Reviewed: H. Landemore et J. Elster dir., "La sagesse collective", Raison publique, Paris, Presses de l’Université Paris Sorbonne, mai 2010, 432 p., 14 €.
Issue 12 of Raison Publique  provides a stimulating examination of “collective wisdom.” What makes this topic interesting is that it is located at the crossroads between sociology and political philosophy: the former traditionally studies the voter’s competence and social assumptions, while the latter has, since Plato and Aristotle, considered the ambiguous relationship between number and reason in normative terms. American political sociology, several examples of which are found in this issue, is structured around the debate between the proponents of the homo œconomicus paradigm and those who challenge the relevance of this approach to explaining political behavior, as well as the majority’s incompetence, which they attribute to a lack of the immediate, “selfish” interests that are presumed to motivate political commitment and participation. This approach to democracy’s advantages offers quite a different perspective from that of French political science. If democracy is justified, from this perspective, it is because, as several papers demonstrate, its superiority is epistemic: in other words, the majority, as Aristotle realized long ago, is more likely than a smaller group to be right.
Reason and Number
The primary argument that Hélène Landemore, an assistant professor of political science at Yale, makes in favor of the greater rationality of large numbers of people compared to smaller groups of experts is their “cognitive diversity”—that is, their “capacity to see the world from different points of view” (p. 10). In this way, she believes, larger groups make up for their lack of information. According to Landemore, this claim is as valid when voting is preceded by deliberation as for elections in which no prior discussion occurs to help voters make their decision. Landemore draws on statistics and invokes the “paradox of the jury,” which shows that if voters in an election cast their ballots independently of one another and are sincere about their choices, a majority is more likely to choose the better option than a single individual. This likelihood, moreover, increases as the group’s numbers rise. The “miracle of aggregation,” for its part, refers to a phenomenon in statistics whereby the average of a group’s answers to a precise, quantifiable question proves statistically accurate, since erroneous answers tend to cancel each other out. The most famous example comes from Galton, who asked a village’s inhabitants to estimate a bull’s weight: the average of their answers proved correct, give or take 500 grams. Aside from the objections that can be made against these two statistical laws, it is unfortunate, given the importance Landemore attaches to the “miracle of aggregations,” that she fails to explain how it might apply to elections involving discrete choices rather than quantifiable evaluations (p. 35).
Information and Political Competence
Landemore’s debate with Bryan Caplan, a professor of economics at George Mason University, on the question of electoral competence echoes, to some extent, controversies in French political science. Landemore points out that information is not synonymous with competence: empirical studies purporting to show that voters are incompetent are premised on a particular conception—namely, that of specialists—of what political competence means. Caplan replies that if one rejects all the usual criteria for measuring competence without proposing alternatives, it becomes impossible to prove democracy’s epistemic superiority; indeed, the latter can only be measured if one accepts that political questions admit right and wrong answers. It should be noted that nearly all of Caplan’s examples are drawn from economics and involve questions admitting only one right answer. However, Caplan also challenges the notion—yet without developing this key point—that lack of information has no bearing on “important decisions” (to use the language of Arthur Lupia, who maintains that genuinely political questions relating to key social issues require no special knowledge), despite the fact that they are based on moral choices rather than knowledge.
This tightly argued and stimulating discussion on the connection between information and competence is continued by Caplan and Gerry Mackie, a political science professor at the University of California. The latter attacks the economic approach to political behavior, particularly the concept of “rational ignorance” (coined by Anthony Downs, one of the pioneers of this approach  , which maintains that, given the miniscule impact one’s vote has on the overall result, it is rational to be uninformed. Thus paradoxically, it is precisely because voters are rational that they are incompetent. Yet given the levels of participation in Western democracies, other motivations besides the hope of having a direct impact on the result must be taken into account in explaining why people vote. One incentive is the desire to express a preference. In other words, one must abandon the theory that voters are driven only by selfishness and that they make purely instrumental calculations when casting their ballots. In conjunction with his main thesis, Mackie also challenges the claim that the average voter is incompetent, drawing on arguments familiar to contemporary sociologists : even when they lack information, voters will use heuristic shortcuts—a candidate’s self-presentation, her vocabulary, and so on—to form an elementary but sufficient sketch of their political positions.
