Making Dumb Groups Smarter PDF

HBR.ORG DECEMBER 2014 REPRINT R1412F...

HBR.ORG DECEMBER 2014 REPRINT R1412F t os rP Making Dumb yo Groups Smarter op The new science of group decision making tC by Cass R. Sunstein and Reid Hastie No Do This document is authorized for educator review use only by Prabakar S, Other (University not listed) until Jun 2023. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860  t os rP yo op tC MAKING No DUMB GROUPS Do 2 Harvard Business Review December 2014 This document is authorized for educator review use only by Prabakar S, Other (University not listed) until Jun 2023. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860  FOR ARTICLE REPRINTS CALL 800-988-0886 OR 617-783-7500, OR VISIT HBR.ORG t os Cass R. Sunstein is the Robert of Chicago Booth School of Business. Walmsley University Professor at They are the authors of Wiser: Harvard Law School. Reid Hastie Getting Beyond Groupthink to Make is the Ralph and Dorothy Keller Groups Smarter (Harvard Business Distinguished Service Professor of Review Press, 2015), from which this Behavioral Science at the University article is adapted. rP yo op tC No The new science of group decision making by Cass R. Sunstein SMARTER Do and Reid Hastie JOSH MCKIBLE December 2014 Harvard Business Review 3 This document is authorized for educator review use only by Prabakar S, Other (University not listed) until Jun 2023. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860  MAKING DUMB GROUPS SMARTER t os SINCE THE BEGINNING OF HUMAN HISTORY, PEOPLE rP have made decisions in groups. As the saying goes, the public consciousness, and it has yet to have a no- two heads are better than one. If so, then three heads ticeable effect on actual practice. It’s time for that to should be better than two, and four better still. With change. We aim to bring behavioral research into direct a hundred or a thousand, then, things are bound to contact with the question of group performance—to go well—hence the supposed wisdom of crowds. describe the main ways in which groups go astray and The advantage of a group, wrote one early ad- to offer some simple suggestions for improvement. yo vocate of collective intelligence—Aristotle—is that “when there are many who contribute to the process Why Do Errors Occur? of deliberation, each can bring his share of goodness Groups err for two main reasons. The first involves and moral prudence…some appreciate one part, informational signals. Naturally enough, people some another, and all together appreciate all.” The learn from one another; the problem is that groups key is information aggregation: Different people take often go wrong when some members receive incor- note of different “parts,” and if those parts are prop- rect signals from other members. The second in- erly aggregated, they will lead the group to know volves reputational pressures, which lead people to op more (and better) than any individual. silence themselves or change their views in order to Unfortunately, groups all too often fail to live up avoid some penalty—often, merely the disapproval to this potential. Companies bet on products that of others. But if those others have special authority are doomed to fail, miss out on spectacular oppor- or wield power, their disapproval can produce seri- tunities, pursue unsuccessful competitive strategies. ous personal consequences. In governments, policy judgments misfire, hurting As a result of informational signals and repu- tC thousands or even millions of people in the process. tational pressures, groups run into four separate “Groupthink” is the term most often applied to the though interrelated problems. When they make poor tendency of groups to go astray. Popularized in the or self-destructive decisions, one or more of these early 1970s by the psychologist Irving Janis, it has de- problems are usually to blame: servedly entered the popular lexicon. But Janis’s con- Groups do not merely fail to correct the errors of tribution is more an evocative narrative than either a their members; they amplify them. scientific account of how groups go wrong or helpful They fall victim to cascade effects, as group mem- guidance for group success. Many researchers have bers follow the statements and actions of those No tried to find experimental evidence to support his who spoke or acted first. specific claims about how cohesion and leadership They become polarized, taking up positions more styles shape group behavior, to little avail. extreme than those they held before deliberations. Since Janis produced his theory, though, psychol- They focus on what everybody knows already—and ogists and other behavioral scientists have built up a thus don’t take into account critical information rich base of evidence on how and when individual that only one or a few people have. decision makers blunder. This work has attained sci- entific acclaim (including several Nobel prizes) and Amplifying Errors widespread popularity thanks to best sellers such With the psychologists Daniel Kahneman and the