X's hidden likes policy fails to boost engagement with risky content

In June 2024, X (formerly Twitter) made likes private, removing public visibility of who liked a post. Researchers from multiple institutions—Yuwei Chuai, Manoel Horta Ribeiro, Gabriele Lenzini, and Nicolas Pröllochs—seized this rare platform-level experiment to test whether hiding liker identity would embolden users to engage with content that carries social or reputational risk, such as political or partisan posts.

The team conducted two complementary studies. First, a Difference-in-Differences analysis of 153,704 posts from 1,045 accounts that accumulated over 324 million likes before and after the policy change. Second, a within-subject survey experiment with 203 X users measuring self-reported willingness to like various content types under public vs. private visibility. Results showed no detectable increase in likes for high-reputational-risk content at the platform level. Survey participants (especially those with higher education and income) reported modest increases in willingness to like such content when likes were private, but these intentions did not translate into significant group-level behavioral changes. The study suggests a gap between user intention and actual behavior limits the real-world impact of hiding like identities.