Prediction Markets and Law: A Skeptical Account

Abstract

Enthusiasm for "many minds" arguments has infected legal academia. Scholars now champion the virtues of groupthink, something once thought to have only vices. It turns out that groups often outperform individuals in aggregating information, weighing alternatives, and making decisions. And although some of our legal institutions, such as Congress and juries, already harness the power of the crowd, others could be improved by multiplying the number of minds at work. "Multiplying" implies a simple mathematical formula for improving decisionmaking; modern many minds arguments are more sophisticated than that. They use incentive analyses, game theory, and statistics to study how and under what circumstances groups make better choices than individuals do. The models propose to solve various information problems, such as determining guilt or innocence, deciding on a course of regulation, or estimating a value that is difficult to measure directly.

Most ambitious, perhaps, has been the attempt to aggregate knowledge to predict the future. Uncertainty is a painful part of reality; it is only natural that the wisdom of the crowd would be summoned to battle it. The most popular model on that front has been the "information market" or "prediction market." (The terms can be used interchangeably.) In particular, scholars have argued that such markets may alleviate uncertainty in legal and policy analysis. This Note argues that enthusiasm for prediction markets in law is misplaced. No one thinks prediction markets are perfect; even their proponents concede that they will fail under certain circumstances. But with their concessions they give up the game, at least as applied to legal problems: the circumstances in which prediction markets are inaccurate are precisely the circumstances in which law needs them most.

Part I surveys information markets - their success stories and their limitations. Part II begins by outlining the ambitions scholars have for information markets and law. Part II then develops the thesis of this Note: that the performance of prediction markets is inversely correlated with how valuable their predictions would be. This Part argues that if a future event is secret or knowledge about its likelihood is thin, if it depends on the idiosyncratic action of an individual, or if it is catastrophic but unlikely, a prediction market will probably not produce accurate information. Finally, Part III defines the niche, smaller than scholars imagine, in which prediction markets shine.