In recent years, the idea of a legislature composed of ordinary citizens has attracted renewed attention. Political theorists and philosophers have advanced normative reasons for devolving legislative powers traditionally reserved for elected officials to randomly selected citizens. These include fairness and equality (Stone Reference Stone2016), protection against corruption (Dowlen Reference Dowlen2015; Guerrero Reference Guerrero2014), new forms of representation (Sintomer Reference Sintomer2023, 252–59), or even policy-making and problem-solving performance that is superior to that of elected assemblies (Landemore Reference Landemore2012). Although many scholars favor hybrid, bicameral models (Abizadeh Reference Abizadeh2021; Gastil and Wright Reference Gastil and Wright2018; Vandamme and Verret-Hamelin Reference Vandamme and Verret-Hamelin2017), others have advocated pure forms of lot-based democracy (Bouricius Reference Bouricius2013; Guerrero Reference Guerrero2014; Reference Guerrero2024; Landemore Reference Landemore2020).
All such arguments rest on the assumption that ordinary citizens can competently engage in policy making and lawmaking. Yet a long tradition in political science from Le Bon, Lippmann, and Schumpeter onward has entrenched skepticism about this assumption.Footnote 1 Later studies have documented, among ordinary citizens, widespread apathy (Almond and Verba Reference Almond and Verba1963; Finifter Reference Finifter1970), ignorance (Converse Reference Converse2006; Kuklinski et al. Reference Kuklinski, Quirk, Jerit, Schwieder and Rich2000; Somin Reference Somin2016), and tribalism (Campbell et al. Reference Campbell, Converse, Miller and Stokes1960; Huddy, Mason, and Aarøe Reference Huddy, Mason and Aarøe2015; Iyengar, Sood, and Lelkes Reference Iyengar, Sood and Lelkes2012; see also Achen and Bartels Reference Achen and Bartels2017; Brennan Reference Brennan2016; Caplan Reference Caplan2011).
Recent empirical studies of deliberative settings, however, have begun to validate the trust that deliberative democrats place in citizens’ capacities. When given proper incentives and opportunities, ordinary citizens have been shown to be capable of developing informed and reasoned judgments (e.g., Esterling, Fung, and Lee Reference Esterling, Fung and Lee2021; Fishkin and Luskin Reference Fishkin and Luskin2005; Gerber et al. Reference Gerber, Bächtiger, Shikano, Reber and Rohr2018).
Another serious concern in the literature is the risk of undue influence or capture—either by organized interests outside the assembly or by experts within (Bagg Reference Bagg2023; Landa and Pevnick Reference Landa and Pevnick2025). Critics fear that citizens might prove more vulnerable to such pressures than professional politicians.
So far this worry remains mostly theoretical. Mini-publics have rarely been empowered enough to attract serious capture efforts. Most assemblies in the “deliberative wave” of the last 20 years (OECD 2020) have been purely consultative and tasked with generating broad policy recommendations—such as promoting renewable energies or revisiting abortion laws—rather than binding legislative work. The 2004 British Columbia Citizens’Assembly on Electoral Reform came closest to drafting a legislative proposal, recommending proportional representation over the established first-past-the-post system (Warren and Pearse Reference Warren, Warren and Pearse2009). Yet its proposal merely informed the provincial legislature and remained subnational in scope. Subsequent national citizens’ assemblies, such as Ireland’s assemblies on marriage equality (2012) and abortion (2016), had more limited mandates still, producing general recommendations rather than specific legal texts.
In 2019, however, the French government convened 150 randomly selected citizens as part of the Citizen Convention for Climate (CCC) to deliberate on socially just ways to curb CO₂ emissions. Although it followed precedents set by Canada and Ireland, the French experiment also broke new ground by placing citizens in the position of co-constructing legislation with the government.
This article’s first and central contribution is interpretive: to clarify what was at stake in this first French citizens’ assembly. Although some commentators have recognized the CCC as an outlier for its size, scope, involvement of experts, and degree of politicization (Smith Reference Smith2022), we claim that the CCC members acted as de facto legislators. We use the term “de facto” as opposed to “de jure,” because this legislative power was grounded in controversial legal grounds and is not recognized in the French constitution. Our second contribution is to put forward the concept of “citizen legislators” as a subcategory of Mark Warren’s (Reference Warren, Warren and Pearse2009) “citizen representatives” to characterize the role of the participants in the CCC. Third, we propose criteria for assessing legislative proposals. Finally, we identify a normative principle—“citizens on top, experts on tap”—as having structured the relationship between citizens and experts throughout the CCC, protecting the former from capture by the latter.
The article proceeds as follows. We first present our research method and then make the case for our interpretation of CCC participants as de facto “citizen legislators.” The following section assesses the quality of their proposals, suggesting that their work disproved the classic objection of citizen incompetence. Next we show that citizens resisted both undue influence and capture; we refine these claims by analyzing expert roles in the Legal and Governance Committees in which the risk of domination remained high. We conclude by defending the French case as a proof of concept for involving ordinary citizens qua legislators in more inclusive forms of democracy.
Research Method
To make our case, we deployed a multimethod approach combining normative political theory and empirical work. Our approach was grounded in seven weekends of immersive fieldwork with the CCC, which were distributed over the roughly nine months of its duration. During these weekends, we attended plenary and small-group sessions. Landemore conducted 14 open-ended citizen interviews during the CCC. We also drew insights from 16 structured interviews that were conducted after the CCC ended by 51 other involved researchers (including Landemore). Altogether, we engaged directly, including through informal exchanges, with at least half of the 150 citizen participants; most of the facilitators and key experts; and at least three members of the Governance Committee, including one who served as a co-chair of the committee. This sustained presence and contact with participants allowed us to develop a rich, context-sensitive understanding of the CCC’s internal deliberative dynamics and how participants gave meaning to the process.
To ensure validity, we drew on best practices in interpretive research design (Bevir and Rhodes Reference Bevir, Rhodes and Bevir2011; Schwartz-Shea and Yanow Reference Schwartz-Shea and Yanow2013). We triangulated across observations, interviews, and documentary sources. We checked our intepretations through ongoing dialogue with participants, organizers, and other researchers and grounded our normative claims in rich descriptions of specific empirical realities. We also practiced reflexivity, interrogating how our own position as researchers shaped our interpretation of facts and behaviors. Finally, we documented our research protocols and reasoning in line with established norms in qualitative and comparative research (Della Porta and Keating Reference Della Porta and Keating2008; Ragin Reference Ragin1987). Our work thus sits within a pluralist methodological tradition that values theoretical insight developed inductively from deep empirical engagement (see appendix A for more details about our methods).
