Deliberative Democracy

Deliberative democracy has a distinctly different justification than either competitive or participatory models. The term has roots in the writings of philosophers such as Jürgen Habermas and John Rawls. This model seeks to have members of society enter discussions as political equals, willing to present arguments that reasonable people might accept while remaining open to changing their views based on counter-arguments, where the “unforced force of the better argument” can ideally lead to consensus. In the ideal form, this deliberation would be spread broadly throughout society. The concept is dismissed by many as Utopian. While many sortition advocates use the term “deliberative democracy”, they often have no familiarity with this Utopian term of art in philosophy, and are using the everyday meaning of the words—representative democracy by everyday people, using discussion rather than just voting.

Some philosophically oriented “deliberative democrats”, such as Cristina Lafont, hold fast to the idealized broad society concept and roundly criticize empowered mini-publics as an unacceptable “shortcut” (Lafont 2020). She asks why citizens should “blindly defer” to deliberative juries that make decisions they may disagree with.

Eric Shoemaker notes that according to the logic of Lafont’s argument, “it is not true that minipublics in particular ask citizens to blindly defer. On this description ‘blind deference’ is the condition citizens find themselves in whenever they do not get their way in their democracy” (Shoemaker 2024b). Indeed, it impugns the notion of majority rule itself. While Lafont believes sortition bodies that shape or decide public policies would be illegitimate due to this “blind deferral”, James Fishkin, the inventor of deliberative polling, responds that:

To accept the result from such an institution, should one exist, is not to follow others “blindly” but on the basis of articulated reasons which have been weighed against other articulated reasons in a process that represents everyone and in which you can see your viewpoints and interests represented. If such institutions were to become commonplace, I believe they would acquire empirical (psychological and sociological) legitimacy as people came to understand them by participating in them. (Fishkin 2018, 196–197)

The key feature of sortition that makes it a workable means of supporting both participatory and deliberative democracy, which is also epistemically superior, is that it does not ask all citizens to participate in deliberation all the time, but for focused periods of time—sequentially—in a manner that facilitates informed, deliberative, and representative decision-making. As I will discuss in later chapters, it is possible to institutionalize sortition in such a manner that huge numbers of short-duration panels are used, each dealing with distinct policy matters, thus allowing nearly all citizens to directly participate in “ruling” at various points in time. I agree with Barber’s recommendation for a democracy where “all of the people can participate some of the time in some of the responsibilities of governing” (Barber 2009, 267).

Good Deliberation

Good deliberation is different from discussion, oratory, rhetoric, negotiation, persuasion, and debate. Effective deliberation is a challenging undertaking, which does not often arise spontaneously. According to deliberative democracy theorists, deliberative conversation is a cooperative venture seeking truth or the common good, rather than seeking to “win” or defeat an opponent. It is thus quite distinct from election-based debate.

Deliberation is not merely argumentation among people with different views. As Bernard Manin observed, it must be carefully structured. “Diversity of views is not a sufficient condition for deliberation because it may fail to bring opposing views into contact. It is the opposition of views and reasons that is necessary for deliberation, not just their diversity” (Manin 2005, 17). As a participant at the randomly selected 2009 Australian Citizens’ Parliament said of preferences, “you’ve kind of just got your own, because you don’t really get a chance to question your own points of view, it’s only when you’re confronted with another point of view that you can even deliberate [in your own mind]” (quoted in Felicetti, Gastil, Hartz-Karp, and Carson 2012, 18).

David Ryfe, a professor of media and journalism, wrote that “a good design of deliberative democracy should establish rules to maintain the theoretical aims, allow people to tell stories to make cultural meanings in addition to making cognitive sense, encourage leadership that facilitates deliberation, endeavor to relate the outcome of deliberation to the participants, and finally create environments to facilitate learning how to deliberate” (Ryfe 2005). The hundreds of citizens’ assemblies convened in recent years around the world have generally embraced this goal, with procedures intended to promote it, and appear to have mostly achieved it.¹

Ideally, deliberation should be broader than a simple two-sided pro and con argument. A problem with limiting debate to pro and con is that each side has a psychological incentive to hide from the decision-makers any information that is not beneficial to their side. As Elster points out, “If speakers on all sides of a question withhold relevant information, the quality of the debates and the decision can suffer” (Elster 2013, 70). Conversely, he notes, “Speakers who are genuinely motivated to reach a good decision rather than to have their proposal chosen should volunteer information even if it counts against the position they defend”. It is this aggregation of widely distributed information that Ober largely credits for Athens’s flourishing beyond non-democratic cities (Ober 2010). So, at some stage, deliberation should allow for the crafting of revised proposals that open the door to new, win–win alternatives.

