Historical Trends in Macro-Jurisprudence: A Language Model Assessment, 1870-2023

Edward H. Stiglitz & Rosamond Thalken

Today, observers agree we live in a period of legal formalism, in which for example textualism and originalism determine the meaning of key legal sources. But it was not always this way. Other periods of our history featured a type of jurisprudence that was more sensitive to social, political, and economic considerations and saw the law as a device to address pressing problems. Significant works in legal history assess long-term patterns in these broad classes of jurisprudence, even as this literature is criticized by historians and legal theorists for relying on unrepresentative samples of legal material. This analysis applies a foundational large language model to study trends in jurisprudence empirically: using a novel dataset of annotated historical Supreme Court opinions, we make predictions with a validated, in-domain, fine-tuned large language model for the universe of Supreme Court opinions from 1870–2023. The resulting measure of jurisprudence supports two downstream analyses. First, the measure allows us to assess the consensus periodization, which maintains that the period after the Civil War was highly formal; that the mid-twentieth century was inventive and open-ended, or “grand” in one scholar’s terminology; and that we currently live in a period of high formalism. Second, the measure permits us to arbitrate points of disagreement and confusion in the literature: the timing and nature of the early-twentieth century transition away from formalism; the degree to which the judicial “revolution” of 1937 was sudden or, instead, gradual and evolutionary; the mid-century rise of inventive and open-ended jurisprudence, and the degree to which it is associated with Chief Justice Warren; and the timing and nature of the formalist revival in the 1980s, including the degree to which it is associated with Justice Scalia. Our findings provide a basis for the development of a legal theory of changes in macro-jurisprudence and suggest a focus on underlying structural determinants.