Despite this important empirical component, however, we did not consider our work to be comparative politics or ethnography. Our goal was not to identify causal mechanisms so that we could answer a precisely formulated “Why” question. Instead, we characterized our approach as “political theory in an ethnographic key” (Herzog and Zacka Reference Herzog and Zacka2019; Longo and Zacka Reference Longo and Zacka2019) or “inductive political theory” (Landemore Reference Landemore2020, 20–21; Reference Landemore2021; Reference Landemore, Baderin and Miller2026). We approached our interviews as political theorists interested in exploring normative questions—for example, What role should citizens occupy in a modern democracy? What are they capable of?—and in making sense of a promising new democratic experiment. In doing so, we followed in the footsteps of predecessors like Tocqueville (Reference de Tocqueville1835), Arendt (Reference Arendt1963), and Mansbridge (Reference Mansbridge1980), who also derived new normative insights from the observation of unprecedented or understudied political processes. Our work is firmly rooted in this tradition of empirically grounded political theory.
This kind of political theory often requires letting go of prior theoretical frameworks and hypotheses and reconfiguring research questions in the moment. Indeed, we arrived at the CCC site in October 2019 with a preexisting theoretical framework anchored in deliberative democracy. This framework led us to pay attention initially to the ways some of the predictions made by theorists in these traditions—for instance, about citizens’ deliberations and the collective intelligence of diverse groups—did or did not materialize. However, observing the CCC prompted us to go beyond these preexisting theoretical expectations and develop a bolder interpretation of what we witnessed than we could have anticipated. Our theoretical expectation was that citizens would set a general agenda on climate for legislators to follow up on. But that is not what we found. Another interpretation slowly emerged from observations and interviews, even those geared toward other hypotheses: that citizens were acting as legislators themselves.
CCC Participants as de facto “Citizen Legislators”
The French CCC was the culmination of a process that began with the massive social backlash—the so-called Yellow Vests movement—against a carbon tax the French government tried to enact in 2018.Footnote 2 As part of his response to this backlash, President Macron decided to convene, via the Office of the Prime Minister, a sample of ordinary citizens tasked with accomplishing what his government could not: creating a plan to reduce greenhouse gas emissions that would not trigger mass protests.
Over the summer and early fall of 2019, 150 citizens were randomly selected from throughout France, including its overseas territories, on the basis of six demographic criteria: age, gender, education level, geographic origin, settlement (urban versus rural), and employment.
In October 2019, the 150 met for the first time in the Iena Palace in Paris, the seat of the Economic, Social and Environmental Council (CESE), a consultative assembly sometimes referred to as France’s third legislative chamber. The CESE was put in charge of hosting and managing the logistics of the convention under the direction of an independent Governance Committee, whose 15 members included 7 CESE staff.
The 150 citizens met over seven weekends between October 2019 and June 2020. Among the 149 proposals that came out of the process were the recognition of the crime of “ecocide” (or environmental murder); an ambitious proposal to make energy-efficient housing renovations mandatory by 2030; a ban on publicity for heavily polluting goods; the promotion of vegetarian meals in public schools; constraints on soil artificialization; and a proposal to add a paragraph to the first article of the French constitution specifying that the French Republic “guarantees the preservation of biodiversity and the environment, and fights against climate change.” After the CCC ended, its proposals were put through a lengthy process of debates and negotiations with lobbies in both ministers and parliament, which ultimately resulted in the 2021 Climate and Resilience Bill. The Senate, however, vetoed the citizens’ request for a referendum on their proposed change to the constitution. Because the process was widely publicized and is now well documented, we refer the reader to these publications (Giraudet et al. Reference Giraudet2022; Landemore and Fourniau Reference Landemore and Fourniau2022) and appendix B for more details.
We argue that the CCC is the first example, in the modern age, of a contemporary assembly of “citizen legislators”—that is, ordinary citizens entrusted with formulating legislative proposals—at least at the national level and in a large, multicultural democracy. This interpretive claim builds on several pieces of evidence.
A first piece of evidence is the (in)famous promise of “no filter” by President Macron early in the process. In the April 2019 press conference where he announced the CCC, he said, “What will come out of this convention, I promise, will be submitted without filters to either a vote in Parliament, a referendum, or applied via executive actions” (our translation). This promise implied that at least some of the proposals needed to be readily implementable quasi-bills. How else could they be immediately put to a vote in parliament or a referendum or be turned into regulations? The participants were thus asked to be competent not just on the substance of climate change and social justice but also on the best way to translate their ideas into legal language. In other words, they were seemingly asked to operate as if they were lawmakers.
This promise was, of course, judged extremely controversial and impossible to keep. When meeting the CCC members for the first time in October 2019, the prime minister had to clarify that there were constitutional filters (aka, the government and parliament) that could not be bypassed. Most academic studies and much of the media since then have focused on the discrepancy between the presidential promise of “no-filter” and the actual delivery of proposals (Courant Reference Courant2021; Duvic-Paoli Reference Duvic-Paoli2022; Galván Labrador and Zografos Reference Galván Labrador and Zografos2023), implying the president’s betrayal of the CCC. We focus instead on the empowering effect of that promise on the CCC members themselves. Even though the promise was bold and likely to disappoint, it had the effect of legitimizing and endowing the CCC participants with de facto legislative authority.
The second piece of evidence comes from the CCC’s official mandate. According to the prime minister’s mission letter, it was “to define structuring measures to successfully reduce, in a spirit of social justice, greenhouse gas emissions by 40% of their 1990s levels by 2030.” The letter also included the expectation that, as a final product, a report would be produced containing conclusions and “the set of legislative and regulatory measures that it will have judged necessary to attain its objective,” including those “it will judge opportune to submit to a referendum.”Footnote 3 Unlike prior and subsequent citizens’ assemblies, the CCC’s mandate was not simply to generate an advisory and general policy recommendation: instead, it was expected to create a set of legal proposals and regulations accompanied by readily implementable policies with experts’ assistance.