Do as Daniel Kahneman’s Thinking, Fast and Slow, Dan late Amos Tversky in the vanguard, behavioral scien- Ariely’s Predictably Irrational, and Nudge (which tists have identified some common mental shortcuts one of us, Sunstein, coauthored with the economist (known as heuristics) and biases that lead individuals Richard Thaler). astray. The planning fallacy, for example, leads us to A smaller but nonetheless substantial body of underestimate how much time projects will take and research—some of it our own—has focused on the how much money they’ll cost. Overconfidence leads decision-making strengths and weaknesses of groups us to believe that our forecasts are more accurate and and teams. But little of this work has trickled into precise than in fact they are. The availability heuristic 4 Harvard Business Review December 2014 COPYRIGHT © 2014 HARVARD BUSINESS SCHOOL PUBLISHING CORPORATION. ALL RIGHTS RESERVED. This document is authorized for educator review use only by Prabakar S, Other (University not listed) until Jun 2023. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860  FOR ARTICLE REPRINTS CALL 800-988-0886 OR 617-783-7500, OR VISIT HBR.ORG t Idea in Brief os THE PROBLEM WHY IT HAPPENS THE SOLUTION Groups often fail to live up to their potential Group members take informational signals Leaders can structure group deliberations as decision-making bodies. Instead of from what others say, even when the to make them more likely to succeed. aggregating the knowledge and wisdom of information is wrong or misleading, while One very simple way is to let others speak their members, they end up making bigger reputational pressures can cause them first. Another is to assign specific roles errors than individuals would. to silence themselves or change their views or areas of expertise to members of the rP in order to fit in. As a result, groups often group. The key is encouraging individuals amplify individual errors, stampede to share their diverse knowledge rather toward bad decisions, foment polarization than suppress it. and extremism, and ignore information that isn’t already widely held. yo leads us to seize on whatever springs most readily to see others making the same errors. What they see mind, because it is memorable or we recently experi- serves as “proof” of erroneous beliefs. Reputational enced it. The representativeness heuristic leads us to pressures play a complementary role: If most mem- believe that things or events or people that are similar bers of the group make errors, others may make them in one way are similar in other ways, too. Egocentric simply to avoid seeming disagreeable or foolish. bias leads us to exaggerate the extent to which our Fortunately, we have evidence that deliberating tastes and preferences are typical. The sunk-cost fal- groups can correct or reduce certain biases. For prob- op lacy leads us to stick with a hopeless project because lems that have “eureka” solutions (the right answer, we have already invested so much in it. Framing effects once announced, is clear to all), groups do well even influence our decisions according to the semantics of if individual members start out biased. Groups are how the options are presented. (For example, people also better than individuals at overcoming egocen- are more likely to agree to an operation if they are told tric bias. An individual will focus on his own tastes— that 90% of people are alive after five years than if they what he likes and what he doesn’t. If he consults tC are told that 10% of people are dead after five years.) with others, he is likely to learn that his tastes are For our purposes, the central question is idiosyncratic. In such cases, group deliberation sup- whether groups can avoid or mitigate these errors. plies an important corrective. Note that we’re less Experimental evidence indicates that they usually apt to get that corrective if the group consists of like- do not. The psychologists Roger Buehler, Dale Griffin, minded people. The influence of the availability heu- and Johanna Peetz have found, for example, that ristic, too, is slightly reduced in groups. Individual the planning fallacy is aggravated in groups. That members may each rely on “what comes to mind,” is, groups are even more optimistic than individuals but they are likely to have different memories, yield- No when estimating the time and resources necessary ing a more representative sample at the group level. to complete a task; they focus on simple, trouble- The larger point, however, is that many individual free scenarios for their future endeavors. Similarly, biases are not systematically corrected at the group Hal R. Arkes and Catherine Blumer have shown that level and often get worse. The mechanisms by which groups are even more likely than individuals to es- errors are compounded can usually be found in the calate their commitment to a course of action that other three problems of group decision making. is failing—particularly if members identify strongly with the group. There is a clue here about