Gender Bias, Feedback, and Productivity (from Marita Freimane)

• I explore how gender biased feedback affects the productivity of workers in an online labor market. Using a design change on YouTube where the platform removed public displays of how often a video has been disliked, I show that — while dislike counts were public — female content creators received significantly more negative feedback on comparable content than male content creators. This gender gap in negative feedback is eliminated after the design change. Using detailed video- and channel- level data and a fuzzy difference-in-differences identification strategy, I show that the removal of excess negative feedback significantly and persistently increased the productivity of female content creators and consumer demand for their content. Relative to men, women produce 8.4 percent more videos after the platform design change. The increase in productivity coincides with an even larger increase of 15.5 percent in demand for content produced by women. Investigating mechanisms, I show that the reduction in negative feedback is primarily driven by changes in the upper tail of the distribution of dislikes and is consistent with the platform’s objective of reducing harassment through ‘dislike attacks’. Finally, I show that there are limited spillover effects on toxicity in other feedback channels and provide evidence from a placebo-test to confirm that productivity effects are indeed driven by the reduction in dislikes.
• The inverse of this can also be done to attack those who are doing constructive work and amplify those who are being destructive.

12:00 IEEE seminar

Work on area 3 of the proposal

SBIRS

• Should mostly be demo development – good progress

• Got sucked into meetings the rest of the day