Algorithmic Kaleidoscope: The Floating Algorithmic Self and Self-Discrepancy of Recommendation System Users

Human self-concepts are multi-dimensional, while algorithmic curation reflects an algorithmic self, a digital reflection of identities from recommendation systems. Through 23 in-depth interviews, this study examined how users interpret their algorithmic self, its composition, and psychological impacts. Findings reveal that algorithmic self is floating, evolving with engagement and life trajectories, incorporating aspects of users’ actual self (with extended and missing parts) and ideal self (guiding self-growth). While single platforms capture only fragments of identity, multiple systems co-construct users’ algorithmic self. Drawing on self-discrepancy theory as an interpretive framework, we identify psychological responses to mismatches between actual and algorithmic selves, including self-enhancement, chilling effects, and identity threats. Respondents reflected that self-esteem and self-efficacy help mitigate these negative responses, enabling self-growth. Capturing these dynamics, we propose the concept algorithmic kaleidoscope, where algorithms capture users’ floating, multifaceted self, and curate exploration and insights, meanwhile users actively manage their algorithmic self through selective reflection.