Algorithmic selection and supply of political news on Facebook
研究了Facebook在2013和2014年更新新闻推送算法后，德国报纸在平台上发布实质性政治新闻的数量和多样性变化，发现数量增加约30%，且政治观点集中度下降。
Facebook has been criticized for exposing its users to low-quality and harmful information, including fake news, hate speech, and politically one-sided content. In December 2013 and again in August 2014, the platform updated its news feed algorithm to increase user exposure to quality content of news publishers, while curbing the proliferation of non-informative posts. This paper uses a sample of German newspapers to investigate the conjecture that these modifications raised the incentives to publish quality news stories on the platform, focusing on the number and diversity of news story posts about substantive political issues. Using the newspapers’ print editions as a counterfactual, our results indicate an increase in the amount of substantive political news on Facebook by approximately 30%. This expansion occurred in a politically balanced way, except that the outlets disproportionately increased their Facebook coverage of the formerly underrepresented Linke (Left Party). Consequently, the within-outlet concentration of political viewpoints decreased by about one half of the standard deviation of our concentration indices.