Maximizing Collective Wisdom
According to Adrian Vermeule (John Watson Professor of Law at Harvard Law School), arguments in favor of collective rationality should be used to promote the United States Congress over the growing power of the Supreme Court. From this conception of collective rationality, Vermeule draws three normative conclusions: first, maximizing epistemic quality requires a major reallocation of legal authority from supreme courts to legislatures, given their size and greater cognitive diversity. Furthermore, in order to make the most of the “paradox of the jury,” the number of Congressional representatives must be increased: Vermeule maintains, drawing on List and Goodin’s calculations, that while the average representative’s competence is 0.51 when two possible choices exist, increasing the number of representatives from 101 to 601 would raise the body’s collective competence from 0.58 to 0.69. Finally, the Supreme Court should have members who do not have legal backgrounds, in order to “reap benefits in terms both of competence and group diversity” (p. 199).
Jon Elster, a professor at the Collège de France, seeks for his part to define the ideal conditions for constitution-writing: relatively large assemblies, in addition to their epistemic advantages, minimize the role of interests, as they afford little weight to horse-trading and vote trading. Elster thus encounters one of the concerns of the proponents of the economic paradigm (with which, nevertheless, he takes his distances) when he points out that an assembly must not, however, be too large, lest representatives fail to inform themselves, due to the relative insignificance of their votes.
How to Make Reason Matter?
Philippe Urfalino, an associate professor at EHESS (Paris) partially reverses the perspective by considering not the wisdom of crowds, but the rationality of small groups. He asks: how can one make the most of the competence of panels of experts—i.e., “wise men”—when they do not rely on numbers or majority rule. Drawing on the example of the medieval Catholic Church, which, for the election of canons, bishops, and popes, weighted majority and “saniority”—i.e., the number of voters and their presumed wisdom—Urfalino demonstrates that drug approval panels in France and the United States share many of its characteristics: in both organizations, emphasis is placed on giving reasons rather than stating preferences; the decision-making process in both instances is twofold, with each proposal of the French commission being collegially discussed, while each vote of an American advisory committee is either accepted or rejected by the Food and Drug Administration; and hierarchy is ubiquitous, whether in considering arguments or in making the final decision. The stimulating question posed by Urfalino is why, when the church had to abandon the notion of sanior pars, which threatened its unity due to the dubious wisdom invoked in support of some candidates, these two organizations are so efficient at reaching decisions in the cases submitted to them. In the American committees, in which members vote, there is no conflict between numbers and wisdom because the experts largely agree with one another; in the French case, which is more intriguing as members are required to reach consensus, the members’ cognitive homogeneity greatly facilitates unanimity (as does the special role of the chairman). Cognitive homogeneity is particularly interesting: when cognitive diversity increases, as occurs in most European drug approval panels (which bring together specialists from various European states), consensus becomes more difficult to achieve. Though Urfalino does not himself explore this idea, it is possible to conclude that the argument advanced in the other papers—namely, that cognitive diversity tends to increase the probability that the majority will strike upon the best option—is in need of qualification. In small, very well informed groups, it can, to the contrary, have obstructive effects.
We will limit our conclusion to one brief observation. It remains unclear how applicable the “paradox of the jury” or the “miracle of aggregation” are to preferences, a question that the papers rarely address. Politics, far from being simply an epistemic process, is first and foremost a choice of values. Therefore, if one admits that in the realm of moral decisions, there is no definitive criterion of right and wrong, the question of democracy’s possible epistemic superiority becomes pointless, as there is no other way to justify democracy than by invoking the necessity of individual autonomy. Politics is inseparable from the question of justice; unless we entertain the idea, as Plato did, by referring to a cosmology in which we no longer believe, that justice is a knowledge that some handle better than others, the equal participation of all citizens in the distribution of goods and tasks must be the most fundamental condition for a proceduralist form of justice.