Against Algorithmic Clarity: Law Beyond Specification

Abstract

As algorithmic tools increasingly assist in legal decision-making, they impose structural demands that often conflict with the normative character of law. Legal concepts must be translated into computational form to function within algorithmic systems, and such translation requires increased conceptual specificity. This paper argues that the specificity demanded by these systems undermines a core virtue of legal reasoning: its principled vagueness. We identify two primary concerns with algorithmic specification. First, the authority problem, that algorithmic categorization is achieved through extra-legal processes that lack interpretive authority. Second, the openness problem, that the elimination of vagueness undermines the ‘open-texture’ of the law, a crucial feature that allows it to be responsive to moral and social development. In response, this paper defends legal vagueness not as a flaw to be corrected, but as a jurisprudential strength. Drawing from Gadamerian hermeneutics, we contend that legal meaning emerges through dialogical interpretation, rather than static definition. Vagueness is what allows legal language to remain responsive, contested, and morally alert. Rather than optimizing algorithmic systems for clarity and determinacy, we propose design principles that support productive vagueness. These include underspecification, interpretive pluralism, and mechanisms for contestability. In the criminal law context in particular, where concepts like ‘culpability’ and ‘risk’ resist compression into fixed categories, computational tools must be carefully constrained. The goal is not to eliminate vagueness but to preserve its function as a vehicle for collective interpretive engagement.