Optimal Patent Jurisprudence

Abstract

We model judicial learning about optimal patent policy. The court is infinitely lived; the plaintiff and defendant are short lived. Litigated cases provide the court with information about the optimal rule. Different cases provide different sorts of information. Opinions influence the stream of future cases likely to be litigated and, as a result, change the flow of information to the court. In structuring opinions, courts make decisions whether to learn fast or slow. We have three main results. First, patent law will stabilize even if the court places zero value on the "predictability" of legal rules. Second, path dependence of law is a rare outcome. It occurs only when the court stops learning and decides that the error costs (the losses from some cases going the wrong way) are lower than the decision costs. Finally, the law can be optimally incoherent in the short run. The court will pay lip service to prior holdings, while dramatically altering the legal landscape. Patent opinion incoherence, which is often the subject of much scholarly critique, makes sense because it facilitates future learning from a population of cases most important to the court for policy-making.