In a world where circumstances never changed and where every judicial decision was unassailably correct, applying the doctrine of stare decisis would be a breeze. Fidelity to precedent and commitment to sound legal interpretation would meld into a single, coherent enterprise. That world, alas, is not the one we live in. Like so much else in law, the concept of stare decisis encompasses a series of trade-offs. Prominent among them is the tension between allowing past decisions to remain settled and establishing a body of legal rules that is flexible enough to adapt and improve over time.
Notwithstanding pervasive disagreement over the application of stare decisis to particular disputes, the doctrine is well established in American jurisprudence. Indeed, the Supreme Court has gone so far as to describe stare decisis as indispensable to the rule of law. But as Jeremy Waldron skillfully reminds us, justifying the doctrine requires more than platitudes. In a recent article, Professor Waldron concentrates on developing what we might think of as the rule-of-law case for precedent. Central to his project is the recognition that rule-of-law benefits arise at several distinct points along the path from initial ruling to subsequent application.
In this brief essay, I wish to build on Professor Waldron’s thoughtful analysis by saying something more about the other side of stare decisis. The rule-of-law benefits of stare decisis are invariably accompanied by rule-of-law costs. In light of those costs, the ultimate question is not whether there are ways in which stare decisis promotes the rule of law. Rather, it is whether stare decisis advances the rule of law on net. Some departures from precedent can promote the rule of law, and some reaffirmances can impair it. Even if the rule of law were the only value that mattered, excessive fidelity to flawed precedents would be cause for concern. That rule-of-law ambivalence, I will suggest, should be brought to bear in calibrating the strength of deference that judicial precedents receive.