Fifty-three percent of French people would be willing to replace the Senate with an assembly chosen by lottery, and yet 32% of respondents judge that the government’s adoption of proposals from the Citizens’ Convention on Climate is a clear failure. The figure of 53% comes from DeCoDe, a survey conducted in May 2025 and published in December 2025; the figure of 32% concerns the evaluation of a failure to adopt proposals from the Citizens’ Convention on Climate. These capture the essence of a contradiction that liberal democracies have yet to resolve. According to Dimitri Courant, respondents favor greater citizen influence over the agenda and the ratification of decisions, particularly through petitions and referendums. Multiplying deliberative spaces without giving them real influence over decisions can produce the opposite effect from what is sought: growing mistrust, because the promise of being heard without the reality of follow-up resembles mere theater.
The Essentials
- Multiplying citizens’ assemblies without giving them real influence over decisions can produce the opposite effect from what is sought: mistrust.
- The OECD has follow-up data for only 55 processes from its 2020 study; it does not allow conclusions about the absence of follow-up for part of the 733 cases in the later database.
- In 2026, 51% of French people believe there is nothing to be proud of in the democratic system, according to the CEVIPOF Political Confidence Barometer; trust in government falls to 17%.
- The absence of a stable majority makes trade-offs and text adoption more difficult. In citizens’ assemblies, the absence of decision-making power can reduce the impact of deliberation.
- The Irish case shows that a consultative assembly can be connected to parliamentary review and then, for certain constitutional reforms, to a referendum.
Deliberative Processes Without Measurable Follow-Up
What these cases reveal goes beyond a simple lack of political will. It signals a design problem: a significant number of these processes were launched without the follow-up question being resolved beforehand. In the absence of formal obligation, the executive’s discretion can be substantial, but some mechanisms provide for a legal obligation or public commitment to respond. Yet this very discretion is precisely what participants end up perceiving. When a mechanism’s architecture provides no constraint on the executive, deliberation produces a paradoxical result: it mobilizes participants’ attention on real issues, then signals to them that this attention commits no one else.
The disappointment that follows is proportional to the gap between the promise of being heard and the lack of follow-up. This gap is structural, and no amount of care devoted to the quality of deliberation can bridge it.
The OECD has documented 733 deliberative processes since 1979 and has limited follow-up data: 55 processes for its 2020 study, which shows the incompleteness of available data. Participants deliberated. The government archived.
This figure contradicts the hypothesis that has guided the proliferation of these mechanisms over the past twenty years: better informed and better heard, citizens would trust institutions more. Dimitri Courant, a researcher specializing in citizens’ assemblies, writes that respondents favor greater citizen influence over the agenda and the ratification of decisions, particularly through petitions and referendums. A consultation with no effect on the final decision sends a precise message: your opinion was collected, it commits no one.
The Citizens’ Convention on Climate in France illustrates the mechanism on a large scale. Launched in 2019, it brought together 150 randomly selected citizens for nine months. Their 149 proposals were handed to the government with a promise of adoption “without filters.” On the most ambitious measures, the filter was applied. Participants saw it.
Part of the general public did too.
Parliamentary Impasse and Deliberative Impasse Reinforce Each Other
Benjamin Morel, a specialist in parliamentary institutions, particularly notes the limits of a caretaker government’s ability to make trade-offs. This institutional diagnosis resonates beyond Parliament itself.
The same deficit of binding power operates in both spaces. The absence of a stable majority makes trade-offs and text adoption more difficult, without necessarily preventing all decisions. In citizens’ assemblies, the absence of decision-making power can reduce the impact of deliberation. The two mechanisms converge: The Barometer was administered in January 2026; it notably reports 51% of respondents finding nothing to be proud of in the democratic system. Trust in government falls to 17%.
These figures measure a perceived gap between institutional supply and real capacity to change things, not a lack of interest in politics.
It would be convenient to stop there and conclude that representative democracy is exhausted. Reality is more nuanced. The Fifth Republic has shown remarkable resilience in the face of successive crises, and Morel’s diagnosis is that of a reformist, not a liquidator: French institutions need adjustment, not replacement. The question is the nature of that adjustment.
The Irish Model: What It Proves and Its Limits
An additional limit concerns subject selection. Irish assemblies addressed societal questions on which parties avoided taking stands, precisely because the divide did not follow usual party lines. This configuration is favorable: it reduces the risk that political actors perceive the mechanism as a direct threat to their position. Extending the model to domains where partisan interests are strongly structured would expose the deliberative chain to different kinds of resistance. Extending the model to domains where partisan interests are strongly structured—redistributive taxation or economic regulation—would expose the deliberative chain to different kinds of resistance, and drawing a general conclusion from this about the capacity of binding assemblies to restore trust regardless of domain would be premature.
Ireland offers the most solid comparison available. Its Constitutional Convention of 2012-2014 addressed notably same-sex marriage; its citizens’ assembly of 2016-2018 addressed the constitutional article on abortion, aging, climate, referendums, and fixed terms. Some constitutional recommendations contributed to subsequent referendums, while others were addressed to the Oireachtas for review. The positive results of constitutional referendums have legal effect, but the assemblies’ recommendations were consultative. Participation in debates was high, as was perceived legitimacy.
