Scott Page's new book is one of the most enjoyable and, at the same time, challenging books that I have read in the last decade. It is of considerable importance for political scientists, as well as other social scientists studying difficult decision-making processes. Page's work is relevant for the study of small teams designing future research, for all forms of collective action, for how citizens make decisions about policies to be adopted, for legislative decision making, and for judicial decision making by juries and panels of judges. His work is fundamental for understanding why highly decentralized systems such as markets and democracies are likely to produce better outcomes than monopolies or dictatorships.
Early in the book (p. 4), Page posits the “Diversity Conjecture” as a jumping-off point. The conjecture is that “diversity leads to better outcomes.” He models a variety of settings to show how diversity does improve outcomes. He also ingeniously illustrates many of his arguments with well-known old sayings like “Two heads are better than one” versus “Too many cooks spoil the broth.” These classic sayings illustrate one of the fundamental tensions that he reveals throughout the book. Given that the preferences and major goals of participants are relatively similar, substantial diversity leads to better problem solving. On the other hand, when goals differ, diversity may add to the costs of making decisions, as well as adversely affecting the quality of the decisions. Thus, he is careful to avoid making diversity itself into a panacea. Too much attention has been paid, however, in the social science literature to the costs of diversity, and the author directs our attention to the potential benefits of diversity leading to better decisions.
Page unpacks the concept of diversity into a set of four tools: Perspectives are the ways that people represent problems and relevant situations; interpretations are how people categorize or cut up perspectives; heuristics are the almost automatic ways that people adopt solutions to the problems they face in light of the interpretations they hold of a perspective; and predictive models help people infer the cause and effect of complex systems based on the other three tools. In light of his useful conceptual distinctions, Page makes three major claims: 1) Diverse perspectives and tools enable collections of people to find more and better solutions and contribute to overall productivity; 2) diverse predictive models enable crowds of people to predict values accurately; and 3) diverse fundamental preferences frustrate the process of making choices (p. 13).
He has an incredible array of empirical examples, mathematical models, and cogent illustrations of what he is discussing. He also brings a lovely sense of humor to his work.
Page's analysis of the wisdom of crowds outperforming the knowledge of experts is backed up by a recent article by Kenneth Arrow (and 21 other notable scholars), “The Promise of Prediction Markets,” in Science (May 16, 2008, pp. 877–88). Arrow and colleagues describe how prediction markets, such as the Iowa Electronic Market (IEM), have outperformed other tools, such as the Gallup poll surveys, in predicting the two-party vote shares for the presidential elections since 1988. By enabling diverse participants to buy “stocks” related to political events, the IEM is able to begin predicting outcomes as early as 150 days before an election. The IEM has been able to predict the outcome with an average error of only 1.5 percentage points in the week before an election, compared to the average error of 2.1 percentage points for the final Gallup poll p. 877). The coauthors (including co-winners of the Nobel Prize Tom Schelling and Vernon Smith) are urging state and federal governments to change their “gambling” laws to facilitate, rather than retard, such innovative tools as the IEM and other predictive markets so as to enhance the accuracy of predictions of both market and political events over other tools that do not draw extensively on diverse participants.
Page stresses that including diverse participants with different perspectives and tools in making decisions is better than simply providing the “insurance” generated when people invest in a diverse investment portfolio. Having a portfolio of tools is one important factor leading to better outcomes. Portfolios are not interactive, however, and do not generate the “superadditivity” that comes from establishing diverse problem-solving groups. Diverse groups, who are in an environment where participants exchange ideas, can build on an initial innovative solution made by one participant by suggesting improvements that were not included in it. And then, others can suggest still further improvements, building on an evolving solution and drawing on their multiple perspectives and skills.
It is this process of building improvements on improvements that Page wants scholars to understand and take seriously. Science itself, when organized in an open forum with diverse participants challenging one another's work, is a superadditive process. So I urge scholars to read The Difference and to build on the multiple insights that he rigorously develops. Instead of strengthening the boundaries between disciplinary “silos,” we could enhance superadditive knowledge development by increasing the diversity of perspectives we bring to our work.
