Debate is distinct from discussion due to its adversarial nature—we control uniqueness and the internal link is linear, multiple factors predispose us to avoiding clash, so you should seek to guarantee it as much as possible—the mere presence of limits or concessionary ground lapse into confirmation bias at best and mutual ignorance at worst.

Let us begin with improving the quality of collective decisions. A long tradition of thought, including, in particular, the reflections of John Stuart Mill and Karl Popper, has highlighted what we can call the epistemic virtues of criticism. Several arguments have shown that to submit an idea to criticism constitutes one of the best means of testing its validity. This holds for practical ideas. A decision is more likely to be of a high quality – whether in factual and technical terms, or in terms of values – if the proposals for action have been submitted to criticism beforehand. Criticism permits the elimination, or at least the reduction, of proposals involving false factual beliefs, logical errors, or objectionable moral choices. We do not need to repeat here the arguments establishing the epistemic merits of criticism. No one today would deny that criticism is one of the best means at our disposal to test the quality, technical and moral, of practical proposals. Nevertheless, the conclusion that we generally draw from these arguments is that it is enough simply to establish the freedom to express criticism to produce its benefits. This is without doubt how Mill reasoned. We find an even more striking expression of this position in the famous free-speech dissent of Supreme Court Justice Oliver Wendell Holmes. Men, he wrote, will eventually realize “that the ultimate good desired is better reached by free trade in ideas – that the best test of truth is the power of the thought to get itself accepted in the competition of the market.”7 Yet the conclusion that the free exchange of ideas is a sufficient guarantee is not justified. Freedom of speech on its own does not ensure that the right to criticize will be exercised. Furthermore, as we shall see, the fact that criticisms are put forward does not guarantee that they will receive proper consideration. Several mechanisms can prevent the ability to criticize freely from leading to its exercise. I group these together here under the label of conformism. Social psychology has long told us (with elaboration in Rousseau) that people want to be liked. Being esteemed and approved of by others also provides a gratifying image of one's self. It follows that when people perceive that, in a given social environment, others' opinions lean in a certain direction, they tend to bend their own expressions in the same direction to gain the approval of their peers. They avoid expressing dissenting or critical views, reinforcing the mechanism of the “spiral of silence.”8 Even in the context of discussion, we observe the disposition to conform to what is perceived as the norm within the group. This phenomenon is at the heart of what has been named, since the works of psychologist Serge Moscovici, the polarizing effect of group discussions.9 Thus, even if criticism is formally free, a powerful social force works to marginalize, or even to stifle, its expression. Contrary to what liberal theorists often affirm, social or cultural diversity within a group does not suffice to assure a confrontation of opinions critically opposed to one another. Suppose, for example, a deliberating body whose members, while being diverse with regard to social position, education, and beliefs, also share a fear of some danger. Let us imagine, furthermore, that this assembly discusses a measure that would contribute to the reduction of this danger; for example, strengthening the powers of the police. In this context, it is unlikely that collective deliberation will bring forth many arguments against the adoption of this measure, however much social or cultural diversity there may be in the group. Rather, the discussion will produce an accumulation of reasons in favor of increased police power, with various members finding, from within their own particular perspectives, diverse reasons for adopting this course of action that others, differently situated, may not have seen on their own. Yet even if increasing the prerogatives of the police did, in fact, contribute to the realization of the desired end, the measure might also present undesirable effects or features in other ways. Collective deliberation should precisely bring to light these potential negative effects and weigh them in the balance against the benefits of the measure. But in our case, the assembly will systematically underestimate these possible negative features even though members of the assembly have the liberty to oppose the measure and criticize one another. It appears, then, that if one wants to obtain from political deliberation the favorable epistemic effects of criticism, the expression of opposing opinions must be encouraged, not merely permitted. But there is another reason to ensure that the participants in a deliberation are actually confronted with opposing points of view; it concerns the reception of arguments, rather than their production. Studies in social and cognitive psychology show that, confronted with new information or evidence, people have a systematic propensity to see in it a confirmation of their previous beliefs. In a now-classic experiment, psychologists presented the same ensemble of documents and studies concerning the death penalty and its effects to two groups of subjects selected on the basis of their antecedent opinions: one group composed of subjects favorable to the death penalty, the other subjects who are rather hostile to it. After being confronted with these documents, the group that was favorable to the death penalty became more favorable to it, and the group hostile to it became still more hostile.10 This phenomenon