Many Minds, Many MDL Judges
Fitzpatrick, Brian T.
Over his long career, Francis McGovern was a leading supporter of decentralizing the fact finding that goes on in multidistrict litigation (MDL). His advocacy of letting torts "mature" gave rise to the sampling that takes place in today's bellwether jury system.' More recently, he advocated selective remands of cases from MDLs to other judges so the litigation could proceed in parallel before multiple judges rather than serially before one. In this Article, I try to formalize and extend the intuition behind McGovern's ideas and ask why, if decentralization is good for fact finding, is it not also good for the legal decisionmaking that takes place in MDLs? My answer is that decentralization is indeed just as good on the legal side. I therefore analyze two ideas for how we might capture the benefits of decentralized legal decisionmaking without incurring too many costs. This is an important matter because the federal MDL statute now concentrates more power in the hands of a single person than perhaps any other part of our judicial system. A single judge can end up resolving hundreds, thousands, or even hundreds of thousands of individually viable cases. This has benefits-most notably efficient case processing, uniformity of results, and the facilitation of global peace-but it also has costs. My focus here is on a cost that has been surprisingly neglected by scholars but may be the greatest cost of them all: the accurate adjudication of legal claims and defenses. I suspect it is intuitive to most of us that asking one person to decide something instead of inviting many other people to weigh in probably reduces the quality of the resulting decision. There is a literature that formalizes this intuition called "many-minds" scholarship. It proceeds from a famous mathematics proof known as the Condorcet Jury Theorem. Although some people have questioned the applicability of many-minds theories to legal decisionmaking, if there were ever a legal context in which they could be applicable, I argue it is in the context of our MDL system.