Institutionalizing Folk Theories of Algorithms: How Multi-Channel Networks (MCNs) Govern Algorithmic Labor in Chinese Live-Streaming Industry

TL;DR

This paper investigates how intermediary organizations, specifically Chinese Multi-Channel Networks (MCNs), institutionally construct and deploy folk theories of algorithms to govern live-streamer labor amid opaque platform logics. Through nine months of ethnography in Beijing and Changsha plus 37 interviews, it identifies dual, internally probabilistic and externally prescriptive folk theories that MCNs use to manage risk and motivate streamers. The external narratives translate opacity into actionable guidance linked to observable metrics, equipment, and discipline, while internal theories acknowledge instability and drive batch recruitment as a risk-spreading strategy. The study contributes to CSCW by reframing folk algorithmic knowledge as infrastructural and governance-oriented, with design and policy implications aimed at enhancing epistemic transparency and accountability in algorithmic labor ecosystems.

Abstract

As algorithmic systems increasingly structure platform labor, workers often rely on informal "folk theories", experience-based beliefs about how algorithms work, to navigate opaque and unstable algorithmic environments. Prior research has largely treated these theories as bottom-up, peer-driven strategies for coping with algorithmic opacity and uncertainty. In this study, we shift analytical attention to intermediary organizations and examine how folk theories of algorithms can be institutionally constructed and operationalized by those organizations as tools of labor management. Drawing on nine months of ethnographic fieldwork and 37 interviews with live-streamers and staff at Multi-Channel Networks (MCNs) in China, we show that MCNs develop and circulate dual algorithmic theories: internally, they acknowledge the volatility of platform systems and adopt probabilistic strategies to manage risk; externally, they promote simplified, prescriptive theories portraying the algorithm as transparent, fair, and responsive to individual effort. They have further operationalize those folk theories for labor management, encouraging streamers to self-discipline and invest in equipment, training, and routines, while absolving MCNs of accountability. We contribute to CSCW and platform labor literature by demonstrating how informal algorithmic knowledge, once institutionalized, can become infrastructures of soft control -- shaping not only how workers interpret platform algorithms, but also how their labor is structured, moralized and governed.

Figures (2)

• Figure 1: Tiered technical audio solutions deployed by MCN agencies serve as material markers of professional status and algorithmic worth within the live-streaming economy. (a) Entry-level handheld microphone provided as standard equipment for contracted streamers, a counterfeit version of the Blue E300, with production costs ranging from 60-70 RMB (approximately 8 USD). (b) Mid-tier condenser microphone offered complimentary with streaming studios, including free replacement service for malfunctions; this unit replicates the Neumann TLM 193 with manufacturing costs between 300-400 RMB (approximately 50 USD). (c) Shure SM7B microphone recommended by MCN agencies for talk show hosts, voice actors, and vocal performers with established audience bases; streamers are encouraged to purchase this equipment through the MCN at a one-time fee of 4,000 RMB (market value approximately 2,500 RMB), inclusive of complementary mixing console and technical support services. (d) Flagship audio broadcasting system typically reserved for contracted knowledge-based/educational teams or professional singers; streamers may temporarily rent this equipment for cover or original song recordings at approximately 300 RMB (40 USD) per hour, with post-production and technical assistance available at additional cost. This premium equipment primarily serves as a demonstration of the MCN's professional capabilities and is not available for external purchase.
• Figure 2: Comparative analysis of livestreaming studios with varying amenities and price points. All studios provide basic air conditioning, internet connectivity, and workspace furnishings, while differing substantially in equipment quality and decorative elements. (a) A bare basement dance streaming studio featuring a faux window facing a concrete wall, priced at approximately 2,500 RMB monthly without computer equipment, lighting systems, or professional backdrops beyond a rudimentary fringe curtain. (b) A compact basement studio designed for singing and conversational streaming, characterized by its confined windowless space, basic interior finishing, and pink swivel chair, available at approximately 2,200 RMB monthly. (c) A ground-floor multipurpose streaming studio equipped with fundamental lighting apparatus and computer systems, featuring a dedicated window, priced at approximately 3,100 RMB monthly. (d) A premium spacious studio situated on the second floor, furnished with an independent clothing rack, basic lighting equipment, professional microphone with stand, in-room audio monitoring speakers, dedicated window access, and a vanity mirror, available at 3,500 RMB monthly with the additional advantage of proximity to technical support staff offices for immediate assistance.