Sortition in practice: what permanent bodies chosen by lot have shown so far

Most citizens' assemblies are one-off events: a lottery-selected group meets for a few weekends, reports and disbands. Since 2015 a smaller set of public bodies has tried something more ambitious — standing institutions whose members are drawn by lot and rotated, attached to a parliament or city government and able, in some cases, to choose their own agenda. This study examines what is known about these permanent sortition bodies: the Ostbelgien Citizens' Dialogue, the Brussels mixed deliberative committees, the Paris Citizens' Assembly, Milan's standing climate assembly, Gdańsk's citizens' assemblies, and the earlier Toronto and Madrid experiments; it then sets them against proposals for a full legislative chamber chosen by lot. The evidence is thinner than the rhetoric around "institutionalising deliberation" suggests. Permanent bodies have proved feasible and relatively cheap, and their procedures are generally followed. But they are small, mostly advisory, dependent on the political sponsor that created them, and almost never evaluated against clear measures of representativeness, attrition or influence.

Key findings

• Permanence is now real but small-scale. Ostbelgien, a community of fewer than 100,000 people, has run a statutory citizens' dialogue since 2019, with seven completed assembly cycles by 2026. Brussels' parliaments had held six mixed committees of citizens and MPs by February 2024. Paris has selected three cohorts of a 100-member assembly since 2021.
• Political reversibility is the main risk. Madrid's City Observatory of 49 randomly selected residents, created by regulation in January 2019, was reverted to a body of politicians and officials in February 2020 after a change of government. Toronto's Planning Review Panel paused after two cohorts pending a review.
• Self-selection is the central representativeness problem. Response rates to lottery invitations are low — 5.8% for Climate Assembly UK and around 10% in Gdańsk's early assemblies — and the volunteer pool over-represents groups such as university graduates. Quotas correct the visible demographics, but not necessarily attitudes or motivation.
• Selection algorithms can be fairer than they were. Flanigan and colleagues (2021) showed that the selection method used by a leading practitioner gave some volunteers near-zero chances of selection; their "LEXIMIN" algorithm, used for more than 40 panels by mid-2021, equalises probabilities as far as quotas allow.
• Attrition is rarely reported. Where it is, it is modest for short processes: Climate Assembly UK had 103 of 108 members at its final weekend. No permanent body reviewed here publishes systematic attendance data.
• Public legitimacy is conditional. In France, 68% and then 63% of survey respondents supported a permanent assembly of 150 citizens with a veto over parliamentary texts, but only around 30% were confident such citizens could deliberate productively. Cross-national surveys find support strongest among the politically disaffected.
• Chambers by lot remain proposals. Theorists from Gastil and Wright to Abizadeh argue for a sortition chamber alongside an elected one; no national legislature has adopted one, and the closest real institutions are advisory or agenda-setting.

Research question and scope

The question is: what does the evidence show about permanent or standing bodies chosen by lot — about how representative they are, whether members stay, how they are perceived and what they change? "Permanent" here means an institution established by law, regulation or standing rules that outlives any single cohort of members, even if each cohort serves only months. One-off assemblies are covered in a companion study on the impact of citizens' assemblies. The review covers sub-national bodies in Belgium, France, Italy, Poland, Spain and Canada, and the main proposals for legislatures chosen by lot. It does not cover juries in courts, or ancient and early-modern sortition (Athens, Venice), except as background.

Background

Random selection for public office has a long history, from the Athenian council of 500 to the lotteries used in medieval Italian republics. Its modern revival came through the citizens' jury, the planning cell (Planungszelle) and the deliberative poll, and later through national citizens' assemblies. The OECD's 2020 comparative study counted twelve models of representative deliberative process and treated two of them — the Ostbelgien model and Madrid's City Observatory — as permanent models, alongside ongoing panels such as Toronto's. It identified three routes to institutionalisation: standing bodies; legal requirements to convene a deliberative process under certain conditions (such as Oregon's Citizens' Initiative Review or the 2017 Mongolian deliberative polling law); and rights for citizens to demand one by petition. The OECD also warned that "it will take time to know how these various examples ... work in practice, as most were established only recently."