To help citizens achieve this ambitious goal, the Governance Committee put a Legal Committee at their disposal. According to a document issued by the Governance Committee, the Legal Committee’s role was to “conduct the work of legally transcribing the measures, and later their amendments, put forward by the members of the Citizens’ Convention.” The document further specified the role of legal experts as follows: “to assist the Convention in the transcription of its proposals into legislative or regulatory measures, or at the very least in terms that make them as close as possible to a bill or a regulation.”Footnote 4
A third key piece of evidence for viewing the CCC as a de facto legislature involves the way the citizens conducted their work. Over the months, they moved through stages of agenda setting, expert consultation, drafting, and voting—closely mirroring a parliamentary process. The amendment procedure that took place between session six and seven was directly inspired by parliamentary procedures (see appendix C). These similarities led the copresident of the CCC Governance Committee Thierry Pech (Reference Pech2021) to title his book about the CCC, A Parliament of Citizens—this despite his seeing the work of citizens as “pre-legislative” because he acknowledged and normatively endorsed the de jure limits on their powers.
Some of the CCC’s work also resembled the activity of parliamentary commissions, which interview experts and develop specific proposals from scratch (Buge Reference Buge2022). The random distribution of 150 citizens from the CCC’s outset among five thematic working subgroups corresponding to five major sources of greenhouse gas emissions—Housing, Production and Labor, Transportation, Food, and Consumption—echoed the division of labor between parliamentary commissions. Yet French parliamentary commissions do not typically work in such close and sustained contact with experts. The CCC had an entire legal team and a technical group of experts at its disposal on location throughout the entire process. In that sense it was perhaps closer to the more privileged situation of US Congress members who, unlike their French counterparts, can rely on the nonpartisan expertise of the Congressional Budget Office, at least regarding budget-related issues.
In the French constitutional organizational chart, the CCC best matched the legislative role of the Office of the Prime Minister. Most legislative proposals originate from the executive in the French semi-presidential system, rather than from the parliament, as so-called law projects. The prime minister thus arguably outsourced the governmental prerogative of initiating laws to the convention members, enabling the CCC to act as an extension of executive power both in its policy-making function and in its legislative power of originating “law projects.” This interpretation helps explain the suspicion that the National Assembly members felt toward the CCC, which some parliamentarians saw as yet another way for the executive to undermine their authority (Buge and Vandamme Reference Buge and Vandamme2023).
The fourth piece of evidence supporting our interpretation relates to the recurring disagreement among Governance Committee members about the exact nature of the CCC. There was first a disagreement about how politicized the convention should be. Should it be a safe deliberative space protected from outside influences (as is still the dominant model in Ireland, for example) and more akin in that sense to a focus group? Or should it allow its members to take a position in the public debate? In the end the latter view prevailed. Citizens were encouraged to engage on social media and hold meetings with their local communities and stakeholders, including elected officials and lobbyists. During the COVID-19 crisis, the Governance Committee allowed them to write a collective letter to the public. Most observers underscored the “strong political framing of the CCC,” which spread “a real sense of responsibility and ownership for the outcome of the process” (Cherry et al. Reference Cherry, Capstick, Demski, Mellier, Stone and Verfuerth2021, 51), in contrast with other climate assemblies to date, such as Climate Assembly UK. Members of the CESE who were on the Governance Committee, however, did not embrace this politicized role and resisted it at various junctures.
One of those moments was a heated debate around “authorship” of the legislative proposals that came out of the CCC. Toward the end of the process, the Legal Committee assisting the citizens translated their proposals into legal language to render them “application ready,” as per the president’s request and the prime minister’s mission letter. This intervention raised a procedural question: Should citizens be allowed to vote again on the reformulated texts and so appropriate them as their own?
The CESE members on the Governance Committee argued that it was not legitimate for citizens to be given an opportunity to vote on the legal transcriptions. In their view, the authorship of the legal translations needed to stay with the lawyers.Footnote 5 By contrast, the non-CESE members of the Governance Committee wanted the citizens to be given the opportunity to appropriate the legal translations as their own through deliberation and a vote.
In the end, a compromise was struck. The Governance Committee organized a series of webinars (informally called Session Six Ter; see appendix B), one of which allowed the citizens in each thematic group to discuss the transcriptions with the Legal Committee and reject any wording that seemed to betray their intentions (provided that at least 15 members voted against): 68 citizens participated, and in the end 5 of the 95 transcriptions were rejected (Reber Reference Reber2020, 413–17). Additionally, during the final CCC session, the citizens voted on their original proposals and their legal transcriptions as a package, thereby arguably transferring authorship over the legal transcriptions to the citizens as well. Locating authorship of the legal proposals with the citizens arguably allowed the work of the convention to be recognized in the explanatory memorandum of the 2021 Climate and Resilience Law, which explicitly identifies the 150 citizens as “associated with” the legislative process and the “spirit” of the law.Footnote 6
A fifth and final piece of evidence for our interpretation draws on the proceedings of the next two French citizens’ conventions. One of the Governance Committee meetings of the second convention (on end-of-life issues) directly discussed the goals of this convening. During the meeting, a senior CESE representative remarked, “The prime minister asked a clear question. It demands a simple answer.… We are not going to write a law. We already tried that, with all the frustrations that this caused. We are going to give a political answer” (emphasis added).Footnote 7
This comment makes the most sense if interpreted in reference to the experience of the first CCC, which the representative perceived as a frustrating attempt to write new laws. Similary, in 2025, on the first day of the third French citizens’ convention (on school reform), the president of its Governance Committee told the convention members: “You are not going to write the law,” reiterating the message that the first CCC had erred in trying to do so.
Definition of the Term “Citizen Legislator”
We offer the descriptive category of “citizen legislator” as a refinement of a category already used in the literature, namely “citizen representative.” Coined by Mark Warren (Reference Warren, Warren and Pearse2009), the term “citizen representative” captures the role of nonelected laypeople acting or speaking on behalf of other laypeople.Footnote 8 Warren’s conception emerged from his analysis of deliberative democratic innovations like citizens’ assemblies, which are based on random selection, which should be regarded not so much as participatory institutions as representative ones, characterized by formal authorization, inclusive selection procedures, and discursive accountability. In these settings, ordinary citizens are empowered to speak and even act, in a limited but real sense, on behalf of other citizens, even though they are not professional politicians. Warren argued that these lay-citizens serve a temporary representative function by directly voicing the perspectives and experiences of other citizens. His work has paved the way for an exploration of the nature of democratic representation beyond elections, whether with a focus on formal, randomly selected citizen representatives (e.g., Landemore Reference Landemore2020) or informal, self-selected ones (Salkin Reference Salkin2024).
