The deliberative democracy movement: from a theory of legitimacy to citizens’ assemblies

The deliberative democracy movement is the body of theory and practice holding that democratic decisions gain legitimacy when they result from reasoned public discussion among free and equal people, not only from counting votes. In theory it is associated with Jürgen Habermas’s discourse principle, that valid norms are those to which all affected persons could agree “as participants in rational discourse”.⁠¹ In practice it produced citizens’ juries, first developed by Ned Crosby in 1971,⁠² James Fishkin’s Deliberative Polling, conceived in 1988 and run more than 150 times in over 50 countries and jurisdictions,⁠³ and the citizens’ assemblies that governments now use and that the OECD has documented.⁠^5,6 Critics argue that deliberation can reproduce social inequality and that many citizens prefer it in an advisory role.⁠^7,8

In plain words

Most democracy works by counting. People vote, and the side with more votes wins. Deliberative democracy adds a question: did people actually talk it through first?

The idea came from philosophers. They argued that a decision is more legitimate when people have heard the reasons on all sides, weighed them, and reached a view they can defend to others.

Then people built ways to try it. In a citizens’ jury or a citizens’ assembly, a group of ordinary people, often picked at random, spends days learning about an issue, hearing from experts on different sides, and talking in small groups. At the end they make recommendations. In a Deliberative Poll, a random sample answers survey questions before and after a weekend of this kind of discussion, so researchers can see how informed discussion changes views.

The results are interesting. In one large U.S. experiment, people from both parties moved toward less extreme positions after deliberating.

Critics raise real concerns. Discussion can favor people who are confident, educated, or from the majority group. A few hundred people in a room are not the whole public. And many citizens want these groups to give advice, not make final decisions. This chapter lays out both sides.

Defined terms

• Deliberation. Discussion aimed at weighing reasons and evidence before a decision. See deliberation.
• Discourse principle. Habermas’s statement that “only those action norms are valid to which all possibly affected persons could agree as participants in rational discourse”.⁠¹
• Public sphere. In Habermas’s early account, “a realm of private individuals assembled into a public body” that discusses public affairs and forms public opinion.⁠¹
• Citizens’ jury. A small group of citizens who receive background information, deliberate, and produce recommendations.⁠²
• Deliberative Poll. A random sample surveyed before and after a structured deliberation with balanced materials, moderated small groups, and expert panels.⁠³ See deliberative poll.
• Citizens’ assembly. A randomly selected body that learns, deliberates, and recommends, often commissioned by a government. See citizens’ assembly and sortition.
• Affective polarization. Dislike and distrust of the opposing party, distinct from disagreement on policy.⁠⁴

Evidence

Theory: legitimacy through public reasoning

The Stanford Encyclopedia of Philosophy’s entry on Habermas summarizes his account of how law gains legitimacy: informal discussion in civil society shapes public opinion, which flows into the parliamentary system, where it is worked into laws that citizens can accept on the basis of reasons. In Between Facts and Norms (1992), Habermas argues that constitutional democracy and the rule of law depend on each other, and that discourse should be insulated from coercion and inequality so that only the “unforced force of the better argument” counts.⁠¹ This is a normative standard, not a description of how politics usually works. Its practical influence came from the question it raised: what would a setting look like in which ordinary people reasoned under fair conditions?

The idea that many people reasoning together can judge well is much older; Aristotle’s argument about the wisdom of the many is discussed in Athenian democracy.

History: citizens’ juries

The Jefferson Center states that the Citizens Jury was “invented by our founder Ned Crosby in 1971,” and describes its process as selecting participants who are “a diverse group of their peers,” providing background information, facilitating deliberation, and creating recommendations.⁠²

History: Deliberative Polling

Stanford’s Center for Deliberative Democracy describes Deliberative Polling, originated by James Fishkin in 1988. “A random, representative sample is first polled on the targeted issues.” Participants then gather, often for a weekend, receive balanced background materials that are also made public, develop questions in small groups with trained moderators, question competing experts and political leaders in plenary sessions, and answer the original questions again. The center states that the changes in opinion represent “the conclusions the public would reach, if people had opportunity to become more informed and more engaged by the issues,” and that the method has been used more than 150 times in over 50 countries and jurisdictions.⁠³ The center also notes that Fishkin holds the trademark on the name “Deliberative Polling” and that fees support its research.⁠³ The relation to ordinary polling is discussed in public opinion polling.

Research results: deliberation and polarization

Fishkin, Alice Siu, Larry Diamond, and Norman Bradburn reported in the American Political Science Review (2021) on “America in One Room,” a national field experiment in which more than 500 registered voters deliberated over an extended weekend on five major national issues, with a control group that took the same surveys without deliberating. They found that deliberators moved substantially toward less polarized policy positions and showed a large drop in affective polarization, results the authors note run counter to what much of the literature would predict.⁠⁴ The organizers of the experiment are also the leading advocates of the method, which readers should weigh, and the design measures short-term change.