why com- Cascading to the Wrong Answer panies, states, and even nations often continue with The human brain may be wired from birth to syn- Do doomed projects and plans. Groups have also been chronize with and imitate other people. It is no found to increase, rather than to lessen, reliance on exaggeration to say that herding is a fundamental the representativeness heuristic; to be more prone to behavior of human groups. When it comes to group overconfidence than individual members; and to be decisions and information flow, the favored term more influenced by framing effects. among social scientists is “cascade”—a small trickle Both informational signals and reputational pres- in one direction that soon becomes a flood. sures are at work here. If most members of a group Consider a brilliant study of music downloads by tend to make certain errors, then most people will the sociologists Matthew Salganik, Peter Dodds, and December 2014 Harvard Business Review 5 This document is authorized for educator review use only by Prabakar S, Other (University not listed) until Jun 2023. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860  MAKING DUMB GROUPS SMARTER t os Duncan Watts. They allowed test subjects to listen to and download one or more of 72 songs by new Many groups on the pre-deliberation private ballot. What did juror 4 do when confronted bands. In the control group, individuals were told end up with three second-degree murder ver- rP thinking that nothing about what others had downloaded or liked dicts? He paused for a second and then and were left to make independent judgments. In said, “Second degree.” Juror 7, an un- other groups, the participants could see how many people had previously downloaded particular songs. their ultimate decided vote, suddenly spoke up and asked, “Why second degree?” The re- The researchers were testing how much difference it made, in terms of ultimate numbers of downloads, if convergence on searchers saw a deer-in-the-headlights expression flit across juror 4’s face be- people could see the behavior of others. a shared view fore he replied, “Oh, it’s just obviously was inevitable. It made a huge difference. Although the worst second degree.” We have no doubt that yo songs (as established by the control group) never similar scenarios play out every day in ended up at the very top, and the best songs never at the very bottom, essentially anything else could hap- Beware of that jury rooms, boardrooms, and confer- ence rooms all over the world. pen. If a song benefited from a burst of early down- loads, it might do quite well. Without that benefit, it thought. A reputational cascade has a differ- ent dynamic. Group members think might be a failure. And as the researchers later found, they know what is right, but they these effects occurred even if they lied to the test nonetheless go along with the group subjects about which songs were downloaded a lot. in order to maintain the good opin- op If a project, a product, a business, a politician, or ion of others. Suppose, for example, a cause gets a lot of support early on, it can win over that Albert suggests that his compa- a group even if it would have failed otherwise. Many ny’s new project is likely to succeed. groups end up thinking that their ultimate conver- Barbara is not confident that he’s right, gence on a shared view was inevitable. Beware of that but she concurs because she wishes thought. The convergence may well be an artifact of tC who was the first to speak—and hence of what we might call the architecture of the group’s discussions. Two kinds of cascades—informational and repu- tational—correspond to our two main sources of group error. In informational cascades, people si- lence themselves out of deference to the information conveyed by others. In reputational cascades, they silence themselves to avoid the opprobrium of others. No Here’s an example of an informational cascade in jury deliberations, which has important implica- tions for business. One of us (Hastie) has conducted dozens of mock-jury studies with thousands of vol- unteer jurors, many of them from big-city jury pools. In these studies the volunteers privately write down their preferred verdicts before deliberations begin and indicate how confident they are in their judg- ments. Deliberations then start, as they often do in Do real trials, with a straw vote to see where everyone stands. The vote circles the jury table and frequently begins with a small set of two or three jurors who favor, with increasing confidence, the same verdict. In one mock trial, jurors 1, 2, and 3 endorsed a verdict of second-degree murder both privately and in the straw vote. Juror 4 had voted not guilty and indicated the highest level of confidence in his choice 6 Harvard Business Review December 2014 This document is authorized for educator review use only by Prabakar S, Other (University not listed) until Jun 2023. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860  FOR ARTICLE REPRINTS CALL 800-988-0886 OR 617-783-7500, OR VISIT HBR.ORG t A Tale of Two Cities os To examine the phenomenon of group polarization, the two of us (along with the social scientist David Schkade) created an experiment in group deliberation— one that, we believe, accurately reflects much deliberation in the real world. We recruited citizens from two Colorado cities and assembled them in small not to seem ignorant, adversarial, or skeptical. If groups (usually six people), all from the same city. The groups were asked to Albert and Barbara seem to agree that the project will deliberate on three of the most contested issues of the time: climate change, go well, Cynthia not only won’t contradict them pub- affirmative action, and same-sex civil unions. The two cities were Boulder, rP licly but might even appear to share their judgment— known by its voting patterns to be predominantly liberal, and Colorado not because she believes it to be correct (she doesn’t), Springs, known by its voting patterns to be predominantly conservative. We but because she doesn’t want to face their hostility did a reality check on the participants before the experiment started to or lose their good opinion. Once Albert, Barbara, and ensure that the Boulder residents were in fact left of center and the Colorado Cynthia offer a united front on the issue, their col- Springs residents were right of center. league David will be most reluctant to contradict Group members were asked first to record their views individually and them, even if he’s pretty sure they’re wrong and has anonymously and then to deliberate together in an effort to reach a group excellent reasons for that belief. (Incipient evidence decision. After the deliberations the participants were again asked to record yo their views individually and anonymously. Here’s what we found: indicates that women are especially likely to self- censor during discussions of stereotypically male subjects, such as sports, and that men are especially likely to self-censor during discussions of stereotypi- 1 People from Boulder became a lot more liberal, and people from Colorado Springs became a lot more conservative. Not only were the group “verdicts” more extreme than the pre-deliberation averages cally female subjects, such as fashion. In both cases, of group members, but the anonymous views of individual members groups lose valuable information.) became more extreme as well. “Political correctness,” a term much used by the political right in the 1990s, is hardly limited to left- 2 Deliberation decreased the diversity of opinion among group op members. Before the groups started to deliberate, many of them leaning academic institutions. In both business and showed considerable divergence in individual opinions. Discussion government there is often a clear sense that a certain brought liberals in line with one another and conservatives in line with point of view is the proper one and that those who one another. After a brief period of discussion, group members showed question or reject it, even for purposes of discussion, a lot less variation in the anonymous expression of their private views. 3 do so at their peril. They are viewed as “difficult,” Deliberation sharply increased the disparities between the views of “not part of the team,” or, in extreme cases, as misfits. Boulder citizens and Colorado Springs citizens. Before deliberation, tC The group members in the examples above are, in many people’s opinions overlapped between the two cities. After a sense, entirely rational. They care about their repu- deliberation, group dynamics left liberals and conservatives much tations, but there’s nothing irrational about that. As more sharply divided. noted, however, people use heuristics, which can lead them astray, and are subject to biases. For the purposes of understanding how cascade effects A by-product of availability is “associative block- work, the most important heuristic involves avail- ing” or “collaborative fixation,” whereby strong ideas ability: A vivid idea or example moves rapidly from block the recollection of other information. This phe- No one person to another, eventually producing a wide- nomenon is a big problem when a group sets itself the spread belief within a group and possibly a city, a task of generating creative solutions. The innovative state, or even a nation. thinking of individual members is suppressed by the In the area of risk, availability cascades are com- powerful ideas generated by other members. mon. A particular event—involving a dangerous pes- In the actual world of group decision making, ticide, a hazardous waste dump, a nuclear power ac- of course, people may not know whether other cident, an act of terrorism—may become well-known members’ statements arise from independent in- to the group, even iconic. If so, it will alter members’ formation, an informational cascade, reputational perceptions of a process, a product, or an activity. pressures, or the availability heuristic. They often Do Availability cascades are familiar in business, too. overestimate the extent to which