is particularly marked when the documents presented to the subjects were ambiguous and called for interpretation. The propensity to find support for one's antecedent beliefs is known as “confirmation bias.”11 Research has also shown that group discussion reinforces the effects of confirmation bias. Groups interpret information with more bias than do individuals; and they privilege information that supports their antecedent beliefs to an even greater degree than do individuals.12 Two mechanisms explain this tendency. First, as noted earlier, group settings accentuate tendencies that predominate among individuals. If privileging information supportive of prior beliefs is already the dominant tendency among individuals, it is not surprising that this tendency should be amplified in group discussion. But a second, and more surprising, mechanism is also at work. It seems that groups tend to discuss principally the information that was already known to all the members before the beginning of the discussion. Within groups, it turns out, discussion turns essentially on shared knowledge. Members of the group are reluctant to discuss those bits of information that are known only to one or a few other members prior to the discussion.13 Shared information appears to have more weight in the eyes of the group members, and has a greater chance of being mentioned during the discussion, and thus remembered later.14 Finally, information supporting the position preferred by the greatest number of group members has a greater likelihood of becoming the object of discussion than information supporting the contrary position.15 Collective discussion thus tends to produce a disproportionate volume of information and arguments in favor of the already-dominant belief in the group. If one wishes to check the effects of confirmation bias – a phenomenon to which groups are particularly vulnerable – one must take proactive measures. One can, for example, call special attention to arguments contrary to prior beliefs by highlighting them (literally) or by making them cognitively more salient. Not only is the free expression of a multiplicity of voices not sufficient to assure the confrontation of opposing views, but the mere expression of contrary arguments is not sufficient for others to understand those arguments or consider them objectively. In the absence of measures that actively induce individuals to pay particular attention to evidence and points of view opposed to their own, collective deliberation will have the greatest likelihood of simply reinforcing antecedent opinions. In a political deliberation, in short, we cannot expect that the gathering of diverse points of view will spontaneously produce a clash of arguments pro and contra, nor that it will bring about a balanced consideration of views. Mill was wrong to assume that, in a society or an assembly composed of diverse members, opposing opinions would already be there, waiting to be set against one another once they were allowed to be uttered. Mill wrote: “The most intolerant of churches, the Roman Catholic Church, even at the canonization of a saint, admits, and listens patiently to, a ‘devil's advocate.'”16 He failed to see that the presence of a devil's advocate was required, not merely admitted. And through the requirement, the Church secured that objections to the canonization of a given person were aired and considered, even if no individual would otherwise have spontaneously offered them.17 The confrontation of opposing opinions also has merit beyond eliciting unshared perspectives. It unifies the field in which opinions are formed and expressed, counteracting the fragmentation of the public sphere. In order to be opposed to an opinion and to contest it, it is necessary that one be cognizant of that opinion and take it into consideration. In a society in which points of view are objectively diverse, the open and explicit clash of opposing ideas is neither the natural state nor the sole possible condition. Another configuration is just as likely: mutual ignorance. The German sociologist Georg Simmel therefore argued that conflict between social groups paradoxically served the cause of social integration: first, by placing the conflicting groups into a relationship with one another and, second, by exerting a pressure for unity among the secondary divisions within each group. We can advance a similar argument in matters of opinion. The clash of opinions unifies the field in which beliefs confront one another, creating a space in which those beliefs are addressed to and respond to one another. This task of mutual addressing is harder when the space of opinions is fragmented into a multitude of islets, homogenous within themselves but formed in conditions of little communication with outsiders. Several factors – some older, some of more recent origin – now trend in the direction of this sort of fragmentation. First, we have long known that people are selective in their choice of contacts and social relations. They tend disproportionately to be in contact with people who share their political opinions.18 Psychologically, many fear the face-to-face expression of political disagreement and want to avoid it as much as possible. More recent factors also work in the direction of fragmentation: the development of cable television and its thematic stations, the spread of the Internet, and finally the movement toward residential and territorial segregation. Although the effects of these transformations are still difficult to estimate, they all present an analogous structure: people are now offered, in multiple ways, greater opportunities for communicating and coming into contact only with other individuals like themselves. Cable television and the rise of opinion-based television stations (a phenomenon currently more pronounced in the United States than in Europe) provide viewers with the possibility of receiving a high proportion of their