Hélène Landemore and Hugo Mercier, in Análise Social (2012), address a central worry, the “law of group polarization” associated with Cass Sunstein, that discussion pushes like-minded groups to extremes. They argue that the argumentative theory of reasoning predicts epistemic benefits from discussion with others, and they reply to the objection that polarization undermines deliberation’s value.⁠⁹ The polarization worry and the deliberation evidence concern different settings: like-minded groups talking among themselves, versus mixed groups under structured, balanced conditions.

Current practice: institutionalization

The OECD published Innovative Citizen Participation and New Democratic Institutions: Catching the Deliberative Wave on June 10, 2020,⁠⁵ and in 2021 a policy paper, Eight Ways to Institutionalise Deliberative Democracy, which observes that many authorities are embedding deliberative processes into their structures rather than relying on one-off initiatives that depend on political will.⁠⁶ Specific examples, including their outcomes and failures, are covered in the case pages for Ireland, British Columbia, France, and Ostbelgien.

Competing interpretations

Deliberation and inequality. Tali Mendelberg and John Oleske, in Political Communication (2000), compared two school desegregation town meetings. The all-white meeting kept consensus and framed arguments as serving the common good while mainly advancing group interests; in the integrated meeting, language that white participants presented as universal was heard by minority participants as racist and self-interested.⁠⁷ Their study suggests that the norms of reasoned, “neutral” talk can carry group advantages. Advocates reply that trained moderation and balanced materials are designed to address exactly this, which is an empirical claim each process should test.

Deliberation versus participation. Diana Mutz, in Hearing the Other Side: Deliberative versus Participatory Democracy, argues, as Scott McClurg’s review summarizes, that the deliberative ideal often conflicts with participatory democracy when viewed from individual citizens and their everyday networks, and studies “cross-cutting political talk” as it occurs in real life.⁠¹⁰ The tension is practical: the qualities that make careful deliberation possible are not the same as those that bring large numbers of people to take part.

Rationalist bias. Peter Dahlgren, in Political Communication (2005), critiques deliberative democracy for its rationalist biases, especially regarding politics outside formal channels, and proposes “civic cultures” as an alternative frame.⁠¹¹

How much authority? Saskia Goldberg and André Bächtiger, in the British Journal of Political Science, ran a conjoint experiment with a representative sample of politically inactive German citizens. They found general skepticism toward strong forms of deliberative forums and a preference for forums limited to policy advice, working with existing institutions, and including safeguards such as large group sizes or clear majorities.⁠⁸ Advocates of empowered assemblies argue that advice without power invites the tokenism Sherry Arnstein described; see participation between elections.

Interpretation, this site’s reading. The movement’s strongest evidence is that structured, balanced discussion among a random sample can change views in informed directions. Its weakest point is the link from a small room to the whole public: who chose the agenda, who spoke, and why non-participants should accept the result. Those questions are about governance, not technique.

Application to this framework

These are design implications, not claims about the framework’s performance.

1. Priorities and deliberation are different tools. My Vote Forward collects priorities and voluntary opinions at scale; it does not convene random samples for structured deliberation. Its results should not be described as deliberative outcomes. See evaluation for methods.
2. Balanced information is a requirement. Deliberative Polling’s balanced, public briefing materials.⁠³ are a model for the explanations attached to proposals: sources for each side, published in advance.
3. Watch for whose language counts. The Mendelberg and Oleske findings⁠⁷ argue for reviewing question wording and option labels with people from different groups; see question design.
4. State the authority. Citizens prefer advisory forums with safeguards in at least one experiment⁠⁸ Every result the framework publishes should state that it is advisory and who, if anyone, has committed to respond.
5. Design against polarization. If the framework ever adds discussion, the evidence points toward mixed groups, moderation, and balanced materials rather than open comment threads among the like-minded.⁠^4,9 See polarization.

Hypothetical example. A city publishes priority results showing that public safety ranks first. It then convenes a randomly selected panel of 40 residents who hear from police, social services, and neighborhood groups over two weekends. The panel recommends a mix of street lighting and mental health response. The two outputs answer different questions: what participants rank first, and what a representative group concluded after learning. Publishing both, labeled clearly, is more useful than treating either as the public’s final word.

Unresolved questions

• Do opinion changes from a weekend of deliberation last, and do they spread to people who did not take part⁠⁴?
• Can moderation and balanced materials overcome the inequalities that Mendelberg and Oleske observed⁠⁷?
• What authority should a randomly selected body have, given citizens’ stated preference for advisory forms⁠⁸?
• Can deliberation and mass participation be combined, or is Mutz right that they pull against each other⁠¹⁰?