The Echo Chamber Effect in Twitter: does community polarization increase?

Abstract

A recent article criticized social media platforms for failing to mobilize society into action long enough to address any major global issue. This is attributed to the simplistic design of current social media platforms, which encourage ideas to spread virally but do not support consensus formation which might lead to lasting social change. One reason for this could be the well known echo chamber phenomenon, whereby people tend to discuss issues only with other like-minded people. Social media has been blamed for encouraging the echo chamber effect and increasing polarization in society. For example, in Twitter, it is very common for users to be followed by others with similar views. Is this a reflection of real life or does Twitter actually increase polarization of views? This paper investigates this by comparing the Twitter follows network at two points in time and detecting communities in the network of reciprocated follows relationships. We find that new edges are (at least 3-4 times) more likely to be created inside existing communities than between communities, and existing edges are more likely to be removed if they are between communities. This leads to the conclusion that Twitter communities are indeed becoming more polarized as time passes.