information only from a channel to which they feel ideologically close. Worse still, cable TV allows individuals with little interest in politics to avoid political news altogether and watch only entertainment programs.19 For its part, the Internet has dramatically increased the number and types of people with whom one can enter into contact. But studies on the usage of the Web suggest that contacts and links are established primarily through personal affinities, and in particular through ideological affinities within the political domain. Progressive blogs and forums link to other progressive sites but not to conservative sites, and vice versa. From these islets and networks of like-minded individuals we can expect the increased effects of reinforcement and polarization, because, in general, interacting with people of similar beliefs pushes one more toward the extreme positions of the views common to the group.20 Finally, the movement toward residential segregation, which has already progressed in the United States and is at work today in Europe, further contributes to the fragmentation of the public space of communication. If it is true that opinions are strongly correlated with sociocultural and ethnic factors, then in a neighborhood whose inhabitants share the same sociocultural or ethnic profile, each person is likely, for the most part, to encounter neighbors who share the same opinions. A selective exposure to similar opinions emerges de facto. Faced with these forces of fragmentation, only intentional collective action can be expected to produce a degree of unification of the public political sphere. This second justification for the deliberate encouragement of adversarial political debates is particularly salient today. True, adversarial debate is by nature reductive. Faced with some political problem, the polity usually has a multitude of possible courses of action, not all of which will be mutually exclusive. Yet the reductive character of the adversarial method is also one of its merits. It simplifies complexity, making the choices easier to grasp. There is no doubt, for example, that the current economic situation in Europe and the United States calls for a range of measures that are more or less intermingled and complementary to one another. To obtain a synoptic view of these measures and their relations and to choose among them would require considerable cognitive effort. There are cognitive advantages to presenting the policy response as a choice between reducing public deficits now and maintaining or increasing these deficits in the short term to prevent further decline until the economy has regained its normal growth rate. As democrats, we cannot discount the value of such cognitive simplicity. Groups of experts may be able to deliberate without using the adversarial method. But if we want ordinary (or even well-informed) citizens to participate ably in collective deliberation, the simplification achieved by the adversary system is an almost indispensable instrument. The fourth principle in support of the adversarial method is the value of treating the minority with respect. No matter how conscientiously citizens deliberate, it is likely that disagreement will remain at the end of the process. Decisions will therefore be taken by the majority. The majority of people will get to live with the decision they desired; a minority of people will have to live under a decision they did not support. To be sure, the decision itself formed the minority: it did not exist per se before the vote. But the manner of conducting the deliberation before the vote entails consequences for the treatment of those who, after the vote, will make up the minority. If the deliberation has been conducted as a debate between opposed positions, with each camp presenting its reasons in favor of its position and criticizing those advanced by the opposition, two consequences follow. After a vote has been taken, the minority must obey the decision, but at least the reasons aiming to justify this decision will have been formulated and made public. The minority might still refuse to listen to these reasons seriously and in good faith, but it was given the chance to consider them. The minority members were therefore treated with the respect owed to autonomous beings. Once children have reached the age of autonomy, parents must justify the orders they give them. When they are not yet autonomous, children must obey orders simply because they are orders. So, too, if the minority members have not had the possibility of hearing the reasons for the decision they must obey against their wishes, they are placed in a situation of having to obey the order simply because it is an order, or simply because it obtained the most votes. I do not mean to imply that the members of the minority will consequently be more disposed to obey the decision. Sometimes justifications exacerbate the opposition. But justifying orders shows greater respect for the autonomy of those receiving them. On the flip side, the reasons for not taking the decision that ultimately triumphed would also have been put forward. These criticisms and objections did not prevail, but they were at least articulated and made public. From the majority's perspective, because it won, it will naturally think that it was right; but in the process, it had to listen to the opposition explaining their justification. The members of the majority were at minimum forced to see that there were reasons supporting the other side. It seems reasonable to think that, as a result, the majority will be less inclined to consider the minority as unintelligent or ill-intentioned. Before I proceed to the practical consequences that we can draw from my argument, I must first respond to an objection: that rendering obligatory the presentation of opposing points of view in the public sphere would require constraints on public discourse and encroachments on freedom of