In the folds of algorithmic personalization : an ethnography of a recommender system in public service media
(2026)
Files
In the Folds of Algorithmic Personalization.pdf
Restricted Access - Adobe PDF
- 7.29 MB
Details
- Authors
- Supervisors
- Abstract
- (en) In contemporary media environments, algorithmic recommendations have become ordinary infrastructures through which attention is organized and visibility is distributed. But what happens when a public service media organization—historically shaped by democratic, cultural, and informational missions—sets out to build and implement its own news recommender system? This dissertation addresses that question through an ethnographic study conducted at the Radio-Télévision belge de la Communauté française (RTBF), an organization that explicitly articulated its ambition to develop "public service algorithms." Based on thirteen months of participant observation, six retrospective interviews, and approximately 1,300 pages of internal and external documentation, the study chronicles the biography of the recommender system. It follows its development and deployment as well as the episodes of breakdown, repair, and reconfiguration that form its everyday existence. Instead of treating the recommender system as an autonomous tool inserted into an organizational structure, this study approaches it as a sociomaterial assemblage composed of values, infrastructures, data flows, professional jurisdictions, regulatory constraints, and organizational routines. To make sense of these processes, the dissertation uses the notion of the fold as an analytical lens. First, it shows how public service logic is folded into the algorithmic assemblage through processes of translation and selective materialization while also being partly redefined through that process. Then, it examines how breakdowns, malfunctions, and repairs reveal the fragility of the assemblage and give rise to adjustments that refold the algorithmic assemblage into a new configuration. Finally, the dissertation analyzes how the stabilized system acts back upon the organization through performative torsion, which is a partial, uneven, and negotiated reconfiguration of practices, roles, temporalities, and regimes of knowing. The dissertation makes three contributions. First, it advances the study of the algorithmization of organizations by conceptualizing recommender systems as sociomaterial assemblages shaped by successive folding operations. Methodologically, it advances the ethnographic study of algorithmic systems by highlighting the movement and transformation of assemblages. Finally, empirically, it offers a situated analysis of algorithmic personalization in public service media through a rarely examined case in the literature. More broadly, the dissertation argues that algorithmic recommendation in a public service context is a fragile organizational accomplishment, continually produced, repaired, and justified, rather than the mere importation of platform logic or the continuation of earlier institutional ideals.
- Affiliations
Citations
Carillon, K. (2026). In the folds of algorithmic personalization : an ethnography of a recommender system in public service media. https://hdl.handle.net/2078.5/279794