of content. This can also be looked at how fast fake news can spread throughout facebook if it is controversial and surprising enough for people to reshare. It is always important to see the credibility of the source. There are a lot of click baits that fool people into believing everything people see on their feed.
This related to our topic of information cascades which exists when people make decisions (like to share a photo) depending on other people’s decisions. And in the case of viral photos or fake news, people usually think it is believable. This will help with the growing information cascade because, even if the cascade slowly stops to grow, there will always be more people to share it because of the amount of shares the popular content has received in the past.