Abstract: Debates around China’s social credit systems (SCS)—designed to rate citizens’ trustworthiness through algorithms that surveil and quantify everyday behaviors—exemplify global controversies over the state’s increasing adoption of data-driven systems. Often overlooked, however, is how state data is produced as part of a dynamic social process. In this talk, I illustrate the datafication process within a Chinese SCS based on a 10-month ethnographic study and 104 interviews conducted in a Northern Chinese city. Contrary to common Orwellian interpretations, my findings reveal that the ambitious goals of data-driven governance are often undermined by practices such as selective data collection and data fabrication. I introduce the concept of 'boundary embeddedness' to explain this unexpected outcome, emphasizing the unique positionality of bureaucrats engaged in datafication, situated between the state and society, and shaped by various organizational, institutional, and relational contexts. Ultimately, these practices, alongside other limitations the SCS encounters on the ground, create a distorted data-driven illusion. My talk underscores the need to critically examine the politics of data underpinning AI-driven systems to better understand their limitations and potential avenues for resistance in a data-mediated society.