A crucial element of useful deliberation is the marshaling of facts, rather than merely beliefs or opinions. Too much of what passes for deliberation consists of listing supposed, but unsubstantiated “facts” by opposing sides. While some facts are best gathered from a diverse range of individuals reporting their lived experiences, others must come from researchers, scientists and other “experts”. Even those facts known by some participants may not spontaneously surface in a deliberation, simply due to shyness or other reasons. Frank Fischer of Rutgers University notes in his book, Democracy and Expertise: Reorienting Policy Inquiry, “Citizens have a good deal of information, but much of it remains tacit, fragmented, and disconnected. To be useful, it has to be organized. Professional experts need methods and practices that can help them to do this” (Fischer 2009, 110).

Of course, many “facts” will inevitably be in dispute, or be indefinite probabilities, or subject to revision based on new research. To the greatest extent possible, opposing sides on some policy matter should agree on a set of commonly accepted facts, much the way lawyers stipulate to agreed-upon facts in a court case. This can ameliorate competing experts simply arguing past each other when presenting to a mini-public. Again, professional staff for a minipublic should facilitate the development of agreed-upon facts among experts who disagree on some policy under consideration.

Note that this is very different from having mini-public members stipulate to a set of “facts” they all believe. The number of “facts” that nearly everybody (except experts who have tested them) believes to be true, but are in fact indisputably false, is legion. In an article titled, “Propaganda, Misinformation, and the Epistemic Value of Democracy”, Étienne Brown observes that,

In fact, while it is true that misperceptions are difficult to correct, we should not conclude that all kinds of democratic deliberation are powerless against misinformation. Instead, studies more moderately suggest that certain types of deliberation will more effectively favor the correction of false beliefs than others. When citizens are exposed to information from an individual they do not perceive as ideologically opposed to their worldview, they may engage in epistemic deference, that is, accept this person’s speech as authoritative. (Brown 2018, 14)

At the end, however, once a final version of some proposal has been crafted,

it does make sense to switch gears and limit deliberation to presentations for or against, with a jury weighing the arguments. As I mentioned earlier in this book, the final approve-or-reject decision should be made by a different group of people than those who crafted the final proposal. In brief, the mere act of crafting a proposal makes those participants unfit to then judge their own handiwork, whether due to pride of authorship, information cascades or groupthink. The next section will examine the argumentative theory of reasoning, which suggests why this recommended procedure is important.

Argumentative Theory of Reasoning

In order to design good deliberative and decision-making processes, it is important to understand cognitive biases and other characteristics of human reasoning. In 2011, Hugo Mercier and Dan Sperber published a paper with a startling but highly influential new theory of how reasoning evolved in humans, and how deliberation plays out in the real world (Mercier and Sperber 2011). I mentioned their theory briefly in Chapter 7, but it deserves deeper explication. Their approach is based on evolutionary psychology. They showed how real-world evidence undercut the “common sense” notion that reasoning evolved to allow the individual to find the truth or think through the best course of action in a given circumstance. They noted that the well-documented “confirmation bias”, in which individuals have a powerful tendency to seek out and believe evidence that supports their existing beliefs and dismiss evidence that would tend to disprove those beliefs, suggests reasoning didn’t evolve to work in that solo situation. If it did, confirmation bias wouldn’t exist, and we would each quickly abandon old beliefs as soon as good evidence showed them to be in error. Countless experiments have shown human resistance to that logical rule (Mercier 2016). As I pointed out previously, we can notice this in others, but are generally blind to this tendency in ourselves. Confirmation bias works potently against finding the truth or optimal solution.

In addition, once we have expressed an opinion, we develop a psychological investment in it and feel a prideful compulsion to defend it, even when the weight of mounting evidence is against it. One implication of this is that group members without prior opinions on a matter may be better suited for assessing the validity of conflicting arguments, unlike politicians or their partisan supporters, who come pre-positioned. This does not mean people never change their minds, but that changing a deep belief once adopted is more difficult than we admit to ourselves.

Mercier and Sperber suggest that human reasoning evolved as a social tool for ferreting out flaws in arguments presented by others, and also as a tool for building convincing arguments to present to others in favor of our own existing beliefs (which others can do a better job of evaluating than we can ourselves). In other words, reasoning works best as a social process in which people evaluate other people’s conflicting arguments, rather than merely thinking things through within a single mind. We are much better at detecting fact-based or logical flaws in others’ arguments than in our own. Indeed, we are horrible at recognizing our own cognitive biases. In this social application, reasoning can help find the truth or an optimal solution to a problem. Decision-making is generally superior at the group level than at the individual level.

This understanding of the dynamics of actual human reasoning and cognitive biases requires an update or modification of the philosophy-based theory of deliberative democracy. We should abandon the insistence that we have interlocutors who are willing to fundamentally change their minds in response to arguments superior to their own. That is a fool’s errand. We need other open-minded observers to listen and assess the arguments. They are able to exercise “epistemic vigilance”, a forensic function that is essentially the opposite of confirmation bias.

