ABSTRACT
With a growing number of cases of algorithmic harm being reported, various stakeholders are developing strategies for changing the algorithmic regime for the better. In this contribution we offer ethnographic insights gained when participating in one such effort, a European research and training project. We investigate three dimensions of the algorithmic regime: First, we explore individual, mostly disciplinary interventions to mitigate algorithmic harm and show how interdisciplinary collaborations are activated by attempts to generalize individual accounts of fairness. Second, we demonstrate how a flexibility in the concepts of algorithmic fairness allowed successful collaboration within the project. Finally, we examine attempts to move beyond narrow, disciplinary requirements, and investigate how the algorithmic regime affects such interventions.