The intellectual case for permanence has three strands. First, standing bodies can set agendas, not just answer questions chosen by politicians. Second, they can monitor what happens to earlier recommendations. Third, they build familiarity: a public that knows a body exists is more likely to accept its output. Critics reply that permanence invites capture by organisers or politicians, that small rotating bodies cannot carry real authority, and that the advisory status of nearly all of them leaves elected officials free to ignore them.

Method and sources

The study combines official documents (decrees, rules of procedure, city and parliament web pages), the OECD 2020 report, practitioner accounts and peer-reviewed research. For each body we recorded the legal basis, size, selection procedure, term, powers and documented follow-up. Representativeness, attrition and legitimacy were assessed from whatever data each body publishes and from cross-cutting studies. Limitations are substantial: several bodies publish little beyond descriptions of procedure; some accounts are written by the practitioners who designed the processes (for example, Marcin Gerwin on Gdańsk); and the Milan description rests on the city's own brochure. Where only the sponsor's account exists, this is stated.

Finding 1: the permanent bodies in practice

Ostbelgien: a statutory two-tier model

The Parliament of the German-speaking Community of Belgium adopted the decree establishing its Citizens' Dialogue (Bürgerdialog) unanimously in February 2019, after a design process involving practitioners and academics; the first Citizens' Council met on 16 September 2019. The structure has two tiers. A standing Citizens' Council of 24 members, all former assembly participants and partly renewed at regular intervals, chooses topics, convenes assemblies and monitors follow-up. Each topic is then handled by a citizens' assembly of 25 to 50 people — in practice about 30. Selection runs in two stages: the permanent secretariat has the national register draw a random set of residents aged 16 or over, who are invited in writing; from those who accept, the final members are drawn by quota on age, gender, place of residence and socio-economic background, in proportion to the population.

Niessen and Reuchamps (2022) emphasise the binding element of the procedure rather than of the outcome: recommendations are debated in a joint committee of assembly members, MPs and the responsible minister, who must then state whether and how each will be implemented and justify rejections. The parliament's public record shows seven completed topics — care, inclusive education, housing, digital skills, integration, pupils' competences and older people — each with a public presentation, a political response within roughly two to five months and, for the first six, a review session about a year later. An eighth assembly, on spatial planning, meets in autumn 2026. The record does not say which recommendations were implemented; the review session asks how far recommendations "inspired" politicians' work.

Brussels: mixed committees in parliament

In December 2019 the Parliament of the Brussels-Capital Region amended its rules to create mixed deliberative committees, and the French-speaking parliament in Brussels (the Cocof) made the same change. As designed, a regional committee combines 15 MPs from the relevant standing committee with 45 residents aged 16 or over, of any nationality, drawn by lot (12 MPs and 36 residents in the Cocof). Topics can be proposed by MPs or by a citizens' suggestion backed by at least 1,000 residents; the parliament's Bureau decides and must justify rejecting a citizens' topic. The approved recommendations go to the standing committee, which within six months must publish a follow-up report giving detailed reasons for its choices and present it to the allotted members at a public sitting, according to Reuchamps's account of the rules. By February 2024 six committees had been held: on 5G deployment, homelessness, urban biodiversity and urban noise (regional parliament), and on citizens' role in crisis management and work-linked training (French-speaking parliament).

The Brussels design is distinctive because elected and allotted members deliberate together. That may improve uptake — MPs who helped write recommendations are more likely to defend them — but it also creates a risk that politicians, with more experience and information, dominate discussion. Public evaluation of the six committees is limited; the parliament itself held a colloquium on follow-up in February 2024.