This principle was enshrined in the reconstituted Athenian democracy after 403 BCE, when the authority to adopt new laws was transferred from the mass Assembly (ecclesia) to randomly selected legislative panels (nomothetai). Around ten presenters—an equal number of opponents and advocates of the draft law—were selected to make their case in front of the random panel of as many as a thousand. The panel listened, weighed the arguments, and then voted. They did not engage in active debate among themselves.

This deliberative process is comparable to the invention of the scientific method, in which the key is to avoid the natural instinct of simply seeking evidence to confirm a belief, but rather to seek means of testing to disprove a hypothesis. The reason the invention of the scientific method was one of the greatest inventions in the history of the human race is that it counters our natural tendency to seek confirmation of our hypotheses. A good deliberative process anticipates various cognitive biases, such as confirmation bias, and uses design features that overcome them. Taiwanese researcher, Kevin Chien-Chang Wu, argues that with specific, appropriate design details, “deliberative democracy is one of the best designs that could facilitate good public policy decision-making and bring about epistemic good based on Mercier and Sperber’s theory of reasoning” (Wu 2011). This is particularly important at the final yes/no decision-making step.

In the open-ended “deliberative” process of crafting a proposal, having some presenters who are invested in a particular view, and thus do exhibit confirmation bias, can actually be a benefit to the group. Mercier and Landemore, in their article “Reasoning is for Arguing: Understanding the Successes and Failures of Deliberation”, suggest that according to the argumentative theory of reasoning, this assures that different sides will marshal the strongest possible arguments, which others can then weigh (Mercier and Landemore 2012).

I had an eye-opening experience while serving on the Commerce Committee in the Vermont House of Representatives. Rather than following the usual procedure of taking testimony from one witness at a time, on one occasion, we had a panel of opposing lobbyists essentially conduct an organized argument in front of the committee for a bill on which none of us had prior stances. The lobbyists would call each other out, since they knew when an opponent was intentionally cherry-picking data or skirting an important issue, which none of us, less-informed committee members, could detect. When a question of fact was in contention, the committee could have our non-partisan staff from the Joint Fiscal Office check the research. We learned and comprehended more through that one brief deliberative session than we normally ever learned about any bill.

However, it is also essential to have presenters who are not under the influence of confirmation bias to bring forth a wider array of information, which neither side’s advocates deemed useful to their side. This requirement is rarely achieved in partisan political debate. In electoral politics, the tendency is for one side to over-promise the benefits of a particular policy, while the other side warns of its devastating consequences, with neither side mentioning important information that is irrelevant to their goal of winning the argument but may be vital to making a wise decision.

My contention is that both active discussion by one group and listening and weighing by a separate group are vital to good public policy decision-making. Both should be employed in public policy decision-making—in the appropriate phase of the overall process. The first phase with active give-and-take can aggregate diffuse knowledge and seek out win–win possibilities. But this process doesn’t facilitate forensic evaluation, is prone to groupthink, and can go off track due to deference to high-status or highly persuasive individuals who are locked into confirmation bias. Once a proposal is crafted, a separate jury needs to listen and weigh arguments before voting on final adoption.

Diversity

A key for bringing forth information and insights that might be ignored by confirmation-biased, motivated reasoners, and experts is to have broad diversity within the body that is crafting a proposal. Even if one were skeptical about the importance of having descriptively representative decision-making bodies, instead favoring a more skilled, or expert group—“wise and virtuous” as the Founders would say—diversity is still crucial. Group competence hinges on diversity, rather than a mere summing of individual participant competence.

Elections are often regarded as a way to get the best among us together to solve the problems of society. But this conception is, to say the least, problematic. Margaret Heffernan, a professor of management at the University of Bath, whose interest in productivity drew her to the work of the evolutionary biologist William Muir, gave a TED talk discussing his research (Heffernan 2015). Heffernan described how Muir studied hens, because it is so easy to measure their productivity (you simply count the eggs). To determine what might make hens more productive, Muir devised an experiment: he selected an average flock and let it alone for six generations, but also separated out the individually most productive chickens and put them together in a superflock. In each successive generation, he selected only the most productive for breeding. After six more generations had passed, he found that the first group, the average group, was doing well. The hens were all plump and fully feathered, and their egg production had increased dramatically. In the superflock, however, all but three hens were dead. The three had pecked the rest to death. The individually productive chickens had achieved their success by suppressing the productivity of the others.

While not directly applicable to human societies, it serves as a metaphor about the danger and unintended consequences of seeking out the best. As Heffernan observes, most organizations and some societies are run in accordance with this superflock model, on the assumption that success is achieved by picking superstars and giving them all the resources and power. The result has been the same as in Muir’s experiment: aggression, dysfunction, and waste.

