The Myth of Algorithmic Regulation: An ethnographic exploration of algorithms, actors, and institutions

Abstract

A growing number of public organizations adopt algorithms in the hope that organizational practices will become more effective and efficient. Especially for regulatory agencies, algorithms are believed to have a high potential to further rationalize practices. Such practices are known as algorithmic regulation. Yet, realizing the potentials of algorithmic regulation is not easy and often does not result in desired outcomes. In my dissertation, I have explored the potentials of algorithmic regulation and how regulatory agencies attempt to realize them mainly through ethnographic fieldwork at two regulatory agencies in the Netherlands. To make sense of the results, this dissertation offers a new lens on the adoption of algorithms: the myth of algorithmic regulation. It enables us to better differentiate between algorithmic myth and organizational reality and to better understand the process through which the myth is translated into reality and the actors involved. The research shows that the process of adopting algorithms involves three specific organizational-institutional patterns: decoupling, learning, and integrating. First, decoupling helps to understand why the adoption of algorithms does not always lead to changes in regulatory practices. When actors who work on the adoption of algorithms do not overcome conflicting understandings of meanings, norms, and power relations, algorithms may become decoupled from everyday practice. Second, learning describes two ways of realizing change in organizations that adopt algorithms: single-loop learning when data science is being connected to other forms of expertise and double-loop learning when institutional mechanisms are being established for algorithmization. Third, with the observed pattern of integration I refer to how public sector data scientists react to varying and potentially conflicting institutional logics. The research shows that data scientists integrate a technological logic with domain and political-administrative logics in their work practices. Thus, the work of the data scientists is hybrid, but hybridity varies across organizations and over time. These observed patterns reveal some key mechanisms of institutional transformation in the context of the adoption of algorithms and help to answer the overarching research question of this dissertation: How do actors and institutions shape and are shaped by the adoption of algorithms in public organizations? Actors and institutions are being transformed because of the interactions between actors and their norms, understandings, and power relations on the one hand and algorithms and the powerful qualities attributed to them on the other hand. The myth of algorithmic regulation seems to be a crucial legitimizer of institutional change in contemporary public organizations. For public organizations in general and regulatory agencies in particular, this dissertation demonstrates that using algorithms must not be seen purely as an implementation issue that can be addressed through technological and organizational changes. Adopting algorithms involves making critical value choices beyond effectiveness and efficiency. Safeguarding values, such as transparency, accountability, and non-discrimination, needs to be institutionally enabled. Thus, to benefit from the myth of algorithmic regulation, creating these institutional conditions will be a key challenge for public organizations in the years to come.