speech. In response to this objection I would first suggest turning to an institution that in the relatively recent past followed just this path: the fairness doctrine in effect in the United States from 1927 to 1987. The fairness doctrine, implemented by the Federal Communications Commission, made it obligatory for radio and television stations to give an evenly balanced presentation of “opposing viewpoints on controversial issues of public importance.” The fairness doctrine not only imposed equality in airtime; it also required the presentation of viewpoints opposed to one another. The doctrine did not apply to airtime during electoral campaigns, which was regulated on other terms. It applied instead to any question that became the object of public controversy outside of electoral periods. The constitutionality of this doctrine was upheld by the U.S. Supreme Court in the famous 1969 decision of Red Lion Broadcasting Co. v. FCC.21 The central argument that the Court invoked in support of the constitutionality of the fairness doctrine was that, in regard to liberty of expression on the airwaves, it is “the right of the viewing and listening public, and not the right of the broadcasters, which is paramount.” The Court thus held that listeners and viewers had the right to hear conflicting viewpoints in order to make up their mind on the issues: “Speech concerning public affairs is more than self-expression; it is the essence of self-government.”22 The fairness doctrine was abandoned for two reasons. First, the doctrine led radio and television stations to avoid controversial subjects for the sake of not exposing themselves to lawsuits claiming they had violated the law. Second, the question of what exactly constituted the opposition of one point of view against another became the subject of repeated litigation, and the fcc proved unable to reduce the insecurity and juridical uncertainty that arose on this front. Despite its eventual abandonment, however, the Red Lion decision shows that the obligation for the media to present conflicting viewpoints is compatible with a certain interpretation of freedom of speech in the public sphere, an interpretation that focuses on the rights of the receiving public. That the U.S. Supreme Court has since rejected this interpretation does not mean that the arguments advanced in Red Lion were objectively weak. They are, in any case, consistent with the claims of this essay. What, then, should we do in practice to foster the confrontation of opposing arguments in today's democracies? Without claiming to provide a complete and detailed response to this question, I will conclude by suggesting two concrete means for promoting the adversarial principle in politics. The first is a practice yet to be invented, which would be implemented outside of electoral periods. The second consists of reinforcing a practice already used in electoral campaigns. First, my suggestion for the future. Outside of electoral periods, civil-society actors (such as foundations or think tanks) could organize adversarial debates on subjects of public interest. These debates would not be regularly scheduled, but would be organized only when a question sparked significant interest from the public (as with such topics as nuclear energy, assisted suicide, or, in certain countries, the wearing of the hijab) or when a large number of citizens mobilized in favor of a cause. More generally, these public debates would not aim to replace any existing democratic practices (such as electoral campaigns or parliamentary debates), but would complement them. Neither the exposition of conflicting viewpoints nor communication across ideological divides can be made mandatory. This does not mean that it is useless to try to facilitate them. Indeed, the probability of being confronted with opposing points of view matters: it tends to make one's thoughts more anticipatory, careful, and subtle.23 In contemporary circumstances, this probability tends to diminish. The active promotion of adversarial debates aims to counteract this pernicious tendency. Given that the goal of these debates would be to further the formulation and diffusion of arguments for and against a given public decision, they should be guided by the following principle: speakers should defend or criticize a given policy or position only with reference to its own merits, and not in response to reasons external to the policy or position. The arguments advanced in these debates should concern the advantages or disadvantages – whether technical or moral – inherent in the decision. I call this the principle of relevant reasons. This principle has two implications: the first concerns simplifying the debate to one issue; the second concerns choosing the right participants. In order to encourage citizens to take account of and weigh the reasons for and against a given decision, each question that can be defined objectively and independent from other questions should be debated separately. Multidimensionality and the bundling of different questions undermine the coherence of the arguments. To be sure, at election time, the voter will vote for a candidate or party that has bundled questions without an objective connection between them. Such grouping may be desirable, because it permits negotiations between different strands of the party. Nevertheless, to understand the bundling and negotiation well enough to cast an informed vote, the voter needs to have thought through the different issues separately, aided by adversarial debate. It is probably too difficult to completely exclude nonrelevant reasons – that is, reasons not substantively linked to the policy in question – at the moment of organizing a deliberative debate. But the principle remains valid: nonrelevant arguments should be sidelined as much as possible. As a consequence, each debate should focus on a specific theme, rather than on platforms comprising multiple dimensions.