The role of recommender systems in the formation of internet echo chambers

Abstract

On social media, users are free to engage with anybody they feel inclined to. However, echo chambers may be undesirable for internet discourse. Echo chambers are situations where users are in an environment where the same opinion gets reinforced by repeating it without due criticism or exposure to the opposite opinion. The research question this thesis addresses is how echo chambers form on a hypothetical social media website. The hypothesis is that a stronger opinion reinforces a grouping tendency with those of the same opinion. This group then reinforces the opinion which in turn reinforces the grouping tendency, keeping users stuck in an echo chamber unless external factors take these users out or dissolve these groups entirely. This paper models user behaviour on a hypothetical social media website to research which situations lead to internet echo chamber formation. This model differs from others in that it specifically models user movement and grouping behaviour. Experiments on the model included simulating social mobility and utilizing the recommender system to serve content of the opposing opinion. The results show that both social mobility and serving the opposite opinion work as measures against echo chamber formation.