Statistical Precedent: Allocating Judicial Attention
Suffering from a well-covered “crisis of volume,” the U.S. Courts of Appeals have patched together an ad hoc system of triage in an effort to provide cases with sufficient attention. For example, only some cases are assigned to central staff, analyzed by law clerks, orally argued, debated over by judges, or decided in published opinions. The courts have evaded overt disaster by increasing the number of active, senior, and visiting judges, but adding personnel poses its own demands on attention—judges must also pay attention to one another in order to coherently develop and apply the law. With too little time and too many voices, they have increasingly abandoned the effort to coordinate that uniform approach to judging: the courts now create traditional precedent in less than 10% of cases, some larger courts have stopped the practice of circulating opinion drafts to the full court, and en banc proceedings are initiated at a minuscule rate.
This Article explains and illustrates how courts can leverage advances in artificial intelligence to more fairly and effectively allocate attention. A machine-generated mapping of a court’s historical decision patterns—what I term “statistical precedent”—can help a circuit court locate the district court, agency, staff attorney, law clerk, and panel decisions that are most incompatible with the court’s collective jurisprudence. Statistical precedent can also aid the court in identifying areas of law that are most in need of development. With the ability to locate likely errors and opportunities for law development, the circuit courts could distribute attention so as to revitalize their contribution to the rule of law.
Ryan W. Copus