By citizen legislators, we mean a category of citizens who are neither professional politicians nor part of an elite group that would make them demographically distinct from the rest of the population. Although Warren implies that being elected disqualifies someone from being a “citizen representative,” we believe that in contexts where elections require neither professionalization, nor the ability to raise massive amounts of money, nor the need to be a notable or celebrity, the label of “citizen legislators” could also apply to elected citizens. Our label is thus compatible with the use of the term in Rosenthal (Reference Rosenthal2006)’s analysis of state legislatures before their professionalization in the 1970s, when they still included firefighters, farmers, small business owners, and other people for whom politics was not their primary activity and source of income. It would also apply to the Icelandic constitutional council composed of ordinary citizens who, although elected, had no prior political experience (Landemore Reference Landemore2015).
But the term would not apply to assemblies of notables and lawyers in eighteenth-century France, even though the members were not full-time politicians nor paid a living wage. Conversely, groups of ordinary citizens who get together to launch a popular initiative in countries that have this mechanism, like Switzerland, could also be characterized as “citizen legislators.” But corporate-funded professional activists paid to gather signatures should probably not be. The clearest case of a citizen legislature, on our view, is a randomly selected assembly, even when that assembly includes the occasional city mayor or member of parliament, and the term “citizen legislatures” can apply to departures from that ideal only up to a point.
By citizen legislators, we mean a category of citizen representatives whose function is, at a minimum, to formulate legislative proposals and, more broadly, to draft laws. The term “legislator” refers to the specific task that assembly members were given in the French case—a task that went beyond the broad agenda setting or advisory policy role that most citizens’ assemblies are confined to but was not strictly equivalent to the work of elected parliamentarians. Fulfilling a legislative function comparable to that of both parliamentarians and ministers, the citizens in the CCC were de facto, if not quite de jure, given the power to initiate the law.
By de facto, finally, we refer to the directness of the expected connection between the CCC’s decisions and policy and legal outcomes, which was the product of both an empowering government mandate and President Macron’s promise of “no filters.” Note that we say “expected directness” between the CCC’s decision and outcomes because de facto legislative proposals can still be thwarted given the constitutional filters that are still in place, which, in the case of the Climate Convention, did interfere with the final outcomes. Additionally, even after the convention members voted on their proposals, the de jure authorization could only come from constitutionally established legislative authorities (government or parliament).Footnote 9 By de jure, in contrast, we mean a constitutional empowerment that would have made the status of citizen legislator explicit and that was missing in this case, even on the most charitable interpretation of the government mandate. Only a de jure status would have guaranteed a direct link between the legislative proposals and decisions of the convention and actual law and policy outcomes.Footnote 10
We think the category of “citizen legislator” is useful to identify not only what happened in the French case but also to name what was missing in most other citizens’ assemblies, in which citizens may have acted as citizen representatives but not as citizen legislators. It could also prove useful in the future if countries implement citizen representation of a different kind from the CCC that still amounts to an establishment of what we call citizen legislators. Thus, we propose this category as not only capturing something essential about the CCC but also transcending its specifics.
Assuming that the CCC members did operate as legislators of a new kind, did they prove up to the task? Were the legislative proposals they came up with any good?
Were Ordinary Citizens Competent Legislators?
The evaluative criteria for legislative proposals are, of course, varied, complex, and controversial. At an abstract level, however, it seems reasonable to assume that a political decision is prima facie bad (or a mistake)—regardless of its content—when it is rejected by the citizenry; fosters corruption or conflicts of interest among political officials; produces the opposite of its intended outcome; or leads to avoidable famine, economic collapse, political breakdown, or similar catastrophes (Estlund Reference Estlund2008, 162–67; Landemore Reference Landemore2012, 211–13). Accordingly, our aim in the following discussion is modest: to show that the CCC’s outcome was neither absurd nor bad in some of these respects.
What we can say first is that the proposals were honed over several months of hard work and were reviewed, annotated, commented on, and assessed by various experts. Ultimately the proposals were translated into legal language by the legal team supporting the citizens’ work. This process ensured enough substantive and formal quality that a small but non-negligible 10% of the proposals were turned, as per the president’s original promise, into regulations or were otherwise directly applied by the end of the summer 2020 or by February 2021.Footnote 11
It is impossible to ascertain whether the rest of the measures were rejected or reformulated because of technical deficiencies or for political reasons. However, no single measure was publicly dismissed by a qualified expert on the ground of its deficiencies or impracticability. President Macron publicly vetoed three of them for strictly political reasons. Even the most hostile reactions in the media questioned the CCC’s political representativity or its autonomy, rather than challenging the technical soundness of its proposals. For example, the right-wing media accused the CCC of being governed by Marxist pro-degrowth ecologists but did not attack the competence of the citizens or the quality of their proposals per se.Footnote 12
A second way to evaluate the quality of the proposals is to ask about their political viability, as measured by popular support. By June 2020 two-thirds of the French people had heard of the convention, and 74% approved of the main proposals it put forward, including the seemingly coercive proposal for mandatory housing renovation. They only rejected the proposal to reduce speed limits on highways.Footnote 13 Three of five French people considered that body legitimate to produce recommendations on behalf of the French public.Footnote 14 Additionally, by February 2021, 55% of the French public thought recommendations emanating from citizens’ assemblies should be automatically implemented by government.Footnote 15
One possible reason for this congruence between what the convention proposed and what the French population was ready to support is the diversity in “user knowledge” that this type of assembly brings to the table.Footnote 16 Whereas the carbon tax was decided by a government full of wealthy, urban people who would never feel its sting, the 149 proposals put forward by the convention passed through the filter of a wide range of life experiences—roofer, farmer, truck driver, airline pilot, mayor, insurer, student, renter, owner, and so on—ensuring that the major objections and downsides to each proposal were discussed and, when possible, taken into account.
