James Surowiecki's acclaimed book The Wisdom of Crowds offers a good overview of cases in which decentralised group processes lead to excellent collective judgments, sometimes better than those from individual experts.
Surowiecki's book begins with a 1906 contest to guess the weight of an ox at an English fairground. The scientist Francis Galton was amazed to discover that the average of the 787 guesses made was almost perfectly accurate, even though the 787 contestants were mostly non-experts. The book then moves to the Iowa Electronic Markets, in which hundreds of self-selected traders buy futures to predict election results - a process that often produces better results than either polls or pundits.
Surowiecki's argument is that many ordinary individuals can collectively make good decisions under conditions of diversity, independence and decentralisation - provided there is an appropriate method of aggregating the results.
Why doesn't this argument simply generalise to mass democracy? The many contestants who guessed the weight of the ox were not experts. Indeed, Galton, on Surowiecki's telling, treated them as "voters" and was interested in discrediting democracy - until he found out how well the public had done. But voters do not seem to do as well with elections and referenda as in Surowiecki's examples. Yet voters are, in fact, diverse - or at least as diverse as the entire electorate in a mass democracy. And they are independent - at least in the sense that they can make whatever decisions they like by secret ballot, and their decision processes are decentralised in terms of millions of individual choices aggregated by the results from polling booths. So why can't we transplant the wisdom of crowds directly to mass democracy? Or can we?
James S Fishkin is director of the Center for Deliberative Democracy in the department of communication, Stanford University. He is a pioneer of "deliberative polling" - a practice of public consultation that employs random samples of the citizenry to explore how opinions would change if they were more informed.
Among his books are The Voice of the People: Public Opinion and Democracy (Yale University Press, 1995) and (with Bruce Ackerman) Deliberation Day (Yale University Press, 2004)A conventional conundrum
Surowiecki confronts democracy explicitly in his last chapter, by focusing on the contrast between conventional polling and what I call deliberative polling (or DP). He treats the latter as an idealistic and somewhat "utopian" project, but also one that succeeds in demonstrating the public's collective wisdom. It borders on being too idealistic, he believes, only because of the practical difficulties of getting large numbers of people to spend the necessary time deliberating. And he repeats economist and lawyer Richard Posner's criticisms that the public is just not interested in the public good; it is more concerned with self-interest.
However, in framing the issue this way, he still leaves our question unanswered. If deliberative polling embodies a form of collective decision that does, in fact, illustrate the wisdom of crowds, why would not conventional polling or conventional voting behaviour do so as well? All of these processes seem to fulfil Surowiecki's conditions: diversity, independence, decentralisation and a plausible method of aggregation.
A deliberative answer
Deliberative polling is an attempt to respond to four limitations in public opinion as we find it reported in conventional polls.
First, there is the problem that the public is usually not well informed on most issues. It tends to suffer from what the economist Anthony Downs termed "rational ignorance". If I have one vote among millions, why should I become informed about the details of public policy or the competing positions of candidates or parties? My individual vote or my individual opinion will not make any appreciable difference to the outcome, so why should I spend a lot of time and effort making it more informed?
A second problem is that many of the opinions reported in conventional polls do not exist. They are phantom opinions, or what political scientist Philip Converse termed "non-attitudes". Or if not non-attitudes, they can be very much top-of-the-head, reflecting only a vague impression of sound bites and headlines. Respondents to polls do not wish to admit that they "don't know", so they will often pick an alternative virtually at random rather than acknowledge ignorance. This problem was dramatised by political scientist George Bishop's study of opinions about the Public Affairs Act of 1975. The public answered survey questions about it, but the subject was fictional. There was no "Public Affairs Act of 1975".
A third limitation of public opinion is selectivity of sources - selectivity both of conversational partners and of news sources. Even if people tend to talk about politics or policy, they will tend overwhelmingly to talk to people like themselves, people from similar social backgrounds and with whom they are likely to agree. And if they look up news on the internet, they are increasingly likely (if you believe the plausible arguments of Cass R Sunstein in his book Republic.com) to select news sources with which they agree. The web increases the opportunities for people to exercise choice and as a result confirm their positions rather than challenge them.
