HCL+PrioR mtg. Kylan Rutherford. Monday 11/24 1:30-2:30pm ET

[Antonio Alonso Arechar]

Good morning everyone,

Kylan Rutherford will present next week (11/24 1:30-2:30pm ET; see blurb below).

Lab meeting will be hybrid, with a zoom link (https://mit.zoom.us/j/91655448125) and in person in the conference room on the 15th floor of E94 (1579; 245 First Street).

Best regards,

Antonio

---

In this preregistered field experiment, we modified content-ranking algorithms on Facebook, Reddit, and X (formerly Twitter) to prioritize prosocial content identified by five different algorithms and examined the effects on political polarization. Three of these algorithms were selected through a peer-reviewed competition with monetary rewards for the finalists. Each algorithm modified existing platform feeds by boosting and/or adding different kinds of constructive, informative, or cross-cutting content. Using a browser extension installed by over 5,000 participants, we randomly assigned users to either a prosocial ranking algorithm or a control condition (unmodified feeds). Averaged across all five algorithms, we find that prosocial ranking reduces affective polarization among users (d=-0.04, p < 0.05). We find no effects on secondary outcomes of well-being, news knowledge, perceptions of and support for partisan violence, outgroup empathy, and perceived value of the platform. We saw mixed results on platform usage. All interventions appear to decrease time spent on Facebook, but increase time spent on Twitter. We find mixed results with Reddit. Our findings underscore the social implications of ranking algorithms -- showing that promoting civility and cross-cutting content can improve societal outcomes without necessarily conflicting with the engagement-driven business model of social media.

[Antonio Alonso Arechar]

Good afternoon everyone,

Sorry, there will be no lab meeting today. We will resume next week.

Best,

A