the views of others Reports of a success or a failure may spread like wild- are based on independent information. Confident fire within or across companies, leading to judgments (but wrong) group decisions are a result. about other, apparently similar events or products. If a movie (Star Wars?), a television show (The Walking Polarizing Groups Dead?), or a book (involving Harry Potter?) does well, Polarization is a frequent pattern with deliberating businesses will react strongly, eagerly looking for a groups. It has been found in hundreds of studies in proposal or a project that seems similar. more than a dozen countries. We found it in dramatic December 2014 Harvard Business Review 7 This document is authorized for educator review use only by Prabakar S, Other (University not listed) until Jun 2023. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860  MAKING DUMB GROUPS SMARTER t os form when we conducted an experiment in which residents of two Colorado cities discussed their po- Leaders can areas of the law, an even better predic- tor is who appointed the other judges litical beliefs (see the sidebar “A Tale of Two Cities”). refuse to take on the panel. rP a firm position The earliest experiments on the polarizing effects Why does group polarization occur? of deliberation involved risk-taking behavior, with There are three principal reasons: a clear finding that people who are initially inclined to take risks become still more so after they deliber- at the outset, The first and most important in- volves informational signals—but with ate with one another. (Examples of risky decisions include accepting a new job, investing in a foreign thus making a few twists. Group members pay at- tention to the arguments made by other country, escaping from a prisoner-of-war camp, and space for more group members. Arguments in any information running for political office.) On the basis of this evi- group with an initial predisposition will yo dence the conventional wisdom became that group inevitably be skewed in the direction of deliberation produced a systematic “risky shift.” Later studies called this conclusion into ques- to emerge. that predisposition. As a statistical mat- ter, the arguments favoring the initial tion—and created a puzzle. On many of the same position will be more numerous than issues on which Americans made a risky shift, those pointing in another direction. Taiwanese participants made a cautious shift. Even Individuals will have thought or heard among American participants, deliberation some- of some but not all the arguments that times produced cautious shifts. Cautious shifts emerge from group deliberation. Thus op took place most often in decisions about whether to deliberation will naturally lead people marry and whether to board a plane despite severe toward a more extreme point in line abdominal pain. with what they initially believed. What explains these unruly findings? As the psy- The second reason involves repu- chologists Serge Moscovici and Marisa Zavalloni tation again. As we have seen, people discovered decades ago, members of a deliberating want to be perceived favorably by tC group will move toward more-extreme points on the scale (measured by reference to the initial me- dian point). When members are initially disposed toward risk taking, a risky shift is likely. When they are initially disposed toward caution, a cautious shift is likely. A finding of special importance for business is that group polarization occurs for matters of fact as well as issues of value. Suppose people are asked No how likely it is, on a scale of zero to eight, that a prod- uct will sell a certain number of units in Europe in the next year. If the pre-deliberation median is five, the group judgment will tend to go up; if it’s three, the group judgment will tend to go down. Even federal judges—experts in the law and sup- posedly neutral—are susceptible to group polariza- tion. Research by one of us (Sunstein, along with David Schkade, Lisa Ellman, and Andres Sawicki) Do has found that both Democratic and Republican appointees show far more ideological voting pat- terns when sitting with other judges appointed by a president of the same party. If you want to know how an appellate judge will vote in an ideologically contested case, you might want to find out whether she was appointed by a Republican or a Democratic president. It’s a pretty good predictor. But in many 8 Harvard Business Review December 2014 This document is authorized for educator review use only by Prabakar S, Other (University not listed) until Jun 2023. