Infotopia: How Many Minds Produce Knowledge (review)

It takes a vivid imagination to see Condorcet's Jury Theorem, Hayek's view of markets as information-aggregating devices, Wikis, prediction markets, open source systems, blogs and deliberative assemblies as examples of the same phenomenon. Cass Sunstein, prolific and wide-ranging legal scholar at the University of Chicago, has the requisite imagination. In his accessible Infotopia, he explores a distinctive class of social arrangements. In these arrangements, advocates and users suppose they are improving collective decision making by aggregating information that no single individual, however well placed and powerful, can assemble, much less evaluate, alone.

Many minds, the book's subtitle declares, produce knowledge. Condorcet figures because of his demonstration that so long as the average participant in a binary decision process is more likely to be right than to be wrong, the larger the number of participants the more frequently the aggregate arrives at the right answer. Hayek enters the scene with his anti-socialist claim that markets, which draw on the wide range of knowledge available to individual buyers and sellers, necessarily arrive at superior decisions no central planning can rival. Never fear. Sunstein provides nothing like a doctrinaire attack on governments or a technical treatise on collective decision making. On the contrary, in a conversational style with many an aside, he offers well-informed but tentative conclusions concerning the strengths and weaknesses of competing means for aggregating individual information. Deliberation, in his account, comes off less well than its democratic advocates claim because it remains vulnerable to bandwagon effects, to group interests in outcomes, to reticence on the part of well informed minorities, to interpersonal pressures and to faulty information.

Prediction markets score better for Sunstein than deliberation. Sunstein inventories markets for influenza prediction, Hollywood box office success and a number of other intriguing outcomes in a convenient appendix. He demonstrates the great popularity they have gained over the past decade. Yet he never really explains the process by which giving people competitive incentives in such markets produces more accurate predictions than simply polling the populace and computing the distribution of predictions.

These earlier sections of the book provide background for Sunstein's enthusiastic treatment of wikis and open source systems as well as his more qualified assessment of blogs, which he finally judges to be subject to many of the same faults as collective deliberation. They often fall prey to information cascades, ideological bias, polarization and unreliable information. But Sunstein's preferred alternatives don't escape unscathed. Wikis, open source, and similar arrangements, it turns out, only work with significant central coordination – authorized monitors who examine new contributions for value and/or "copyleft" agreements in which new contributors guarantee that they will keep their contributions in the public domain. In short, they fall far short of the automatic information aggregation so admired by Hayek.

Four features of Sunstein's analysis will jar many sociologists. First, it assumes that aggregators of individual choices are looking for crisp correct answers, which makes sense for such questions as who will win the next election, but less sense when the object of aggregation is to produce consensus, participation or reduced ignorance. Second, despite its general treatment of cascades and polarization, the analysis pays almost no attention to interpersonal processes by which people pass information – good or bad – among themselves, and thus create relatively homogeneous cultural clusters. Third, while acknowledging that strict self-interest fails to explain why people make positive contributions to collective knowledge in wikis and elsewhere, Sunstein provides no coherent account of the incentives, impulses or pressures that promote such contributions. Fourth, aside from a few concessions that markets sometimes succumb to fads and bubbles, it displays surprisingly little curiosity about what markets actually produce and how: prices, yes, but in what sense does the distribution of prices in today's markets maximize wisdom, well being, or even social efficiency? The analogy between market-driven price-setting and prediction of technical innovations or baseball championships strikes me as weak and misleading. Extending Harrison White on the...