A characterization of political communities on reddit

Abstract

The social news aggregator Reddit is among the most popular websites on the internet. Many online users use the platform to anonymously share and discuss (mostly US-centric) political content. In this ongoing work, we perform a comparative large-scale analysis of political subcommunities (subreddits) on Reddit using a dataset of more than 100 million posts from around 5 million users. In particular, we investigate these communities with respect to (1) the content posted, (2) their relationships to other subreddits, and (3) the distribution of attention received in these subcommunities. We find that left-leaning communities use derogatory language less often than right-leaning communities, but are more focused on news sources reflecting their own political leaning. We also observe that right-leaning communities are more interconnected with right-leaning subreddits on European politics. Finally, the attention of individual submissions (as measured by their number of up-votes or comments received) is spread more evenly in right-leaning communities.