Even if we naively believed that the US Congress was made up of the best and brightest, the homogeneity of the members is not only unrepresentative but also a liability in terms of group competence when making decisions. Legal scholar Cass Sunstein, in his book Why Societies Need Dissent (Sunstein 2003), explains how research consistently shows the importance of diversity to optimal decision-making. The observation that diversity trumps homogeneity in groups of problem-solvers is hardly startling: two heads are better than one only if the two are thinking somewhat differently from one another. But the importance of diversity goes far deeper.

As James Surowiecki notes in The Wisdom of Crowds, diverse groups outperform homogeneous ones in many situations, even when the homogeneous group is made up of experts. By definition, experts are those who have experience solving certain kinds of problems, but while that experience may be helpful it may also act like blinders (hence the adage that generals are always fighting a previous war of their experience). One is better off relying on a diversity, rather than a specialization, of experience and expertise; as Surowiecki observes, “if you put together a big enough and diverse enough group of people and ask them to ‘make decisions affecting the general interest’, that group’s decisions will, over time, be intellectually superior to the isolated individual, no matter how smart or well informed he is” (Surowiecki 2005, xvii).

Exactly what kind of diversity (e.g., cultural, age, cognitive style) is desirable depends on the task at hand. For certain tasks, both diversity and expertise may be irrelevant. For other tasks, a diversity of life experiences is highly relevant. Within the realm of public policymaking, seeking out any particular sort of diversity (or expertise) is futile, since it often cannot be known in advance what sort of diversity is important for a particular decision—leaving random selection as the optimal strategy.

Even a conservative such as Edmund Burke saw value in diversity within a legislature, with deliberators representing every relevant point of view. Awareness of the value of diversity dates back at least to Aristotle, who wrote that “for the many, of whom each individual is but an ordinary person, when they meet together may very likely be better than the few good, if regarded not individually but collectively, just as a feast to which many contribute is better than a dinner provided out of a single purse” (Aristotle [350 BCE] 2016, 76). Aristotle seems to foreshadow Wikipedia and “crowdsourcing”.

At the time of the debates over the ratification of the United States Constitution, the issue of diversity was raised implicitly in many arguments of the Anti-Federalists. Melancton Smith, a prominent merchant and major landowner in New York, argued in 1788 that good governance required that legislators not be restricted to members of the “natural aristocracy” or “first class”. To have the necessary knowledge to adopt wise public policies, the representative body must include the middle classes. In addition to the knowledge held by the wealthy and well-educated, the legislature “should also comprehend that kind of acquaintance with the common concerns and occupations of the people, which men of the middling class of life are in general much better competent to than those of the superior class” (quoted in Storing and Dry 2008). Including members of racial, religious, and cultural minority groups in deliberation has the potential to increase understanding and sympathy. This result is not automatic, however, and some research suggests there is also a potential for increased polarization. The details of how the deliberation is structured and conducted are crucial.

Diversity does not merely refer to familiar demographic parameters such as race, class, and sex that exactly mirror the community. Political scientist Joel Parker put it this way:

We don’t need representatives to ‘represent’ directly any particular set of static interests. That interests may in fact be shaped by the representative process, and not simply reflected, is ultimately an argument for the importance of resemblance to representation. We need representatives to practice good judgment, perhaps more than we need them to hold precisely the same preferences, in the same combinations, as the rest of us. Interests should arise from, rather than drive, this collective good judgment. (Parker 2011, 151)

Theorists such as Hélène Landemore have stressed “cognitive diversity” or problem-solving styles as well (Landemore 2012b). Cognitive diversity is not just a matter of different perspectives arising from different upbringings and life experiences, which may be associated with categories such as race, sex, or class, but differences in personality and even brain function—different ways of seeing, interpreting, and making predictions about the world. Landemore has written extensively about the epistemic benefits of maximizing cognitive diversity in political decision-making bodies. She contends that this can optimally be achieved through random selection of group members (Landemore 2012a). The ability of random selection to generate a far more representative body than elections is, she argues, self-evident, and also confirmed by all sorts of pilot sortition implementations in the form of citizens’ assemblies. This representativeness is not symbolic or token, but deep. It reflects not only the sex, race, class, age, ethnic makeup, values, life experiences, religious beliefs (or lack thereof), occupations, interests, and political philosophies present in the community, but also its range of cognitive styles, in a way that no elected chamber could ever match.

Diversity Can Trump Ability

The notion that diversity can trump even expertise contradicts the “common sense” assumption that groups of experts should outperform panels of non-experts. The “Delphi method”, named for the ancient Greek oracle and advanced by the RAND corporation at the beginning of the Cold War, was based on this assumption. Experts were polled about a particular issue, such as the effects of technological changes on future wars. Initial results were shared with the experts, who then had the chance to modify their previous responses in a series of iterative rounds.