A third criterion of quality is more substantive: Were the “solutions” developed by the 150 smart enough; that is, likely to solve the problem? It is of course hard to measure this in practice, though the support team of experts did try. They provided a rough assessment of the fitness of each proposal for the GHG reduction target (on a rough scale of 1 to 3, with 3 being the highest contribution). The established think tank IDDRI, with the mission of facilitating the transition to sustainable development, praised the proposals for presenting original and coherent trade-offs, as well as offering a new systemic approach to ecological transition.Footnote 17 Giraudet and Guillemot (Reference Giraudet, Guillemot, Courant and Reber2024) similarly concluded that on the “housing” front, although the CCC proposals were rarely original, they were more ambitious and responsive to participants’ firsthand, socially informed concerns than existing proposals put forward by experts and the subsequent government-initiated bill.
At this point, however, another question emerges. Although these legislative proposals may have been good enough to answer the common objection of citizen incompetence, were they truly the product of citizens’ ingenuity, or were they instead the work of the experts who supported and potentially even guided them? After all, the citizens can only be said to be the actual authors of the law-like proposals if they were not controlled and dominated by experts.Footnote 18 Experts’ input has been consistently identified as the primary source of influence in minipublics (Elstub et al. Reference Elstub, Carrick, Farrell and Mockler2021; Goodin and Niemeyer Reference Goodin and Niemeyer2003; Luskin, Fishkin, and Jowell Reference Luskin, Fishkin and Jowell2002), which raises concerns about the integrity of the process in terms of the autonomy of citizens.
Citizen and Expert Relations in the French CCC
Experts played an enormous role in the CCC, more so than in other comparable experiments. The Governance Committee brought in around 130 experts (in addition to those serving on the committee itself) to testify in front of the assembly at various points and to help them write their proposals. Some of the experts were constituted as a support group of 14 members (Technical Advisory Group). The Legal Committee made up of seven lawyers was created to help citizens formulate their proposals in a manner compatible with French and international law. In order to minimize framing effects, the Governance Committee had initially considered structuring expert testimonies as adversarial debates but eventually concluded that this was not the appropriate format for the topic of climate change. Instead, it decided to mitigate the risk of expert-introduced biases by bringing in a diversity of expert and stakeholder profiles, including some chosen by citizens themselves. It also established a clear rule that the Technical Advisory Group and the Legal Committeee were to assist, not steer, the participants.Footnote 19
These innovative features were specific to the CCC and not found in other citizens’ assemblies and mini-publics. The ratio of experts to assembly members in the CCC was close to 1:1, compared with approximately 1:2 to 1:3 in other climate assemblies.Footnote 20 The relationships between experts and citizens in the CCC were also more extensive than in previous citizens’ assemblies, with experts embedded in the protocols at almost every stage of the process (Smith Reference Smith2024, 59). The difference is even more pronounced when compared with James Fishkin’s deliberative polls, in which experts only intervene upstream of the process, in the preparation of briefings materials, and as external speakers brought in on panels to address questions generated by citizens in small groups. None of these interventions allow experts to mingle with the participants, work closely with them, or take any active role in the output of the deliberation (Fishkin Reference Fishkin2011, 395; Reference Fishkin2025, 53). Paradoxically, the more limited yet structuring role of experts in deliberative polls arguably makes their input more constraining, framing debates and priming citizens without the possibility for citizens to push back and question frames over a sufficiently sustained period of time (Gleason Reference Gleason2011). By contrast, the multiple ways in which CCC participants could, over a long period of time, engage with experts throughout the convention helped build trust between the two groups and demystify expert knowledge. This sustained and close relationship, in turn, arguably gave citizens greater freedom to challenge expert framing and mitigate priming effects.
A major concern evoked by the participation of so many experts is the danger of capture, a type of corruption that happens when a political actor is overly influenced by a special interest or corporate group. The less experienced citizens are, the more vulnerable they are to capture by experts, steering bodies, and other stakeholders (Bagg Reference Bagg2023; Landa and Pevnick Reference Landa and Pevnick2021; Umbers Reference Umbers2018). As one of the very few cases of maximally empowered citizens’ assemblies, the CCC was more likely than most citizens’ assemblies to be the subject of attempts at capture, thereby allowing us to test the resilience, vulnerability, or both of lay citizens to such a risk.
Importantly, however, the relevant comparison point to assess the role of experts in the convention is not a citizens’ assembly entirely insulated from experts’ influence but rather existing parliaments, whose members routinely rely on the help of countless experts and lobbyists when writing laws in parliamentary commissions. Despite the enormous influence of experts in shaping the law, we still credit parliamentarians with authorship of the law, assuming that they are capable of relying on experts without abdicating judgment and responsibility.Footnote 21
Before proceeding, we need to introduce two analytical distinctions. A first distinction is between blind and rational deference. Blind deference occurs when surrendering one’s judgment entirely in the face of claimed expertise; rational deference, by contrast, consists of suspending one’s judgment for second-order reasons rooted in experts’ credentials and reliability but with the persistent awareness that the statements to which one defers can always be openly challenged and contested. In the first case, authority is perceived as indisputable, whereas in the second, authorization and trust are accompanied by a background context of critical scrutiny involving both experts and nonexperts (Warren Reference Warren1996, 56). For citizens, blind deference amounts to taking what experts say at face value; through rational deference, however, citizens exercise their own judgment in deciding when and to what extent to defer to the experts’ judgement (Goodin Reference Goodin2020, 26).
A second important distinction is between influence and undue influence. As a citizen put it during an interview, not every influence jeopardizes independence of judgment: “So there are people who say that we are being influenced.… Yes of course, but then in that case you put your children in school they are influenced by the teacher, by the education they receive.… Since experts enlighten us on the basis of their experience, their knowledge and scientific technical training we are influenced. After that you decide if you agree or not.” (male, around 50)
Undue influence, however, takes place whenever experts take advantage of their audience’s ignorance or abuse their authority to advance their own agenda and manipulate the citizens’ deliberations. If undue influence by experts had been the norm at the CCC, the content of its proposals could not be said to have been developed by the citizens and instead should be described as having been manufactured by the experts.
The subjective perception of the influence of experts varied among citizens but is well epitomized by this statement from one participant: “I have a quite easy relationship with experts, they are here to help us. I don’t feel influenced at all given the experts’ panel we had and given the panel that we can consult in private. That is, we are solicited by associations, by civil society, and it’s our responsibility to enlarge our spectrum to have all the information that prevent us from being influenced.” (male, age 30)
Our ethnographic observations of the proceedings and that of other researchers at the convention (Giraudet et al. Reference Giraudet2022) confirm citizens’ subjective perception that they were not captured by experts, for a variety of reasons.
