Political Bias and Factualness in News Sharing across more than 100,000 Online Communities

Studying news shared online is challenging due to the massive volume of content which is shared. In this work, we conduct the largest study of news sharing to date, analyzing more than 550 million links to news articles posted on reddit over four years. Using non-partisan news source ratings of sources' political bias and factualness, we find that up- and down-voting mechanisms generally decrease the visibility of biased and low factual content, and that such content is very concentrated, being mostly shared in only a small fraction of all communities.