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860  FOR ARTICLE REPRINTS CALL 800-988-0886 OR 617-783-7500, OR VISIT HBR.ORG t os other group members. Sometimes their publicly Negative information about the winner and positive stated views are a function of how they want to information about the losers (revealed to only one present themselves. Once they hear what others be- or two group members) did not reach the full group. rP lieve, they will adjust their positions at least slightly While many hidden-profile experiments involve in the direction of the dominant position in order to volunteers from college courses, similar results preserve their self-presentation. have been found among real-world managers. In a The third reason stresses the close links among study of hiring by high-level executives conducted three factors: confidence, extremism, and corrobo- by Susanne Abele, Garold Stasser, and Sandra I. ration by others. When people lack confidence, they Vaughan-Parsons, the experimenters did not con- tend to be moderate. The great American judge trol the information the executives had about the Learned Hand once said, “The spirit of liberty is the various candidates; instead, the executives did their yo spirit which is not too sure that it is right.” As people own information searches. As a result, some infor- gain confidence, they usually become more extreme mation was known to all, some was shared but not in their beliefs, because a significant moderating fac- by all, and some was held by only one person. tor—their own uncertainty about whether they are The finding? Common information had a dispro- right—has been eliminated. The agreement of oth- portionately large impact on discussions and conclu- ers tends to increase confidence and thus extremism. sions. The executives gave disproportionately little weight to valuable information held by one person Focusing on “What Everybody Knows” or a few, and as a result made bad decisions. op Our last group problem may be the most interesting The study also found that some group members of all. Suppose a group has a great deal of informa- are “cognitively central,” in that their knowledge is tion—enough to produce the unambiguously right held by many other group members, while other outcome if that information is elicited and properly group members are “cognitively peripheral,” in that aggregated. Even so, the group will not perform well their information is uniquely held. To function well, if its members emphasize broadly shared informa- groups need to take advantage of cognitively periph- tC tion while neglecting information that is held by one eral people. But in most groups, cognitively central or a few. Countless studies demonstrate that this re- people have a disproportionate influence on dis- grettable result is highly likely. cussion. A simple explanation for this is that group “Hidden profiles” is the technical term for accurate members prefer to hear information that is com- understandings that groups could but do not achieve. monly held—and prefer to hear people who have Hidden profiles are a product of the “common knowl- such information. Cognitively central people are edge effect,” whereby information held by all group thus accorded high levels of credibility, whereas cog- members has more influence on group judgments nitively peripheral people are accorded low levels. No than information held by only a few. The most obvi- ous explanation of the effect is that common knowl- Making Groups Wiser edge is more likely to be communicated to the group. A central goal in group decision making should be to But faulty informational signals play a big role as well. ensure that groups aggregate the information their Consider a study by Ross Hightower and Lutfus members actually have and don’t let faulty informa- Sayeed on how groups make personnel decisions. tional signals and reputational pressures get in the Résumés for three candidates for the position of mar- way. Here are six ways to achieve that goal, starting keting manager were placed before three group mem- with the simplest: bers. The experimenters rigged the résumés so that Silence the leader. Leaders often promote self- Do one applicant was clearly the best for the job. But each censorship by expressing their own views early, thus test subject was given a packet of information contain- discouraging disagreement. Leaders and high-status ing only a subset of attributes from the résumés. members can do groups a big service by indicating a Almost none of the deliberating groups made willingness and a desire to hear uniquely held infor- what would have been, if all the information were mation. They can also refuse to take a firm position considered, conspicuously the right choice. The win- at the outset and in that way make space for more ning candidates tended to be those about whom all information to emerge. Many studies have found three test subjects were given positive information. that members of low-status groups—including December 2014 Harvard Business Review 9 This document is authorized for educator review use only by Prabakar S, Other (University not listed) until Jun 2023. