Misinformation on Twitter
It is clear that popular social media sites like Twitter and Facebook have a profound impact on the behavior and beliefs of its users. These sites act as news sources for many people and, as a result, both true and false news can be proliferated throughout these large social networks; this particular phenomenon is known as an information cascade.
In the article Viral news–true and untrue–moves equally through Twitter from the Cornell Chronicle, the author explains that cascades are the “measurement of paths viral tweets take from the original poster down through the network via retweets” and how these cascades can occur on Twitter.
As learned in class, there is one initial node in the network that engages in a certain behavior and, depending on the thresholds of its neighbors, can influence others to engage in this behavior. This initial behavior has the potential to spread throughout the entire social network; but, the larger this social network is, the more likely there are to be dense clusters in the network that can prevent this behavior from being a complete cascade.
In the context of Twitter, a node can represent a Twitter user and edges connecting these nodes mean that they follow each other. If one initial node tweets a piece of misinformation, this tweet can be retweeted by one of his neighbors/followers; this neighbor’s neighbor could also retweet this piece of misinformation, and this is how the tweet gets spread throughout the social network. Very rarely will a complete cascade occur in a social network of this magnitude due to the highly clustered structure of the network, where one cluster has the potential to prevent the tweet from going completely viral if they choose not to retweet it (i.e. their beliefs do not coincide with the content of the tweet). An extension of the information cascade studied in class would be each individual node having their own thresholds. A threshold, in this case, would be how many of my followers would have to retweet a piece of information for me to choose to retweet it myself. Certain Twitter users may have higher thresholds, where it takes more of their followers to retweet something before they themselves feel compelled to retweet it, whereas some users can have lower thresholds (i.e. it takes less of their followers for them to feel compelled to retweet something).
Containment of Misinformation Spread in Online Networks also focuses specifically on the proliferation of misinformation on Facebook and Twitter and seeks to minimize this. What the authors sought to do to counteract this spread was follow the above logic and develop a method to find the smallest disjoint set S of highly influential nodes in the social network. In this way, this set of highly influential nodes have the power to counteract the damaging effects of the misinformation already spread by correcting the misinformation in a tweet.
In focusing on highly influential nodes, these nodes (like celebrities, for example) have many followers whom they influence; these followers also have followers, who can see what they retweet. Because this particular node is so influential and the nodes that follow it are easily influenced by this node, the tweet will cascade throughout the network from their followers, to their followers’ followers, and so on. Though it is still unlikely that a complete cascade will occur due to the highly clustered structure of Twitter’s social network, focusing on having these highly influential nodes in set S be the initial nodes to correct a widely spread falsehood has a higher possibility of counteracting the damaging effects that come as a result of spreading misinformation.
Sources:
https://news.cornell.edu/stories/2021/11/viral-news-true-and-untrue-moves-equally-through-twitter
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.722.2176&rep=rep1&type=pdf