Close

Paper presentation: Predicting children at risk: The future as resource in social work decisions

Helene Ratner and Kasper Trolle Elmholt presented a paper in the stream on ‘The algorithmic turn of prediction‘ at this year’s Society for Social Studies…

Helene Ratner and Kasper Trolle Elmholt presented a paper in the stream on ‘The algorithmic turn of prediction‘ at this year’s Society for Social Studies if Science (4S) conference.

Here’s the abstract: Public sector use of predictive algorithms has paved way for hopes of better precision and a more timely intervention in casework. Specifically, they have opened up for new modes of relating to the future through preemptive and preventive interventions. How the future is mobilized in the present and what kind of decision-making practice it compels, however, differs with respect to which administrative practice that is to make use of predictive algorithmic technologies. In this paper, we explore how predictive algorithms in child protection work give rise to different enactments of the ‘timely intervention’, with different modes of evoking the future in the present. Conversely, these different temporalities rely on different data infrastructures. Theoretically, this implies theorizing algorithmic temporal orientations not only as a matter of discursive articulations but as a socio-material phenomenon, entwined with digital data infrastructures. Therefore, we theorize algorithms as wider socio-material infrastructural arrangements rather than abstract, computational formulas. Empirically, we compare two Danish developments concerning algorithmic prediction of children at risk. While they are similar in terms of their aim and target group, predicting children at risk of maltreatment, they differ in terms of their temporal orientation and infrastructural arrangements. Whereas one administrative practice seeks to use predictive algorithms to support already existing practices of risk assessment, using data from only the administrative sector in question, the other case endeavored to predict children’s risk of maltreatment from the moment of their birth, requiring data from other administrative systems.  With this analysis, we bring attention to two different modes of conjuring the future in the present: Whereas the first case uses prediction to tame the uncertainty in the present, supporting existing decision-making practices, the second case illustrates a rather different invigoration of the future. Namely, to eliminate undesired futures by preempting problems before they emerge or grow severe.