Washington Law Review


The “watershed” doctrine gives prisoners a constitutional basis to reopen their cases based on a new due process protection that would have made a difference had it been announced before their appeals were exhausted. The Supreme Court has imposed nearly impossible conditions, however, for any new rule of criminal procedure to apply retroactively to a final conviction or sentence. No such rule can be backdated unless it enhances not only the accuracy of criminal verdicts, but also “our very understanding of the bedrock” tenets of fairness in criminal trials. The Court refers to rules that satisfy both these requirements as “watersheds.” In the quarter-century since it established this doctrine, the Court has denied the accuracy-and-fairness credentials to every one of the dozens of new rules it has characterized as procedural and whose watershed status it has considered. Scholarly consensus accordingly casts watershed doctrine as exceptional, esoteric, and insignificant. This Article challenges that consensus. We use the dynamic concentration model of game theory to show how watershed doctrine counteracts the structural undersupply of constitutional due process rules. The Court maintains too small a caseload to scrutinize more than a fraction of due process violations or specify every such procedural demand. That institution is accordingly ill equipped to rein in the punitive tendencies of elected state judges who owe their jobs to electorates that tend to value crime prevention more than defendants’ rights. Watershed doctrine potentially mitigates this enforcement problem by creating an extreme, if low-probability, threat of repealing scores of final convictions. By issuing a single new watershed rule, the Court can mandate sweeping retrials or release of prisoners into the public. This existential threat provides an overlooked reason why state courts might insulate their states’ criminal procedures against Supreme Court incursions. To achieve the desired insulation, state courts can create constitutional safe harbors by trying to align their procedures with watersheds they project the Court might announce in the future. Indirect support for this theory comes from our comprehensive study of the hundreds of watershed decisions that state courts have issued since 1989. We narrowed this list down to the 228 controlling decisions about whether to backdate distinct due process rules across different jurisdictions. Our analysis found that twenty-seven, or more than one in nine, of these decisions inflate the retroactivity rights of criminal defendants.

First Page