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860  MAKING DUMB GROUPS SMARTER FOR ARTICLE REPRINTS CALL 800-988-0886 OR 617-783-7500, OR VISIT HBR.ORG t os less-educated people, African-Americans, and wants to obtain the information that its members sometimes women—have less influence within de- hold, they should be told before deliberations begin liberating groups (and may self-silence). Leaders that each has a different and relevant role—or at least rP who model an open mind and ask for candid opin- distinctive information to contribute. ions can reduce this problem. Appoint a devil’s advocate. If hidden profiles “Prime” critical thinking. We have seen that and self-silencing are sources of group failure, a when people silence themselves in deliberating tempting approach is to ask some group members to groups, it is often out of a sense that they will be pun- act as devil’s advocates, urging a position that is con- ished for disclosing information that runs counter trary to the group’s inclination. Those who assume to the group’s inclination. But social norms are not that role can avoid the social pressure that comes set in stone. Social scientists have done a lot of work from rejecting the group’s dominant position, be- yo on the importance of “priming”—that is, triggering cause they have been charged with doing precisely some thought or association in such a way as to af- that. But be careful with this approach: Authentic fect people’s choices and behavior. In experiments dissent and a formal requirement of devil’s advo- on group decision making, engaging participants in a cacy are different; the latter does far less to improve prior task that involves either “getting along” or “crit- group performance, because members are aware ical thinking” has been shown to have a big impact. that it’s artificial—a kind of exercise or game. When people are given a “getting along” task, they Establish contrarian teams. Another method, shut up. When given a “critical thinking” task, they related to appointing a devil’s advocate but shown op are far more likely to disclose what they know. So if to be more effective, is “red teaming.” Red teams the leader of a group encourages information disclo- come in two basic forms: those that try to defeat sure from the beginning, even if it goes against the the primary team in a simulated mission, and those grain, members will probably do less self-silencing. that construct the strongest possible case against a Reward group success. People often keep si- proposal or a plan. Red teams are an excellent idea lent because they receive only a fraction of the ben- in many contexts, especially if they sincerely try tC efits of disclosure. Careful experiments have shown to find mistakes and exploit vulnerabilities and are that incentives can be restructured to reward group given clear incentives to do so. success—and hence to encourage the disclosure of The Delphi method. This approach, developed information. Cascades are far less likely when each at the RAND Corporation during the cold war, mixes individual knows that he has nothing to gain from the virtues of individual decision making with social a correct individual decision and everything to gain learning. Individuals offer first-round estimates (or from a correct group decision. The general lesson is votes) in complete anonymity. Then a cycle of re- that identification with the group’s success is more estimations (or repeated voting) occurs, with a re- No likely to ensure that people will say what they know, quirement that second-round estimates have to fall regardless of whether it fits “the party line.” (This, by within the middle quartiles (25%–75%) of the first the way, is one reason that prediction markets work round. This process is repeated—often interspersed and deserve careful attention.) with group discussion—until the participants con- Assign roles. To understand one especially verge on an estimate. A simple (and more easily ad- promising strategy, imagine a deliberating group ministered) alternative is a system in which ultimate consisting of people with specific roles that are judgments or votes are given anonymously but only known and appreciated by all members. One person after deliberation. Anonymity insulates group mem- might have medical expertise; another might be a bers from reputational pressures and thus reduces Do lawyer; a third might know about public relations; a the problem of self-silencing. fourth might be a statistician. In such a group, sen- Group failures often have disastrous conse- sible information aggregation would be far more quences—not merely for businesses, nonprofits, likely, simply because every member would know and governments, but for all those affected by them. that each of the others had something to contrib- The good news is that decades of empirical work, ute. Indeed, experiments have found that the bias alongside recent innovations, offer some practical in favor of shared information is reduced when test safeguards and correctives that can make groups a subjects are openly assigned specific roles. If a group lot wiser.  HBR Reprint R1412F 10 Harvard Business Review December 2014 This document is authorized for educator review use only by Prabakar S, Other (University not listed) until Jun 2023. Copying or posting is an infringement of copyright. [email protected] or 617.783.7860