Actuarial Review Return to Main Page

Quarterly Review


Twenty Heads are Better than One

The Wisdom of Crowds: Why the Many Are Smarter than the Few and How Collective Wisdom Shapes Business, Economies, Societies, and Nations by James Surowiecki
[Doubleday, 2004, $24.95]

Reviewed by Curtis Gary Dean

Contestants on the TV game show Who Wants to Be a Millionaire have three "lifelines" to help them as they answer multiple-choice questions to work their way up the money ladder. If they are stumped on a question, one lifeline is to call a friend or relative. Most likely the contestant has some faith in the friend's ability. According to the author, the friend is correct about 65 percent of the time, a pretty good result given that a random selection would be correct 25 percent of the time. Another lifeline is to ask the audience. The audience votes on the four possible answers and vote counts are tallied and displayed for each answer. The audience does better, much better, with the answer getting the most votes being correct 91 percent of the time!

This simple audience vote illustrates the author's thesis: the aggregated opinion of a group of people regularly beats the experts in solving problems, and this competence extends to a wide range of situations. (It may be a stretch to call the friend an expert, but it is intriguing to note how much better the group does than the friend.) In fact, experts often do poorly with complex problems, while a group of people, many with limited expertise, may produce excellent solutions. The author is a staff writer for The New Yorker. He provides many interesting and diverse examples showing the astonishing accuracy of group solutions. He supports his arguments with results from academic and business studies.

How can the solution from a group be so accurate? The author's contention is that the solution for each member of a group consists of two components: one component reflects the actual knowledge of the member and the other component is random error. The key concept is that over a group the average of the random error components tends towards zero, letting the total average reflect the sum of the actual knowledge within the group. Just like in ratemaking, aggregating a group of policies gives us a better picture of the expected value.

For the crowd's wisdom to shine, four conditions should be present:

  1. diversity of opinion
  2. independence
  3. decentralization, and
  4. aggregation
Diversity of opinion expands the set of possible solutions. Independence is necessary for the error components of individual solutions to average out to zero, and independent individuals are more likely to have new information. Decentralization reduces the likelihood that a central authority can control the outcome or thinking of the group.

Some mechanism is necessary to aggregate the crowd members' individual opinions. In the example above, aggregation is accomplished through voting: one vote per person. In the Iowa Electronic Markets the aggregation is accomplished by prices for real-money futures. These markets are operated by faculty at the University of Iowa Tippie College of Business. Participants can make small bets on the outcome of political or economic events. Check out www.biz.uiowa.edu/iem. As of the middle of the day on July 27, 2004, the last trade price for a DEM04 contract was $.502. This contract will pay $1 to the holder if the Democratic presidential candidate receives the highest popular vote in the November '04 election. The last trade price for a REP04 contract was $.505. In the world's financial markets, asset prices reflect the "voting" of market participants. The search engine Google uses a weighted voting process where each link to a Web page counts as a vote, and the weight of a vote depends on the importance of the other linked page, which again is determined by voting.

The author also discusses the pitfalls of collective decision making. Dissent is often buried as a group tries to reach a consensus. The author explains that even one dissenting opinion can force a group to act more wisely because the group now has to go back and examine the facts and inferences. Disagreement and conflict of ideas are more likely to result in an eventual good collective decision than early consensus or compromise.

Groupthink may push a group further away from a good solution into more extreme positions, a condition referred to as group polarization. Another problem is that groups often give too much weight to the rank and status of particular members. Many groups are sorely lacking in cognitive diversity. Poorly structured or managed groups can produce disastrous results as in the Columbia shuttle tragedy, a case discussed at length in the book.

Our modern society is a mixture of coordination, collaboration, and competition. There is often a conflict between maximizing personal gain and contributing to social welfare. The author discusses some of the conflicts and the methods that have evolved to handle them.

Actuaries are regularly involved in important and complex decisions, both as individual experts and as members of a group. This short and easy-to-read book should be of interest to most. We regularly participate in collective decision making but rarely think deeply about the structure and dynamics of the process. The author provides plenty to think about as he discusses decision making in corporations, markets, and societies.

Click here to write a Letter to the Editors