The Delphi method had mixed results, sometimes accurate and sometimes far off the mark. By keeping the experts anonymous and isolated from each other, it was hoped they could avoid the groupthink dynamic that promotes agreement, regardless of whether the point of convergence is right or wrong. But confirmation bias among experts is still a problem. Also, the method is intended to predict what is likely to happen, not what we want to happen. Policy decisions are grounded in community values, which the method isn’t intended to tackle. So, at best, it can only advise and inform decision-makers. The Delphi method also suffers from initial assumptions of who qualifies as an “expert”. Relatively homogeneous experts often have gaping holes in their knowledge of matters outside their area of expertise that nevertheless are of key importance.

James Surowiecki explains this counter-intuitive reality:

Bringing new members into the organization, even if they are less experienced and less capable, actually makes the group smarter simply because what little the new members do know is not redundant with what everyone else knows. (Surowiecki 2005, 31)

Professors of business and finance, Lu Hong and Scott E. Page, published a paper in 2004 offering a theorem that shows the inherent weakness of the approach to problem-solving that seeks to find a group of “best” problem-solvers, who maximize individual abilities, and who inevitably also are less diverse than a group of randomly selected problem-solvers. “Ultimately, the gain in individual abilities is more than offset by the functional [cognitive] diversity of a group of randomly selected people. It is in this sense that we might say diversity trumps ability” (Hong and Page 2004). In Page’s subsequent book, The Difference: How the Power of Diversity Creates Better Groups, Firms, Schools, and Societies, he describes how he discovered this unexpected dynamic while running a computer model about problem-solving (Page 2008). He found that a group of random agents outperformed a group made up of the best individual performers in a problem-solving task. This didn’t make intuitive sense to him, and he at first assumed he had made some coding error in writing the computer program. But there was no error. The result kept repeating in scenario after scenario. With deeper examination, the benefit of diversity in solving difficult problems became clear. He bluntly named this the “Diversity Trumps Ability Theorem”.

There are, of course, limits to the benefits of diversity. To consistently outperform the group of “experts”, the groups in Page’s computer model couldn’t be too small, and the members of the diverse group also had to be reasonably smart. “These conditions all made intuitive sense. A collection of third graders probably has little to add to the study of global warming” (Page 2008, 137). Although the attempt to create a mathematical proof, which relies on specific conditions and assumptions, may be erroneous, and it might be an over-reach to apply this theorem to the real world—and indeed mathematicians have criticized the paper (Thompson 2014), there is real-world experience that supports their initially counter-intuitive conclusions. For this theorem to be applicable in the real world, there are many conditions that must be met. Two of the most challenging are the ability of the participants to effectively communicate with each other and to properly evaluate what they learn. This is where an understanding of group social dynamics and good process design comes in. The problem being tackled by the group can be difficult, but can’t be so arcane that only specially trained experts can understand or evaluate potential solutions.

As I stated earlier, and bears repeating, “experts” are those who have experience solving old problems, and this experience may be helpful or act like blinders in dealing with new problems. Since future political challenges are inherently unpredictable, we can never know in advance which sort of problem-solving styles, or experts, might be most beneficial to have in a deliberative body. Diversity allows us to cover all the bases.

Even when diversity does not introduce new and more varied information to a group, research has shown that merely by reducing homogeneity, better decision-making is possible. Research by social psychologists such as Katherine Phillips of Northwestern University and others indicates that it is not simply a matter of diverse members (outsiders, if you will) bringing new information and perspectives to the deliberation. They have found that “even when out-group members do not bring minority viewpoints to the table, the mere presence of social category diversity in task groups can fundamentally change the behavior of the social majority and enhance group performance” (Phillips, Kim-Jun, and Shim 2012, 257).

MIT professor Evan Apfelbaum and a team of fellow researchers conducted studies with experimental markets in Singapore and replicated them in the United States, in which participants could earn money through virtual trading of stocks. Homogeneous groups were far more likely to overvalue stocks and develop market bubbles. In a paper published in November 2014 in the Proceedings of the National Academy of Sciences, they wrote, “In our experiments, ethnic diversity leads all traders, whether of majority or minority ethnicity, to price more accurately and thwart bubbles”. Rather than minority members bringing new information to the fore, in these cases diversity promoted better individual decision-making in general. “It is clear that people work harder in diverse settings, socially and cognitively. So when it comes to creating the conditions that promote cold hard accuracy in decision-making, diversity may be an invaluable asset” (Massachusetts Institute of Technology 2014). The paper notes that “the evidence may inform public discussion on ethnic diversity: it may be beneficial not only for providing variety in perspectives and skills, but also because diversity facilitates friction that enhances deliberation and upends conformity” (Levine et al. 2014).