Paris: a 100-member city assembly

The Council of Paris created the Paris Citizens' Assembly (Assemblée citoyenne de Paris) by resolution in October 2021, following a 2019 citizen conference that gathered more than 9,000 contributions. It has 100 members drawn by lot. Cohorts have been renewed roughly annually: the first served from November 2021 to March 2023, the second from July 2023 to July 2024 and the third from September 2024. Members can put a topical question to the Council of Paris, submit motions (vœux) and propose a draft council decision (délibération) through the deputy mayor responsible for participation. The first cohort worked on the city's reading plan, energy renovation of housing and a Paris street code; the second produced what the city calls the first citizen deliberation, on homelessness, and a motion to create 100 "garden streets". The city commits to a "right of follow-up" (droit de suite), with annual public meetings on implementation. The available account is the city's own; no independent evaluation of implementation was found.

Milan: a standing climate assembly

Milan's Permanent Citizens' Assembly on Climate accompanies the city's Air and Climate Plan to 2030. According to the city's brochure, it has 90 members drawn by lot to represent the city's population, subject to periodic rotation; members are invited by a personal letter and serve for six consecutive months after a short training. The assembly meets in plenary every two months, with small-group work between plenaries, and the city "commits to respond point by point" to its proposals. Unlike Paris or Brussels, its remit is the implementation of an existing plan rather than new policy. Its role is closer to a standing user panel than to a legislature.

Gdańsk: binding by mayoral pledge

Gdańsk held Poland's first citizens' assembly in 2016, after flooding from an extreme rainfall. According to Marcin Gerwin, who proposed and coordinated it, Mayor Paweł Adamowicz agreed that recommendations supported by at least 80% of members would be binding on him. A second assembly, in spring 2017, produced nine recommendations on air pollution, including a ban on burning coal in household furnaces within five years with support for poorer households; a third dealt with civic engagement and equal treatment. Gerwin reports a response rate to invitations of around 10%. The "binding" status rested on a political commitment by one mayor rather than on law. Gdańsk's municipal rules separately allow residents to initiate participatory processes, including citizens' panels: 1,000 signatures to propose one and 5,000 to make the request impossible to refuse, according to the OECD, which also noted that the right had not yet been used.

Madrid and Toronto: experiments that stopped

Madrid's City Council transformed its dormant City Observatory into a deliberative body by a regulation of 29 January 2019. In February 2019 the council sent 30,000 letters; on 12 March 2019 it drew 49 members from 1,135 volunteers, stratified by gender, age and district. Members were to meet at least eight times a year, review the most popular proposals on the city's Decide Madrid platform and could send proposals to a local referendum. After the municipal elections of May 2019 brought a new governing coalition, the Observatory's composition and function were reverted in February 2020 to a body of politicians and officials.

Toronto's Planning Review Panel ran from 2015 to 2017 with 28 randomly selected residents serving two-year terms (11 full-day meetings after four days of training), and a second cohort of 32 from 2017 to 2019; the OECD reported a pause pending review of the planning department's engagement strategy. The regional transport agency Metrolinx ran a similar 32-member Regional Reference Panel for 11 full-day meetings between October 2018 and May 2020.

Comparison of permanent and standing bodies

Finding 2: how representative are bodies chosen by lot?

The two-stage lottery and its weak point

Almost every body reviewed uses the same two-stage method. A large random sample of residents receives an invitation; those who accept form a pool; the final members are drawn from the pool under quotas. The OECD notes that the first stage typically reaches at least 2,000 people for small populations and around 30,000 for national processes. The weak point is the acceptance stage. Climate Assembly UK received 1,748 positive responses, a 5.8% rate, which its organisers described as within the 3–7% range typical for UK assemblies; Gdańsk's early assemblies had a response rate of around 10%; Madrid's 30,000 letters produced 1,135 volunteers, under 4%.

Who declines matters more than how many. Jacquet's (2017) interviews with people who refused invitations to three Belgian mini-publics identified six logics of non-participation: concentration on private life, low internal political efficacy, avoidance of public meetings, schedule conflicts, political alienation, and the belief that the mini-public would have no impact. Several of these are correlated with social position, so quotas on age, gender and education cannot guarantee that the people selected resemble those who stayed away in attitudes or experience. Climate Assembly UK added a quota on concern about climate change for exactly this reason; few permanent bodies do so.

