Study: TikTok algorithm does not permanently react to negative feedback
New research indicates that even when negative signals are given to TikTok's algorithm, it may not permanently remove content from a user's feed.
Researchers from Northern Illinois University have investigated how effectively TikTok users can influence their For You Page (FYP) feed. Their study suggests that negative feedback provided by users, such as using the 'not interested' feature, has only a temporary effect on the algorithm.
The algorithm aims to personalize user experience heavily relying on implicit signals, like video watch time, as well as explicit signals such as likes. However, the research found that even if a user does not watch or marks a video as 'not interested,' similar content reappears unless negative feedback is consistently given.
Piotr Sapiezynski, one of the study's co-authors, told Ars Technica that his research group specializes in 'algorithm audits' to better understand how online platforms operate. Their objective was to examine user agency on TikTok, as numerous users had reported the ineffectiveness of negative feedback.
The research implies that while platforms offer tools for feedback, their actual impact on a user's content selection may be more limited than anticipated. This raises questions about user control within digital environments.