Predictive algorithms are increasingly being deployed in a variety of governance settings to determine legal status: to confer welfare benefits, to allocate prison sentences, to allow immigration. Such uses of algorithmic systems raise three categories of governance issues. Considerable attention has already been paid to issues involving governance of algorithms, and to issues involving governance by algorithms. Here I take up a third, neglected inquiry, regarding the governmentality of algorithms: how such automated systems reconfigure the people and institutions with which they interact. This inquiry may take two forms: the first assessing the effects of attempting to incorporate legal standards into algorithms, and the second assessing the effects of incorporating algorithms into legal processes. In the first case, algorithmic systems promise to shape and eventually supplant the legal standards they seek to implement. In the second case, reflexive properties of algorithmic systems threaten to undermine the normative foundations of legal judgments. Understanding such effects is vital to identifying situations where the use of such mechanisms should be curtailed or abolished.

Dan Burk's lecture will take place as part of the Berlin Science Week from 1 - 10 November 2019.