Anonymity and Online Commenting: The Broken Windows Effect and the End of Drive-by Commenting

In this study we ask how regulations about commenter identity affect the quantity and quality of discussion on commenting fora. In December 2013, the Huffington Post changed the rules for its comment forums to require participants to authenticate their accounts through Facebook. This enabled a large-scale 'before and after' analysis. We collected over 42m comments on 55,000 HuffPo articles published in the period January 2013 to June 2014 and analysed them to determine how changes in identity disclosure impacted on discussions in the publication's comment pages. We first report our main results on the quantity of online commenting, where we find both a reduction and a shift in its distribution from politicised to blander topics. We then discuss the quality of discussion. Here we focus on the subset of 18.9m commenters who were active both before and after the change, in order to disentangle the effects of the worst offenders withdrawing and the remaining commenters modifying their tone. We find a 'broken windows' effect, whereby comment quality improves even when we exclude interaction with trolls and spammers.