Will the Crowd Game the Algorithm? Using Layperson Judgments to Combat Misinformation on Social Media by Downranking Distrusted Sources
Name
3313831.3376232.pdf
Description
Published version
Size
5.79 MB
Format
Adobe PDF
Checksum (MD5)
09850c2ee9c29dcfead3c8402d1156ee
Author(s) • •
Epstein, Ziv
Pennycook, Gordon
Rand, David
Date Issued
April 21, 2020
Journal
Conference on Human Factors in Computing Systems - Proceedings
Publisher
ACM
Citation
Epstein, Ziv, Pennycook, Gordon and Rand, David. 2020. "Will the Crowd Game the Algorithm? Using Layperson Judgments to Combat Misinformation on Social Media by Downranking Distrusted Sources." Conference on Human Factors in Computing Systems - Proceedings.
Version
Final published version
Abstract
© 2020 Owner/Author. How can social media platforms fight the spread of misinformation? One possibility is to use newsfeed algorithms to downrank content from sources that users rate as untrustworthy. But will laypeople be handicapped by motivated reasoning or lack of expertise, and thus unable to identify misinformation sites? And will they "game" this crowdsourcing mechanism in order to promote content that aligns with their partisan agendas? We conducted a survey experiment in which =984 Americans indicated their trust in numerous news sites. To study the tendency of people to game the system, half of the participants were told their responses would inform social media ranking algorithms. Participants trusted mainstream sources much more than hyper-partisan or fake news sources, and their ratings were highly correlated with professional fact-checker judgments. Critically, informing participants that their responses would influence ranking algorithms did not diminish these results, despite the manipulation increasing the political polarization of trust ratings.
MIT Department
Massachusetts Institute of Technology. Media Laboratory
Sloan School of Management
Massachusetts Institute of Technology. Department of Brain and Cognitive Sciences
Terms of Use
Creative Commons Attribution 4.0 International license
Persistent DSpace Link
DOI of Published Version
https://doi.org/10.1145/3313831.3376232