Not taking one’s hat off in certain circumstances might be sanctioned by public disapproval, signalling the violation of a moral obligation. Such a breach of diffused expectations, however, does not amount to a violation of positive law. By contrast, an action which might be indifferent or tolerable from a moral point of view, i.e. not to stop at red lights in a desert, constitutes a violation of positive law where that is provided by a legal norm expressed into the sources of law and the effects produced by such a violation might be enforced by public authorities invested of the power to exercise public force.
It may be contrary to positive law to enter a vehicle into a park. A data-driven machine trained on thousands of pictures of vehicles might accurately classify new pictures as either representing vehicles or non-vehicles.1 Whatever the performance that the machine achieves in such task, however, the machine output does not automatically produce performative legal effects,2 in that the machine processing does not amount to positive law. To count as a legal decision, the outcome output by the machine must satisfy the set of normative requirements provided for by positive law. If the decision is disputed, the fulfilment of such requirements must be proven through the articulation of a justification which, in turn, complies with the standards of legal reasoning which distinguish legal practice.
P. Gowder, ‘Is Legal Cognition Computational? (When Will DeepVehicle Replace Judge Hercules?)’ in R. Whalen (ed), Computational Legal Studies. The Promise and Challenge of Data-Driven Research (Edward Elgar Publishing 2020), pp. 215-237; M.A. Livermore, ‘Rule by Rules’ in R. Whalen (ed), Computational Legal Studies. The Promise and Challenge of Data-driven Research (Edward Elgar Publishing 2020), pp. 238-264. ↩
M. Hildebrandt, ‘Law as Computation in the Era of Artificial Legal Intelligence: Speaking Law to the Power of Statistics’ (2018) 68 University of Toronto Law Journal 12. ↩