Manipulation of elections through information cascades in social media
A year after the results of the last election, and still everywhere one can find articles, soundbites and opinions on the outcome of the election. One of the reasons for this is that the result of the election was unexpected. People are still shocked or in disbelief, and many are actively searching for reasons why every pundit was wrong and why no one could accurately predict the outcome.
A particular theory that seems to have gained some attention is of a foreign country interfering with the election. This is purported to have happened through use of online social media platforms. Specifically, exponents of this theory claim that various countries or groups influenced the election by buying ads on social media websites that would subconsciously act on the mind of the reader and influence them to vote in a certain way. More information is coming out about this theory, and the Senate even hosted a hearing to question the companies in charge if they know anything about the meddling and what they are doing to prevent this kind of behavior in the future. The three companies who were represented at the hearing were Facebook, Google, and Twitter, all giants in the tech world, and companies any layman would recognize by name.
The companies were asked how they would remove hate speech without impeding on users’ rights of freedom of speech. A representative of Facebook, Colin Stretch responded that his company would use algorithms to block hate speech and stop bots. However, Senator Mark Warner responded by asking if these algorithms were actually part of the problem?
Colin Stretch said that with respect to the Algorithm, their goal is “to provide the most relevant information. It’s primarily driven by friends and families.”
However, deeper analysis of these algorithms shows that they might not be as helpful as the companies make them out to be. The algorithms work through information cascades, which is the event when a person or people make a decision based on the decision of others before them. By the nature of a cascade, valid information can often be overlooked in favor of following a crowd. For this reason, cascades can often and easily make the “wrong” choice or pounce on little information instead of considering new developments that may affect a choice or outcome. Cascades are determined by the structure of the network that they are in, so in the context of social media, if many people that you are connected to share an opinion or buy a product, you would be more likely to do the same.
It has been shown through various Psychological studies that people, even from a very young age, prefer to associate themselves with people similar to them (Tajfel and Turner’s 1974 “Social Identity Theory” , 2012 “Not Like Me = Bad Infants Prefer Those Who Harm Dissimilar Others”). On social media platforms, this idea manifests itself by people being connected to people with similar views as them, and social media companies have taken advantage of these concepts to ads to people that relate to their specific interests and the interests of their friends and connections. The cascade effect comes into play here when new information, which is intentionally agreeable by specific algorithms of the companies, is shared rapidly to friends with similar views on the network. The sites have been accused of creating echo chambers since only agreeable content is shown and spread, and opposing viewpoints rarely manifest themselves to a user, out of fear the user might click out of the app causing the company to lose money in the process.
This is a problem because people are confined to their own viewpoints on social media platforms, and while those users are happy seeing what they enjoy, and the social media companies are happy making money off ads to those users, the valuable sharing of different ideas is lost, so people unexposed to each other’s opposing ideas become more divided and less receptive to new ways of thinking, something of immense importance in a free, democratic society.