A fourth problem is that democratic decisions take place in a context in which key political actors have incentives to manipulate, mislead and distort. Richard Posner, whom Surowiecki cites as offering a more realistic view of democracy, follows the economist Joseph Schumpeter in making political competition the keystone of democracy. And other Schumpeterians, like political scientist Ian Shapiro, have even argued that political competition will better serve "truth" than would deliberative democracy (see Ian Shapiro's book, The Moral Foundations of Politics).
But the United States election of November 2004 was held in circumstances where large portions of the American public thought that there were weapons of mass destruction actually found in Iraq, that Saddam Hussein had a direct role in the terror attacks of 11 September 2001, and that the "Swift Boat Veterans for Truth" (who opposed the Democratic presidential candidate John Kerry) were actually serving truth. It is hard to idealise as collective "wisdom" the public opinion and voting decisions premised on such misinformation. American democracy, born with aspirations of the country's founders for deliberation (at least at the elite level) has led to a democracy of mass persuasion and manipulation. We have gone from James Madison to Madison Avenue - or worse.
Democracy's design
Also in openDemocracy, a dedicated blog on the ideas and experiences informing a democratic experiment: dLiberation is edited by J Clive Matthews and features (among many others) James S Fishkin, Arthur Lupia, Amy Gutmann, and Ian O'Flynn discussing and contesting the merits of deliberative democracy in the context of the Tomorrow's Europe project on 12-14 October 2007.Deliberative polling of the sort Surowiecki describes in his chapter on democracy was designed to respond to these problems. A scientific sample is first given a survey of the conventional sort and then recruited to participate, either online or face to face, in many hours of deliberation in small group discussions, with carefully balanced briefing materials, and also with panels of competing experts who answer questions developed in the small groups.
First, each participant has effective motivation to overcome the incentives for rational ignorance. The premise of the experiment is that each individual's views matter, both in the small group discussions and in the final results. The resulting considered judgments are amplified by television, by the internet and by the press.
Second, phantom opinions are replaced by opinions that represent a lot more information, thought and discussion.
Third, the selectivity of both news sources and discussion partners is overcome. Participants are exposed to balanced briefings, competing experts and all the arguments that can be generated by a representative sample randomly assigned to small groups.
Fourth, the misinformation and the strategic incompleteness of manipulative campaign efforts give way to balanced materials and the opportunity to weigh good information. Deliberative polling employs social science to see what the people would think under transparently good conditions.
These results suggest that Surowiecki has underspecified the conditions that lead to anything that might be called collective wisdom. At least for the kinds of questions the deliberative poll is designed for - questions of collective political will-the problem of democratic institutional design would have to address the four problems outlined above. Deliberative polling is not necessarily the only way to address these concerns. But the empirical research programme at the Center for Deliberative Democracy (CDD) at Stanford University demonstrates that it is at least one successful way.
It is important to distinguish the kinds of questions for which deliberative polling is designed from those prominent in Surowiecki's book (such as guessing the weight of an ox, or the winner of an election). The latter pose factual questions. But deciding who should win an election (rather than who is most likely to win), or what public policy should be, poses a different kind of problem: a question of collective political will. What should be done? Or, what should we, citizens, do? Such questions also benefit from diversity, from independent judgment and from decentralisation.
But questions of public policy or decision also require a modicum of basic knowledge and an openness to discussion with those who have different interests and values. The CDD has tried or is preparing to undertake this process in thirteen countries, including the United States, Britain, Denmark, Australia, Bulgaria, and Canada; the Tomorrow's Europe project on 12-14 October 2007 will engage all twenty-seven countries of the European Union in a single trans-European deliberative poll.
The experience in all these cases is that deliberative public opinion offers informative contrasts with top-of-the-head opinion. Not only does it differ significantly, but the changes in opinion are driven by the acquisition of information in a context where participants weigh balanced and competing arguments. Crowds can be wise. But they can also be foolish - driven by manipulative efforts to pander half-truths and misinformation.
The work on developing the institutional designs where the potential for mass wisdom can be harnessed for democracy has only just begun. The idea of the "wisdom of crowds" contributes to that dialogue, but there is much more to be done to realise the potential of an approach - deliberative polling - that can enhance democratic practice, process and understanding.
