Algorithmic Gatekeeping 2.0: When News Recommenders "Soft-Censor" Political Diversity

Einarsson et al. (2024) found recommenders reduced topic diversity but didn’t diagnose why. This research audits recommender algorithms (e.g., via API testing) to detect whether "soft news" (e.g., lifestyle, entertainment) is prioritized over contentious politics—a form of soft censorship. Unlike their focus on exposure effects, we’d examine algorithmic designs (e.g., engagement metrics) that de-prioritize divisive issues. Comparative analysis across countries (e.g., Denmark vs. Ghana; Darkwa et al., 2023) could reveal how business models shape agendas. The contribution is linking algorithmic bias to agenda-setting theory: if algorithms suppress political issues, they indirectly control public discourse—extending McCombs & Shaw (1972) to black-box systems.

References:

1. Algorithmic agenda-setting: the subtle effects of news recommender systems on political agendas in the Danish 2022 general election. Á. Einarsson, R. Helles, S. Lomborg (2024). Information, Communication & Society.
2. Navigating the Mass Media’s Political Agenda Setting Role: A Case Study of Ghana. Ernest Darkwa, Bridget Acquah, Constance Osafo-Adjei (2023). Inverge Journal of Social Sciences.