ANT-KOL: Reconfiguring KOL Influence within Actor Networks on Algorithmic Platforms

Algorithmic platforms are redefining how influence is produced and negotiated, with Key Opinion Leaders (KOLs) playing a central role in the processes. However, how they form communities, evolve collectively, and exert influence remains insufficiently understood. This study examines a leading Chinese algorithmic platform named "Toutiao" through the lens of Actor-Network Theory (ANT). We conduct a two-wave longitudinal social network analysis of KOLs (N₂₀₂₃ = 165; N₂₀₂₅ = 155), each of whom had more than one million followers. Semi-structured interviews are conducted with 10 volunteers, including KOLs and ordinary users. Results reveal a marked shift from a relatively cohesive KOL community to a sparser and more fragmented structure. Approximately 84.89% of the disappeared edges occurred between KOLs that remained in the networks. KOLs continuously adjust their engagement and content strategies in response to changing platform conditions. These adjustments are shaped by broader social and platform conditions, including revenue policies, perceived recommendation mechanisms, and professional identities. Integrating ANT theory with the KOL community practices, an analytical framework named ANT-KOL is proposed. Three processes are identified, respectively community destabilization, sociotechnical alignment, and asymmetric visibility. This study integrates ANT's relational ontology with insights from KOL studies, revealing that influence is a temporary sociotechnical outcome produced through ongoing negotiation.