First, in the CCC, the relationship between citizens and experts was designed by the Governance Committee—and understood by the 150 citizens—to invert the traditional relationship between experts and amateurs, subordinating expert advice and knowledge to the ends defined by citizens, as recommended by democratic theorists (e.g., Christiano Reference Christiano, Mansbridge and Parkinson2012, 33). This inversion required a transformation of both groups and a relationship of trust and respect anchored in a clear acknowledgment that citizens would provide the direction while the experts were there to help them actualize their visions. We saw this inversion deepen over the course of the convention, with the citizens slowly gaining self-confidence and the experts becoming increasingly humble, as several examples presented later illustrate. We describe this particular organization of hierarchical relations as putting experts “on tap, not on top.”Footnote 22
Second, citizens never simply yielded to experts’ judgments because of their greater knowledge. Instead, they oscillated between partial, rational deference and utter defiance. The Paris airport CEO learned this the hard way when he came on stage in his three-piece suit and started lecturing citizens about climate change since the time of the dinosaurs. When it became clear that his confidence outmatched the relevance of his comments, he was interrupted by a polite but firm remark: “We already know all this. Can you please get to the point: What would it take to reduce the number of flights for short distances?” When he went on to argue that the Paris airports were a model of environmental responsibility and cleanliness, he was met with undisguised skepticism.
The most emblematic case of citizens’ rebellious attitude toward experts’ advice, however, was their position on the principle of a carbon tax. Although this option was defended by experts at every turn and initially was supported privately by a small majority of the convention members as revealed by researchers’ questionnaires, it was spectacularly rejected in a plenary session early in the process, in which the expert who presented it was berated by a few infuriated citizens: “Don’t treat us like children!” Ultimately, the convention decided to put forward 149 proposals, none of which supported a carbon tax at the national level, although they did accept the principle of one at the European Union level. They did so against the recommendations of most of the experts who had talked to them, as well as President Macron himself, whom some suspected of convening the CCC in the hopes of gaining citizens’endorsement of this solution.
Third, citizens were able to identify and discuss precisely, with great nuance, where and to what extent experts were overstepping their boundaries. For example, in an interview conducted among members of the Housing group near the end of the process, citizens complained that “there were things that had been added really a lot and not yet discussed” (male, 30) and “words that did not come from us” (female, 40) in a 34-page text produced by the experts from the Technical Advisory Group on the basis of the citizens’ recommendations. The citizens had just received the document to review and had gone through its first six pages. Pressed to clarify their concerns, another citizen (male, age 60+) said, “Everything that has to do with coercive measures, everything that concerns obligations to do things, penalties [did not come from us]. If I consider the first sentence of the measures, one finds ‘forbidding’ ‘fines’ ‘sanction’ ‘penalty…’ I didn’t note them all. After a while it’s starting to add up.”
The citizens were not denying that they had envisaged coercion as a means of making French people curb their greenhouse gas emissions. But they felt that the experts had too much zeal in endorsing sanctions and other coercive measures and therefore did not capture the nuance of their own deliberations. The citizens insisted on a distinction between coercion applied at the level of individuals (which they were uncomfortable with) and coercion applied at the level of institutions and collective agencies (which they felt was more justified). They also insisted on using coercion only “as a second step”—after incentives had already been used. Later in the weekend the citizens were able to present their grievances to the experts and have them fix the mistakes.
Fourth, whenever citizens followed the advice of experts, there were generally objective reasons for them to do so. For example, the citizens in the working group on Housing quickly rallied to an idea presented to them by experts in the first two weekends: a global renovation project making it mandatory for citizens to make their homes energy efficient by a certain date. Citizens were seemingly more compelled by the merit of the argument than by the charisma of the particular experts presenting the idea. Global renovation is in fact widely seen as one of the most important levers of public policy to fight greenhouse gas emissions in a potentially efficient and equitable manner (Giraudet and Guillemot Reference Giraudet, Guillemot, Courant and Reber2024, 244). Yet, the citizens did not uncritically adopt the proposal. Taking into account the imperative of social justice and the remarks from the representative of an agency working for low-income families, the citizens came up with a plan to subsidize global renovation so as to make it costless for those who could not afford it.
There was also a general resistance to expert control throughout the process, which manifested in the way citizens continuously reopened questions and queried established frameworks. For example, early in the process, after a long PowerPoint presentation by a high-level civil servant on the hierarchy of norms in French law, a citizen asked, “In the table of the hierarchy of norms, you did not note the place of contracts. Why?”Footnote 23 The state councilor replied, “It was to enlighten you on what you are going to propose. A priori you are not going to propose contracts…. Well, that said, yes. That’s an excellent idea. Contracts are also part of what you could propose—contracts between private persons on certain matters. You are right. Don’t bridle yourselves. You are right” (emphasis added).
This 180-degree reversal of the expert’s position was prompted by a single citizen’s question, which revealed a blind spot in the expert’s thinking that had unduly narrowed the deliberative space for citizens.
A second example of citizens’ tendency to reopen questions and frameworks defined or limited by experts emerged at the end of the process during a debate on the possibility of having a referendum on their proposals. One of the organizers presented a PowerPoint discussing the option of a yes or no answer to a single question, the only format ever explored in the French context. However, an alternative came up in the Zoom chat when a citizen remarked that a multiple-choice referendum like the one organized in the wake of the Irish citizens’ assemblies might be preferable. The organizers deflected, arguing that the legal team had chosen not to put the option of a multiple-choice referendum on the menu precisely because it was “uncertain,” given that it had never been used in France before.
A citizen in the Zoom chat then asked, “What is it that’s ‘uncertain?’” A few minutes passed without anybody addressing the question until Cyril Dion, one of the four guarantors—appointed process monitors playing the role of external watchdogs as explained in appendix B—brought up the issue again and asked why the legal team had sidelined the possibility of a multiple-choice referendum (which could have taken the form of multiple single-question referenda held the same day to circumvent the French constitutional constraints). The organizer became defensive. The guarantor insisted, “It is important to keep this an open hypothesis, and it’s up to citizens to decide.” Several citizens in the chat then expressed their agreement. These two examples illustrate the way citizens continuously questioned the experts’ frameworks. The second case also illustrates that the questioning of experts by citizens can be enhanced by good institutional design, such as the presence of guarantors, to prevent takeover or undue influence by experts.
