Information Cascades on Twitter related to Crisis Warnings
In a paper written by Cindy Hui at Rutgers University along with three other researchers from Rensselaer Polytechnic Institute (William Wallace, Malik Magdon-Ismail, and Mark Goldberg), they researched information cascades on Twitter related to disaster/crisis warnings, specifically an example of tweets responding to an armed robbery near the RPI campus in 2010. Their conclusions were that only a small number of original Tweets are the sources of public information, which were then retweeted and seen by a large audience (causing a cascade). I think that this study is interesting, because it reveals that these few original Tweets that are most retweeted are the sole (or one of few) sources of information about the event on social media. A big possible implication and consequence of this is the classic problem of “disinformation.” If one of these Tweets got details wrong about the crisis, then many people’s perceptions of what happened are going to be incorrect, because it is the most retweeted commentary about it; any further corrections are much less likely to be retweeted and shared on Twitter.
This relates to what we learned in class because it is a modern example of an information cascade on social media. Once a few people start sharing a Tweet relating to a crisis, this Tweet will be shared and be believed by many people. If someone issues a correction or learns further details about the crisis later on (private information), most people probably will still believe the original information because many people have shared it (and thus likely believed it to be true, so it is the dominant public information); most people likely won’t even see the corrected and new information.
Paper: http://www.cs.rpi.edu/~magdon/ps/conference/InfoCascadesSWDMwww2012.pdf