Social media newsfeed algorithms can directly affect how much social feedback a given post receives by determining how many other users are exposed to that post. Because we show here that social feedback affects users’ outrage expressions over time, this suggests that newsfeed algorithms can influence users’ moral behaviors by exploiting their natural tendencies for reinforcement learning…. [D]esign choices aimed at … profit maximization via user engagement can indirectly affect moral behavior because outrage-provoking content draws high engagement….28 1 2 3 4 5 6 52. In other words, if a user makes two posts—one containing hateful, outraged, and 7 divisive content and one lacking such content—Facebook’s algorithms will show the hateful, 8 outraged, and divisive post to more users. Consequently, the hateful, outraged, and divisive post 9 is rewarded with more likes, shares, and comments. The user quickly learns that to obtain a 10 reaction to his or her posts, he or she should incorporate as much hateful, outraged, and divisive 11 content as possible. 12 53. 13 On October 5, 2021, Frances Haugen, a former Facebook product manager, testified before Congress: 14 The dangers of engagement based ranking are that Facebook knows that content that elicits an extreme reaction from you is more likely to get a click, a comment or reshare. And it’s interesting because those clicks and comments and reshares aren’t even necessarily for your benefit, it’s because they know that other people will produce more content if they get the likes and comments and reshares. They prioritize content in your feed so that you will give little hits of dopamine to your friends, so they will create more content. And they have run experiments on people, producer side experiments, where they have confirmed this. 29 15 16 17 18 19 20 54. Recently leaked documents confirm Facebook’s ability to determine the type of 21 content users post through its algorithms. After Facebook modified its algorithms in 2018 to 22 boost engagement, “[t]he most divisive content that publishers produced was going viral on the 23 24 25 26 27 28 28 William J. Brady, Killian McLoughlin, Tuan N. Doan, Molly J. Crockett, How social learning amplifies moral outrage expression in online social networks, 7 SCIENCE ADVANCES, no. 33 (Aug. 13, 2021), https://www.science.org/doi/10.1126/sciadv.abe5641. Posts were classified as containing moral outrage or not using machine learning. 29 Facebook Whistleblower Frances Haugen Testifies on Children & Social Media Use: Full Senate Hearing Transcript, REV (Oct. 5, 2021), https://www.rev.com/blog/transcripts/facebook-whistleblower-frances-haugen-testifies-onchildren-social-media-use-full-senate-hearing-transcript. CLASS ACTION COMPLAINT 19 Case No. __________________