Last but not least, the citizens expressed their independence by increasingly challenging the governance of the assembly over time. Early in the process they started to ask about the possibility of inviting experts of their own choosing; for example, the former Minister of Ecology Nicolas Hulot, whose idea of “ecocide” they wanted to hear more about and ultimately took up in their final proposals. They also extended, without prior approval by the Governance Committee, direct invitations to various people, including President Macron, who obliged with a much-publicized visit to the convention in January 2020, and to Greta Thunberg, who never replied.
Strikingly, President Macron’s visit was conducted on the citizens’ terms, rather than the terms of the Governance Committee or even of the president himself. In response to their invitation, the president had sought to bring them all to dinner at the Elysée Palace (the official residence of the French president). The citizens did not like the idea and thought that he should come to them for a work session on equal terms; further, he should be seated at a table with them rather than lecturing from a podium. The Governance Committee was appalled by the citizens’ demands, arguing, “You don’t say no to the President!” In the end, the citizens were able to impose their conditions, and President Macron’s visit took the form of a working session, in which he was seated across from the citizens at a large table.Footnote 24 There were countless other examples of such assertions of autonomy by the citizens regarding both the substantive and procedural aspects of the convention.
Possible Remaining Sites of Expert Domination
Citizens in the CCC proved their ability to retain control over the deliberative process as they moved toward the formulation of proposals to curb greenhouse gas emissions in a socially acceptable way. Two qualifications, however, need to be introduced. The first has to do with the power of the Legal Committee to shape the convention’s outcome through the simple act of translating their proposals into legal language. As we saw earlier, the Governance Committee members were divided about what should be the role of lawyers in reformulating citizens’ proposals in legal language toward the end of the process. The role of lawyers as translators also caused frustration among the citizens, which they expressed in casual conversations and more formal interviews. The second, perhaps more serious, concern relates to the role of the Governance Committee and the facilitators of the event. We consider these objections in turn.
Did the lawyers on the Legal Committee exercise undue influence and try to shape the content of the citizens’ proposals? And, even if they did not do so in the context of the CCC, could they have done so?
The Governance Committee clearly defined the role of the Legal Committee as being in the service of citizens, not as co-constructors of the proposals. Constrained by this official mission, the lawyers themselves seemingly took a humble view of their role. Here is how one of the main members of the Legal Committee, an assistant professor of law, described the role of the Legal Committee as she experienced it: “To transmit a general [legal] culture foundation to the citizens; to listen to citizens in the working groups; to transcribe their proposals into legal language; and to compare these legal transcriptions with those proposed by Parliament [after the Convention was over] so as to be able to explain to citizens the difference between what they had proposed and what Parliament came back with.”Footnote 25
Empirical observations further confirm that the Legal Committee did not visibly attempt to manipulate, modify, or distort the content of the proposals. To the extent that betrayals occurred, they appeared to result from slippages and approximations caused by the fact that the legal team was stretched thin and pressed for time, rather than from any conscious or even unconscious attempt to grab power.
The Legal Committee worked hard to introduce transparency into its methods and to justify and explain its translation choices to the citizens, even producing a methodological memo available to the convention members. Random selection of the CCC’s members also provided a safeguard against (even inadvertent) capture by experts, lawyers included. Among a sufficiently large sample of randomly selected citizens, it was highly like that the group would include lawyers or individuals familiar with legal jargon who thus could be able to keep the Legal Committee accountable: this is indeed what happened at the CCC, as demonstrated in an earlier example. Finally, even citizens with no knowledge of law were extremely attentive to the formulations proposed to them and, as we observed in the final sessions, pushed the Legal Committee to change and reformulate phrases when they felt that the interpretations did not capture their exact intent.
The only domain in which lawyers had a real advantage was the knowledge of what could be translated into law or not. When they claimed that a proposal was not sufficiently precise as to be translatable into legal language, citizens had no choice but to trust them. Although this situation presented a real threat of manipulation by experts, we saw no evidence that the legal experts used it to pursue an agenda different from that of the citizens. Of course, claims about the nonexistence of evidence are tricky to substantiate, and it is possible that invisible forms of manipulation took place. Looking forward, the remedy to informational asymmetry is probably a mix of professionalism, transparency, and ensuring that there are enough experts and that those experts have sufficiently diverse political convictions so that they can keep each other in check.
A more serious threat to the 150 citizens’ claim to authorship arguably lay in the Governance Committee.Footnote 26 The primary function of this body of 15 people was to settle structural questions, such as defining the parameters within which the deliberations would be allowed to take place (Did the mandate cover the question of nuclear energy? Should the citizens be allowed to formulate proposals on sources of financing for their proposed measures?); settling procedural questions, such as the number of meetings, the agenda of each meeting, and the schedule and steps each meeting would require; selecting the experts and interlocutors whom the 150 would be allowed to consult; and interfacing with the government and other institutions. The Governance Committee carried out expert selection in a discretionary manner, based on expert availability and identified needs, and only rarely in response to participants’ requests. Even though citizens’ autonomy of judgment regarding the experts’ advice was preserved, that both the objectives and the means to achieve those objectives were mediated by an opaque committee remains a valid source of concern.
This Governance Committee included its two copresidents, seven members of the CESE, three climate experts (two of which were also affiliated with the CESE), and three participatory democracy experts. The representatives of the two private organizations in charge of organizing and facilitating the sessions (Missions Publiques and Res Publica) were also present during the Governance Committee’s meetings and participated in its deliberations, although they did not have voting rights.
The convention was therefore governed by experts and not its own members. The 15 people on the Governance Committee were able to impose a number of key procedural and substantive decisions. Additionally, all these choices were made without much justification or accountability to the CCC members, although the added presence of two randomly selected citizens after the first session, who were rotated after each session, possibly helped on that front. There was little transparency about the functioning of this committee, because researchers were not allowed to observe them or their deliberations. As a result, we do not know what was the balance of power between the various factions on the committee nor the extent to which the introduction of the two regularly rotated citizen representatives was able to influence the dynamics.
Additionally, known environmental activists on the Governance Committee, as well as an activist serving as one of the guarantors, sometimes appeared more concerned with pushing specific proposals than impartially facilitating a democratic process and committing to it, regardless of its outcomes.Footnote 27 In the Food working group, for instance, the copresident of the Governance Committee Laurence Tubiana lamented that the topic of meat had not yet been included in the agenda (Courant and Baeckelandt Reference Courant, Baeckelandt, Courant and Reber2024, 208–10). There was not a clean separation between the role of experts and other roles in the CCC, likely confusing citizens as much as it did external observers (see appendix D).
