Social media algorithms can curb misinformation, but do they?

TL;DR

It is demonstrated that a series of temporary emergency changes to Facebook's news feed algorithm in the wake of the 2020 U.S. presidential election that were designed to diminish the spread of voter-fraud misinformation systematically reduced the amount of misinformation in the control group of the study, which was using the news feed algorithm.

Abstract

A recent article in $\textit{Science}$ by Guess et al. estimated the effect of Facebook's news feed algorithm on exposure to misinformation and political information among Facebook users. However, its reporting and conclusions did not account for a series of temporary emergency changes to Facebook's news feed algorithm in the wake of the 2020 U.S. presidential election that were designed to diminish the spread of voter-fraud misinformation. Here, we demonstrate that these emergency measures systematically reduced the amount of misinformation in the control group of the study, which was using the news feed algorithm. This issue may have led readers to misinterpret the results of the study and to conclude that the Facebook news feed algorithm used outside of the study period mitigates political misinformation as compared to reverse chronological feed.

Figures (1)

• Figure 1: (a) Average weekly number of views of news from trustworthy and untrustworthy sources, calculated using the Facebook URLs dataset DVN/TDOAPG_2020. Our estimates of untrustworthy news are based on links to sources rated mixed, low, or very low for factual reporting by Media Bias/Fact Check (MBFC) and shared at least 100 times, whereas Guess et al. consider any post by users with two or more reports as untrustworthy. (b) Fraction of views of untrustworthy news among all views. The horizontal dotted lines are averages of the points of the same color. We observe a drop during a period overlapping with the experiment, likely due to the changes in the news feed algorithm.