Cognitive biases are not merely an attribute of individual humans. There are also many less-studied collective cognitive biases. In the paper on diversity vs. homogeneity and stock trading, the authors wrote: “Economists and psychologists have cataloged numerous individual cognitive biases. However, we suggest that biases may stem not only from the limits of individual cognition, but also from the social context in which decisions are embedded. Homogeneity (or diversity) is not a feature of individuals, but of a collective: a team, a community, or a market”. The researchers in that study also note that “the presence of more than one ethnicity fosters greater scrutiny and more deliberate thinking, which can lead to better outcomes. Such friction, however, can cause conflict and complicate collective decisions. The challenge, then, is in establishing rules and institutions to address ethnic diversity and its effects. Without them, conflict can be destructive; with them, diversity can benefit the collective” (Levine et al. 2014). Whether these social cognitive biases and behaviors are innate or culturally created, they are pervasive and need to be anticipated in designing optimal deliberative procedures.

Mere diversity in the room is not enough. One design option that is being experimented with is the use of segmented deliberation spaces. It has been found that when only a few members of a marginalized minority are included, they may not be inclined to fully express concerns about a policy topic. Allowing identity groups to caucus separately may bring forth perspectives and information that the group would not otherwise benefit from. But there are also polarization risks from this procedure, and more research is needed. Among many cultures, if even 50 percent of those present are women, they might contribute less than ten percent of the speaking. In a group of all women, it turns out they have plenty to say, but cultural norms of deferral may trump this. It is beyond the scope of this book to examine the array of facilitation methods and group deliberation designs, but it is important to note that diversity alone is insufficient.

Groupthink

Irving Janis published the first in-depth articles on “groupthink” in the early 1970s, which included an analysis of decision-making dynamics during such events as the Kennedy administration’s Bay of Pigs fiasco (Janis 1971). Janis noted that the term “groupthink” (with a tip of the hat to George Orwell and his neologisms in his novel 1984, such as “goodthink” and “doublethink”) refers to the “development of group norms that bolster morale at the expense of critical thinking”. And arise when participants are

motivated to avoid being too harsh in their judgments of their leaders, or their colleagues’ ideas. They adopt a soft line of criticism, even in their own thinking. At their meetings, all the members are amiable and seek complete concurrence on every important issue, with no bickering or conflict to spoil the cozy, “we-feeling” atmosphere.

This sounds worryingly like the cultural norm that could be generated within a citizens’ assembly (depending on process and facilitation). The mere fact that the participants and facilitators feel good about the work that they have done cannot be taken at face value.

Janis proposed several procedures to minimize the risk of groupthink. He proposed nondirective group leadership, encouragement of dissent (rather than mere toleration), and breaking discussion into separate sub-groups, to see if they all come to similar conclusions (Janis 1982). Thankfully, all of these have become fairly standard in citizens’ assemblies, though some facilitators may elevate the search for consensus above the importance of dissent. Note that these desirable features are contrary to the fundamental character of political parties in electoral systems, which typically develop policy proposals with strong leaders and relatively like-minded partisans.

As mentioned previously, in pseudo-deliberative bodies, such as Congress, there is no interest whatsoever in adopting a good deliberative process. What passes for debate is actually performance and declaration of in-group solidarity in preparation for the next election, rather than an attempt to inform or learn from each other and find common ground. Disinterest in a good deliberative process is also typical in multi-party electoral systems. There is substantial negotiation between party leaders, centered around power. They negotiate over the formation of a coalition government, or on a particular policy—but this is almost never a deliberative process. Different countries, and different political parties within them, have a variety of internal processes, but these negotiations are not deliberative, and rarely, if ever, occur in the legislative chamber itself, among the “back benchers”, or ostensible “representatives”. The crafting of draft legislation is done within political parties, which gather people who think similarly, with a narrow diversity of opinion, forfeiting collective intelligence.

Although a good process can reduce the risk of groupthink, the search for consensus and community spirit in citizens’ assemblies, as well as any inaccuracy of a stratified sample in creating a fully representative group, means that groupthink is always a risk. Miriam Solomon, of Temple University, challenges the view that rational deliberation in a group necessarily improves decisions. In an article based on research in a business context, “Groupthink versus The Wisdom of Crowds: The Social Epistemology of Deliberation and Dissent”, she wrote: “Pressure to reach consensus can, in fact, lead to phenomena such as groupthink and to suppression of relevant data. Aggregation of individual decisions, rather than deliberation to a consensus, surprisingly, can produce better decisions than those of either group deliberation or individual expert judgment” (Solomon 2006, 26).

The Wisdom of Crowds

In a deliberative situation, bringing additional perspectives, information and analytical styles to the fore has the potential to enrich the deliberation and lead to better-informed decisions. Diversity can be helpful in this active deliberation mode. On the other hand, when a group is seeking to find the correct answer to a question (in situations where a correct answer exists), many experiments have shown that avoiding active give-and-take deliberation— keeping participants isolated from the assessments of other participants—is key to getting an optimal collective assessment. Thus, paradoxically, active deliberation can be both helpful and harmful, depending upon the specific task and circumstances.

