Algorithms and Regulation

Algorithms – generally understood as sequences of precise instruction unambiguously specifying how to execute a task or solve a problem – are such a natural ingredient of regulation that some may wonder whether regulation could even be understood without recognising its algorithmic features, and without realising algorithms as a prime subject for regulation. In terms of the algorithmic features of regulation, somewhat simplistically and without suggesting in any way that the algorithmic language captures regulation in its entirety – far from it – algorithms are relevant to the three dimensions of regulation: the regulatory process, the modalities of regulation, and the regulatory approaches (or attitudes). By the regulatory process, we refer to the process that, stylistically, commences with political and economic pressures to find a solution to a certain problem and continues with the formation of policy goals, data gathering, and the mapping of possible regulatory responses to achieve these goals (which ought to include the sub-processes of regulatory impact assessment upon choosing the preferred measure). The chosen measures are translated into regulatory norms and implemented (or enforced), resulting, if all went well, with some improvement of the conditions related to the initial social problem (as can be analysed by a back-end regulatory impact assessment). By regulatory modalities, we mean the set of regulatory measures available to the state (or, more accurately, to state agencies acting as regulators): regulation through (and of) information, permits and licensing, civil, administrative and criminal liability, taxes and subsidies, or insurance schemes. By regulatory approaches, or attitudes, we mean the top-down command and control attitude, performance-based regulation, and themanagerial approach, with the latter two also including co-regulation or private regulation. Algorithms are relevant to all three dimensions of regulation, as they may assist most, if not all, stages of the regulatory process, may inform or even be a component of the regulatory modalities, and may similarly inform and be integrated into the regulatory attitudes. Conversely, algorithms may be the subject matter of regulation.