GOVERNING MENTAL HEALTH CARE: HOW POWER IS EXERTED IN AND THROUGH A QUALITY IMPROVEMENT COLLABORATIVE

We investigated the role of power in public governance using a Foucauldian conceptualization of power, i.e., power is produced by a range of techniques as diverse as language and measuring. We draw on an evaluation study of a quality improvement collaborative, in which different mental health care organizations were encouraged to improve their care in a structured way. We analyzed how the different actors involved in the collaborative were governed and came to govern themselves differently. Measurement instruments were an example of a dominant mechanism by which actors at different levels of the collaborative were governed: by accounting for improvements, introducing or strengthening a certain way of thinking about health care clients, and changing how clients thought about and acted upon themselves. We argue that the focus on consequences of governing techniques is fruitful for studying governmentality and leads to new research questions in the context of public policy analyses.     

Human–algorithm hybrids as (quasi-)organizations? On the accountability of digital collective actors

Models of individual accountability for algorithms’ actions fail when a human–algorithm association comes to be viewed as a collective actor. In some situations, human and algorithmic actions are so closely intertwined that there is no longer a linear connection between the emergent collectivity and the complex interactions of humans and algorithms. In such collective decision-making sequences, individual accountability can no longer be attributed. Therefore, a new perspective on human–algorithm associations that captures their emergent properties and organizational qualities is needed to develop appropriate models of collective accountability. This article seeks to answer a number of questions. How can the encounter between humans and algorithms within such a socio-technical configuration be adequately theorized? Can the configuration itself be understood as a hybrid collectivity? Can actions be attributed to the configuration as a personified collective actor? How will accountability be institutionalized for human–algorithm associations – in centralized or distributed collective forms?