I will now take up this less intuitive storyline, about the benefits of not engaging in give-and-take deliberation. In certain scenarios, this is the key to uncovering the wisdom of crowds. To explain this, it is helpful to briefly consider the discovery made by Francis Galton in 1906, as so wonderfully described by James Surowiecki in his book The Wisdom of Crowds (Surowiecki 2005). Galton was a Victorian scientific polymath who was a pioneer in many fields, including statistics, psychology and eugenics. In the field of meteorology, he devised the first weather map. Galton believed some people were born brilliant and most were not. In one famous experiment, he sought to prove how badly the average person was at assessing matters they knew little about. He utilized the betting slips of a guess-the-weight-of-an-ox game at a fair, the annual West of England Fat Stock and Poultry Exhibition. Around 800 fair-goers, most of whom had no special knowledge, had guessed what the weight of an ox on display would be after slaughter and dressing. While some guesses were wildly high, and others were wildly low, he was amazed to discover that the average of all of the guesses was 1,197 pounds, while the actual weight turned out to be 1,198 pounds!

There are several possible explanations for this, such as that the poor guessers varied randomly on either side of the mark, tending to cancel each other out, allowing the few experienced, skilled butchers the chance to shine through. But the same general result shows up for guessing the number of jellybeans in a jar (a task nobody has practiced skill for), and countless other similar tests in which there is a correct answer. The average of the group is generally better than the guess of even the most skilled among that group. A key factor in such wisdom-of-crowds scenarios is independence of the guessers. If those guessing hear the guesses of other people, that (poor-quality) information tends to influence their own guess, and the average tends to fall further from the mark. People discount their own initial estimate in favor of what they come to believe is a more reasonable guess. In these sorts of situations, the wisdom of the crowd is best discovered when individuals do not consult or communicate with each other.

The tendency of participants to latch onto the assessments of others is one cause of “groupthink”. There are multiple psychological reasons for this “herd” behavior. It can be uncomfortable to be at odds with the group. Psychologists have shown that most people (though not all) will adjust their assessments about such obvious factual questions as “which lines match” in a line-length matching game, and agree with the opinion stated by the rest of the group, even if the other members of the group are unambiguously wrong (having been secretly prompted by the experimenters to make outlandish decisions), just to avoid going it alone (Plumridge 2020). Of course, people are also likely to cede independent judgment and go along with assessments expressed by high-status individuals, leaders of one’s social/political/religious/ethnic group, or acknowledged experts.

Even setting all of these psychological factors aside, there are also fully rational reasons for going along with the group. People often fall back on the heuristic that the opinion that most people hold is more likely to be right. If you come across a crowd of people all looking up into the sky, aren’t you likely to look up there too?

An experiment that demonstrates such informational cascades that can lead reasonable people to wrong conclusions is described in Cass Sunstein’s important book, Why Societies Need Dissent. Imagine an urn filled with red and white balls in a specific ratio. The task is to guess whether there are more red or white balls. Each participant is allowed to withdraw one ball, examine it secretly, place it back in the urn, shake the urn to mix the balls up again, and declare their guess as to whether there are more red or white balls in the room full of other participants. It seems reasonable to assume that after enough people have done this, it will become obvious which color is more numerous.

But suppose that by chance the first two participants happen to draw red balls, even though there are more white balls in the urn. In this imagined scenario, the chances of this happening are less than 50/50 but still pretty good. Both participants have declared they think the urn contains more red than white. The third person secretly draws a white ball. Here is the crucial moment of an informational cascade. Remember, the participant doesn’t simply declare what color ball he or she drew, but makes an educated guess about which color is more numerous overall. This person may well still make the guess that there were more red than white balls, since all the evidence so far indicates a two-to-one ratio. The fourth person may also happen to draw a white ball, but they have heard the three previous participants declare that they think the urn is mostly red balls. So, just like the third participant, it is reasonable for this person to guess that the urn has mostly red balls. As the number of people guessing red mounts, any participant drawing a white ball will tend to discount this one small piece of personal knowledge in favor of the obvious overwhelming number of red balls he or she assumes everybody else is drawing. As Sunstein writes, “people cease relying, at a certain point, on their private information or opinions. They decide instead on the basis of signals conveyed by others” (Sunstein 2003, 55).

Market bubbles and all kinds of other human foibles can be traced to this dynamic. In certain circumstances, each participant would do better by making an independent assessment—making a guess based solely on personal knowledge (the color of the ball drawn)—without the influence of other people’s guesses. Surowiecki gives numerous examples in which independence of participant assessments is essential for optimum group decision-making through the wisdom of crowds. In these situations, if participants discuss a topic among themselves, they may get pulled in one direction or another (perhaps by one participant’s status, eloquence, or other irrelevant yet influential factor) and are likely to end up with a worse guess overall.

