Content selection algorithms of social media platforms have a reputation for creating filter bubbles, where users are primarily exposed to politically like-minded information. In a new paper, Ferenc Szucs and I take a supply-side perspective: We investigate the role of Facebook’s news feed algorithm for content created by media companies. Specifically, we evaluate the effects of two major changes in the algorithm on the political bias of stories posted by German news outlets. In December 2013 and August 2014, Facebook publicly announced measures designed to boost the dissemination of quality news stories. Using a differences-in-differences strategy, our results indicate greater levels of within-outlet ideological diversity of news posted on Facebook after the second algorithm update, relative to the print coverage. The effect can be explained largely explained by a shift of the outlets towards more news and less trivia, which reduced deviations from the distribution of ideologically relevant expressions used in their print editions.