Algorithmic judging: lessons of logic and jurisprudence

This article considers the nature and use of judging algorithms. Based on the nature of formal systems, legal indeterminacy and the nature of law as a socially normative force, we are sceptical about judging algorithms. We first consider the nature of algorithms and formal systems, noting that formal systems of more than trivial complexity are necessarily incomplete. Further, Turing’s halting problem and the P = NP problem provide evidence that legal algorithms are unlikely to be able to resolve some legal problems in a reasonable amount of time, if ever. This claim of legal indeterminacy is not a surprise to a student of jurisprudence, as it is a fundamental tenet of American Legal Realism. The Realists argued, although less formally than mathematicians, that legal systems contain legal propositions that are formally indeterminate. This Realist argument adds strength to our scepticism, by providing an ‘existence proof’ (or – at minimum – strong inductive evidence for the existence) of legally indeterminate propositions. Finally, we consider the social context of law and judging. Algorithmic judging fails to capture the social normativity needed to legitimately resolve common disputes. ‘Tell it to the judge’ might enhance the legitimacy of the judgment rendered; but ‘telling it to the algorithm’ may not.