Big Data, Responsive Regulation and Voluntary Compliance

In this paper, we attempt to create a taxonomy of the ethical concerns, associated with algorithmic approaches that attempt to regulate the level of trust regulators can place in individuals. Such approaches are relevant in various regulatory contexts, for example, fast track licensing of businesses, low-risk certifications (e.g. Israel’s coronavirus vaccine Green Pass regime), and decisions to diminish surveillance, detection, or enforcement measures from a certain sector. In all such contexts, regulators must employ a measure of trustworthiness on either an individual or a collective level, invoking concerns of equality and fairness. Individual-level trust is usually based on information regarding a person’s level of past compliance. Algorithmic tools have the capacity to relatively easily identify and aggregate information on an individual’s compliance levels across platforms and contexts, and to compare this information to other individuals, ostensibly generating an individual “compliance score.” Such means raise diverse concerns, namely the question whether norm violation in one context should affect trustworthiness in another: beyond the empirical question this is also a normative challenge, as barriers to compliance in certain contexts may reflect social disparities that do not warrant penalizing. In this paper we will outline the different behavioral (e.g. accuracy and sustainability of prediction) and ethical (stigmatization, lack of controllability) factors which are relevant for both types of trust and provide the blueprint for how a normative discussion of what predictions should be allowed to be taken into account in a behavioral big data research done for the purposes of regulatory and enforcement dilemmas.