Polarization

There are other sorts of issues where discussion and active deliberation (including the sharing of perspectives, analytical styles, or private information known by only a few participants) are important for making good decisions. This is sometimes referred to as collective intelligence as distinct from the wisdom of crowds discussed above. When crafting a proposal or comparing a large number of options, such give-and-take procedures are essential. But Cass Sunstein and others have argued (and demonstrated) that active deliberation, with the give-and-take of information and arguments, in addition to the risk of groupthink, can instead lead to polarization. In some experiments, active deliberation has caused individuals who were originally closer to the entire group’s mid-point on some issue to move further toward the extremes of one side or the other. In other words, rather than discussion causing participants to come to a center of common ground, groups can polarize into more extreme opposing factions. But this is not inevitable.

James Fishkin of Stanford University found in his deliberative polls that this polarization dynamic does not occur. Perhaps this is because of the specific facilitation and “balanced information” his technique employs. The concerns that group discussion by participants would either result in deference to high-status participants or polarization into conflicting camps, taking more extreme positions, appear to have remedies. Fishkin found that neither fear was realized in any of his numerous deliberative polls, and in the first Chinese citizens’ assembly implementation there was a group movement away from the initial preferences of the higher status members. Fishkin credits this success to good design, with balanced presentations, moderated small group discussions, and secret “voting” in the exit poll.

The danger of polarization that Sunstein warned against may be the result of tribalism or promoted by poor process. Policy discussion, or what passes for “debate” in a group, may entail people lining up to give their reasons and arguments in favor or in opposition to some policy. The side with more adherents often ends up with more active speakers (and crowd applause or cheers), causing undecided people to swing in that direction due to a variety of psychological tendencies. Those who are already for or against some policy will tend to selectively believe speakers who echo their own views due to confirmation bias. In such scenarios, sub-group solidarity can indeed prompt polarization. A good deliberative process would ensure that conflicting sides get an equal number of alternating speakers.

However, due to conflicting values or other deep beliefs, some issues may inevitably lead to polarized positions. So, voting is often required. In some cases, it may be appropriate for a mini-public crafting a proposal to report out competing policies for a final jury to judge. Likewise, even with the best possible process in place, active deliberation may result in a groupthink outcome that another mini-public (or the population as a whole) would soundly reject. This danger of groupthink is one reason why a mini-public that drafts a proposal, tapping into collective intelligence, should not pass final judgment on its members’ own proposal. An independent mini-public, which does not engage in give-and-take deliberation but relies on the wisdom of crowds, with listening and weighing instead, needs to vote on final adoption.

Active or Independent Deliberation

I have been using the term “deliberation” to refer to the active give-and-take process. However, in English, the word “deliberation” also has a second meaning. Rather than requiring a give-and-take of arguments between people, deliberation can refer to an internal thought process, as in deliberating with oneself—an internal reflective process. Professor John Anthony Garry of Queen’s University in Northern Ireland and his colleagues, in describing a theoretical sortition system named “randomocracy”, refer to this as “imaginative deliberation” (Garry, Stevenson, and Stone 2015). But even if participants take no part in giving arguments, it is still essential that they be presented with competing views in order to develop informed preferences. As suggested by the argumentative theory of reasoning, it is our ability to detect flaws in other people’s competing arguments where we as humans do best.

While active give-and-take deliberation within a diverse representative group has the potential for finding consensual win–win policies in the common interest, this is certainly not assured. There is a possibility of the group dividing into polarized factions, on one hand, and there is also a significant danger of succumbing to information cascades and groupthink on the other. A consensus may be the result of a problematic groupthink dynamic that results in a policy, which no other mini-public, or the community at large, would find desirable.

As I proposed above, a good approach to this problem is to use the “check and balance” of a sequential process. One mini-public employs active deliberation (using collective intelligence) for crafting proposals, and an independent jury (using the wisdom of crowds and preventing groupthink) judges the product of the first mini-public.

Note that the function of lottery selection is actually somewhat different for a proposal-drafting citizens’ assembly than for a final adoption jury. The drafting phase is best tackled by somewhat smaller bodies to allow all-to-all communication and to use collective intelligence through active deliberation, which particularly benefits from diversity. However, since this body is not making any final decision about adopting their proposal, diversity is key, and the accuracy of their representativeness is less crucial, so an approximation may suffice. This initial body will need to serve for a substantial term (e.g. numerous weekends, perhaps for months). But the final adoption body, using independent assessment of pro and con arguments, can be substantially larger, since they do not need to engage in all-to-all communication, but only listen. This larger second body may serve for a very short term, but must be accurately representative to earn legitimacy. So for the first body, lotteries enhance diversity, and for the second, representativeness, and for both, protection against self-selection special interest, and corruption.