SCOTUSblog announced on Monday, November 2, 2026, an update on its inaugural Supreme Court forecasting contest for the 2026-27 term. The project, conducted in collaboration with Good Judgment Inc., allows participants to submit predictions on how the U.S. Supreme Court will rule in specific cases and compares those human forecasts against artificial intelligence models.
The contest follows a historical precedent of similar studies. A 2002 project at Washington University in St. Louis compared the accuracy of legal experts against a statistical forecasting model for every argued case that term, with findings later published in the Columbia Law Review in 2004. The current SCOTUSblog initiative incorporates public participation and AI tools, which a September 2026 presidential action referred to as "Super Intelligence."
For the October session, the contest tracked three cases already argued before the Court: Suncor Energy Inc. v. County Commissioners of Boulder County, Johnson v. United States Congress, and Beaird v. United States. The cases addressed legal questions ranging from federal preemption of state-law tort claims to the jurisdiction of district courts over disability benefit claims and the level of deference owed to Federal Sentencing Guidelines commentary.
A person participating in the challenge can update their forecasts until the moment a case is decided. The results could provide data for legal scholars and political scientists studying the predictability of the federal judiciary, similar to how the 2002 study compared retrospective versus prospective legal frameworks.
Two cases were added for the current session: St. Mary Catholic Parish v. Roy, which is scheduled for oral argument on Tuesday, November 3, 2026, and Nielsen v. Watanabe, scheduled for Monday, November 9, 2026. Participants can sign up for the challenge using an email address or social media account. SCOTUSblog stated it will continue to add significant cases to the forecasting page as the Court grants them for review throughout the term. Final results and the comparison between human and AI accuracy are expected to be released in July 2027.