Researcher Pierre Rosanvallon, whose work concerns democratic legitimacy, has shown that institutions gain credibility not through their origin—who represents whom—but through their capacity to produce decisions that citizens recognize as their own. The Irish model rests on a possible sequence of deliberation, political and parliamentary review, then referendum when a constitutional amendment is retained. Each link is visible. Each link is binding.
But the Irish model has conditions of possibility that it would be imprudent to generalize without naming them. Ireland is a country of five million inhabitants with a constitutional system that makes referendums relatively accessible. France has a referendum tradition marked by two resounding failures in 1969 and 2005 that have made successive governments very reluctant to use this tool. Directly transposing the Irish model to France would first require resolving the constitutional framework question. It is feasible—several constitutional lawyers have shown it—but it is a political undertaking in itself.
It would be false to present a transfer of power as an obvious solution: it is a bet, not a recipe.
Mistrust Toward Consultations, a Problem of Transparency
A competing reading deserves serious consideration. According to an interpretation inspired by Dani Rodrik’s work on democratic regulation of markets, the quality of deliberation might depend less on formal power granted than on the transparency of the decision-making process that follows. In other words: citizens might accept that their recommendations are not automatically followed, provided they understand precisely why, with what arguments, by whom, and with what assumed political responsibility.
This hypothesis is testable. Public, dated, and reasoned follow-up is a good practice recommended by the OECD. The Gothenburg Commission in Sweden has produced this kind of systematic documentation on urban planning questions. The OECD reports a general association between subnational participation and higher trust, without publishing here a positive impact measure for individual processes.
Mistrust toward consultations stems partly from an administrative transparency deficit, which can be addressed without constitutional reform. Making public the rate of follow-up on recommendations by government, with an obligation to provide reasons, would be a measure applicable immediately. OECD data show that direct additional feedback to participants is little documented, but do not quantify the number of processes with no feedback at all on follow-up given.
Granting Power or Granting Accountability: Two Paths, One Common Condition
The DeCoDe survey by CEVIPOF goes beyond the purely technical framework of citizens’ assemblies: it concerns what citizens expect from politics. The figure of 53% in favor of replacing the Senate with lottery selection expresses concern about institutions perceived as unresponsive to collective issues. French democracy is going through a representation crisis documented over several electoral cycles.
Two scenarios emerge for the coming decade. The first extends the current trajectory: multiplication of consultative assemblies, low follow-up rates, absence of binding mechanisms. OECD data document the rise and characteristics of 733 deliberative processes. The second bets on the institutionalization of assemblies with formal decision-making power, linked to a referendum or legislative obligation, on specific and circumscribed domains—local taxation, urban planning, climate adaptation. The Irish model links certain recommendations to review by the Oireachtas and sometimes to a referendum, without formally transferring decision-making power to the assembly.
These two scenarios do not entirely exclude each other. A government can start with transparency, make follow-up rates public, publicly justify its refusals, without waiting for constitutional reform. And constitutional reform can be prepared by binding local experiments that test feasibility conditions. The signals to watch by 2030 are concrete: participation rates in citizens’ assemblies, the share of recommendations translated into legislation, the evolution of the CEVIPOF barometer on trust in institutions. Follow-up and implementation data are incomplete and not uniformly available, though comparisons between processes are conducted.
The Two Deficits in Common
The connection is a matter of interpretive analogy; it is not documented by the OECD as a structural link between the two phenomena. The two deficits present points of convergence: the absence of a stable majority in Parliament and the absence of decision-making power in citizens’ assemblies both reduce the impact of deliberation and voting. A contingent majority can adopt a text; the legal link between parliamentary vote and legislative decision is established by the Constitution. Without an institutional mechanism for response and public follow-up, a citizens’ assembly does not meet good practices recommended by the OECD; the actual absence of follow-up must, however, be established case by case.
Trust in democratic institutions depends on citizens’ ability to see the link between deliberation and decision. This chain presents visible breaks in France. Repairing it requires specific political choices, not merely formal innovations. Experimenting with binding referendums on a specific domain, making the follow-up rate for recommendations public, creating a formal obligation to provide reasons when a government rejects the conclusions of a citizens’ assembly: these paths do not require waiting until 2040.
Some can be implemented by decree or through ordinary legislation in the coming legislature.
Sources
- Dimitri Courant, “The French Want More Power, Not More Consultation,” The Conversation, June 2026, read article
- CEVIPOF (Sciences Po), DeCoDe Survey, June 2026, CEVIPOF, Sciences Po
- CEVIPOF (Sciences Po), Political Confidence Barometer, January 2026 wave, CEVIPOF, Sciences Po
- OECD, database on deliberative processes (733 processes since 1979), OECD, Innovative Citizen Participation
- Benjamin Morel, Parliament, Temple of the Republic: From 1789 to Today, LGDJ, lgdj.fr