Most problematically, some conflicts of interest built into the governance structure were arguably reflected in some of the convention’s outcomes. For example, the survival of the CESE as an institution depended on the CCC’s success. If it could prove to be a capable host for the whole process, it had a chance to avoid the fate floated by President Macron in the months before: its replacement by a new “chamber of the future.”Footnote 28 Somehow, surprisingly specific recommendations about making the CESE a chamber of participation, in addition to its existing role as the chamber of organized civil society, found their way into the final citizens’ report, even though the future of the CESE as an institution had no direct link to the convention’s mandate of curbing greenhouse gas emissions.Footnote 29 Similarly, the CESE members’ previously mentioned unanimous stance against the idea of letting citizens vote on the legal formulations of their proposals suggests a corporate interest against the precedent of empowering citizens as legislators. Indeed, what added value would the CESE as an institution have if citizens’ assemblies could speak without its intermediation? Because these 7 CESE members were still a minority on the 15-member governance committee, they were outvoted. But this kind of conflict of interest at the top arguably poses a real threat to the autonomy of citizens.
The role of the service providers in charge of the CCC’s organization also raises some issues. These individuals brought in methodological principles and commitments honed over years of practice. But some of these commitments—for example, to generate ideas and consensus, rather than thoroughly explore disagreements—seemed to lag behind the debates among social scientists on these issues. In some cases, their professional commitments reflected a possible conflict of interest: for example, pleasing the Governance Committee versus pleasing the citizens. Although these service providers were often portrayed as neutral technicians and did not have the right to vote, they wielded a lot of influence on the deliberations of the Governance Committee.
Despite all this, the principle of placing the citizens “on top” and the experts “on tap” was sufficiently respected throughout the process to justify seeing the citizens as authors of the proposed legislation. Most notably, several elements of their proposals, such as the criminalization of ecocide and constitutional reform, went beyond their initial mandate. This political momentum even continued beyond the convention, culminating in the establishment of an association dedicated to advocating for the proposals and overseeing their implementation. In other words, the CCC participants were able to take ownership of their political vision.
This governance model, however, raises fundamental questions about the kind of power and autonomy such assemblies can and should be granted. Consider that parliaments do not have a Governance Committee. They are not told what to legislate about from without. They have full control over their agenda, their deliberations, and their ultimate decisions. It is worth considering whether randomly selected assemblies could not develop the same ability through similar or other rules (see also Landemore, Lacelle-Webster, and Pénigaud Reference Hélène, Lacelle-Webster and de Mourgues2026).
Conclusion
We have argued that the French CCC was the first citizens’ assembly in which participants were able to act as “citizen legislators,” a descriptive category we offer as a refinement of Mark Warren’s notion of “citizen representative.” Whereas citizen representatives act and speak for the larger public generally, citizen legislators specifically write law proposals on behalf of the larger public. In the CCC, however, the status of citizen legislators remained de facto, as opposed to de jure, because the CCC had no clear legal status apart from being a creature of the prime minister’s office, and the French constitution does not yet recognize citizens’ assemblies as valid sources of law.
We also argued that the legislative proposals produced by the CCC were of sufficient quality to defuse the objection of incompetence often leveraged against ordinary citizens and that the CCC participants were able to resist capture by the experts supporting their work. To the extent that they followed expert recommendations, it was on the basis of what we called “rational” as opposed to “blind” deference. We also credit the ability of citizens to resist expert capture to the principle “citizens on top, experts on tap” that, in our observations, governed most of the interactions within the convention. We suggest that reinforcing citizens’ ability to resist capture would require moving from the current model of externally governed citizens’ assemblies to a self-ruling one.
Much work, both normative and empirical, remains to be conducted for citizen legislators to become a normatively attractive and politically realistic proposition. Although citizens may be capable of writing laws, is it the best use of their time and talents, as opposed to, for example, defining a broad agenda for the country? Other questions remain open: What does an actually self-governing citizens’ assembly look like? Would citizens’ assemblies with as much or more power be able to maintain their unity when confronted with more highly divisive issues (Shapiro Reference Shapiro2017; Pénigaud de Mourgues Reference Pénigaud de and Théophile2025)? In an actual citizens’ legislature with an open agenda, how would citizens be able to bundle issues absent the existence of structured parties (Rosenbluth and Shapiro Reference Rosenbluth and Shapiro2018; contra see Guerrero Reference Guerrero2024)? What should the relationship between citizens’ assemblies and the rest of the population look like (Curato and Böker Reference Curato and Böker2016; Lafont Reference Lafont2019)?
Yet, given the growing academic debates about the possibility and desirability of randomly selected legislatures, as well as the demonstration by the French CCC that citizens can act as legislators under at least some circumstances, the ideal of a democracy where some legislative functions are in the hands of randomly selected citizens suddenly seems less utopian and possibly within reach. For all its limitations, the French experiment is thus a proof of concept that could ultimately pave the way for a constitutionalization or otherwise legalization of the de facto status of “citizen legislators” into an official, de jure one.
Supplementary material
To view supplementary material for this article, please visit http://doi.org/10.1017/S1537592725104295.
Acknowledgments
We are grateful to audiences at the 2019 Colloquium on “Random Selection in Politics” at the École Normale Supérieure de Lyon, where early versions of this project were first presented and our initial collaboration took shape; to discussions at Université Paris-Créteil in 2020 with Patrick Savidan and his students; and to participants in the 2020 ACM Collective Intelligence Conference. We also thank the audience of the 2024 Seminar of Practical Philosophy at Université Laval (Canada) and participants in the 2024 APSA panel “Innovating for Democracy” in Philadelphia. Further valuable feedback was received at the 2025 Ohio University workshop “Democracy and Social Change” and at the Yale political science workshop “Politics Across Borders” (Spring 2025). We thank Victor Wu for editorial assistance and Thalsa-Thiziri Mekaouche for research assistance. We are especially grateful to Isabela Mares and Ian Shapiro for insightful comments on the article’s framing and methodology and to the anonymous reviewers at Perspectives on Politics for their constructive suggestions. Finally, we thank the 150 participants in the French Citizens’ Convention for Climate, whose work and civic commitment led us to rethink fundamental assumptions about what modern citizens are capable of.
Open access
