TikTok exhibits attention-capture damaging patterns (ACDPs) that research suggests may tap into users’ psychological vulnerabilities, prolonging engagement and contributing to users’ lack of agency over their usage. Although various approaches to mitigate the effects of such patterns have been proposed, empirical evidence of their impact remains scarce. We conducted a quantitative study of targeted interventions on ACDPs in TikTok, combining behavioral, eye-tracking, user experience, and agency measures (N=48). We added friction to three existing ACDPs: autoplay, infinite scroll, and social investments. Our findings revealed complex trade-offs: while disabling autoplay and increasing scroll friction reduced compulsive engagement patterns, these fundamental disruptions to the established interface architecture decreased user experience with limited or negative impact on agency. Obstructing social investments had negligible effects. Our work validates the argument on negative impacts of ACDPs and demonstrates how to measure them holistically. Future studies can build on these findings to advance understanding of damaging interfaces.
ACM CHI Conference on Human Factors in Computing Systems