Fairness in the final draw

Flanigan, Gölz, Gupta, Hennig and Procaccia (2021) showed that the second stage introduces its own distortion. Because the volunteer pool is skewed — for example towards college graduates — any algorithm that meets demographic quotas must select under-represented volunteers with higher probability and over-represented ones with lower probability. The algorithm previously used by the Sortition Foundation, they found, gave some pool members a near-zero chance of selection even when much more equal probabilities were possible. Their LEXIMIN algorithm maximises the lowest selection probability subject to the quotas; the Sortition Foundation alone had used it to select more than 40 panels by June 2021, and it is offered as open-source software. This does not solve self-selection, but it means that the lottery among volunteers can be made as fair as the quotas permit.

Remuneration and inclusion

The OECD found that 44% of deliberative processes paid participants, 7% compensated transport and 6% covered expenses, while 43% offered nothing — mostly local processes. It cites practitioner evidence that payment helps recruit young and lower-income people, but notes that its own data cannot test this. Climate Assembly UK paid £150 per weekend plus travel, accommodation and care costs. Permanent bodies, whose members may serve for months, face a sharper version of the trade-off: paying enough to include people with jobs and caring duties, without turning membership into a quasi-salaried post.

Finding 3: attrition and continuity

Attrition is the least documented aspect of permanent sortition. Short national assemblies report low drop-out: Climate Assembly UK selected 110, two did not attend the first weekend, one withdrew and was replaced, and 103 of 108 attended the final weekend. Organisers commonly over-sample groups expected to drop out — Climate Assembly UK selected five extra members for this purpose. For bodies whose members serve six months (Milan), a year (Paris, Madrid) or two years (Toronto), the risks are higher: members move, change jobs or lose interest, and replacing them mid-term breaks the link between learning and deciding. None of the permanent bodies reviewed publishes attendance or replacement statistics comparable to those of Climate Assembly UK. This is a significant evidence gap, because a body that loses its younger, poorer or more sceptical members over time may drift away from the population it is meant to mirror.

Continuity of the institution itself is a separate problem. Madrid shows that a regulation can be undone within a year by a new majority; Toronto's panel stopped after review; the German Bundestag's staff unit for citizens' assemblies was dissolved in 2025 after a change of government (see the companion study). Ostbelgien's cross-party unanimity at adoption is the exception and may be part of why it has endured.

Finding 4: legitimacy — inside and outside the room

Public support

Surveys find broad but soft support. Pilet, Bol, Vittori and Paulis (2023), using representative samples of 14,043 people in 15 Western European countries, found the strongest support for sortition-based assemblies among the less educated, those with low political self-confidence and those with anti-elite views — and that support varied with what respondents expected assemblies to decide. In France, a survey run alongside the Citizens' Convention for Climate found 68% (first wave) and 63% (second wave) support for a permanent assembly of 150 randomly chosen citizens with a veto over parliamentary texts, yet only 32% and 28% were confident that such citizens could deliberate productively on complex issues, and 74% of those aware of the convention doubted its representativeness.

The "like me" problem

These figures suggest a gap between support for the idea of ordinary citizens deciding and trust in any particular group of them. Permanent bodies may close or widen that gap. If the public comes to know the institution, as with juries, legitimacy may rest on the procedure rather than on the people. If the body is obscure, or is seen as a creature of the government that created it, the opposite may happen. No permanent body reviewed here has published repeated public-awareness surveys that would show which way it is going.

Legitimacy with elected officials

Elected officials are the other audience. Brussels' mixed committees and Ostbelgien's joint committees try to build ownership among MPs. The Madrid reversal and the German experience show the cost of failing to do so: a body seen as one party's project is vulnerable to the next majority.

Finding 5: proposals for a chamber chosen by lot

Several theorists have proposed going beyond advisory bodies. Gastil and Wright (2018) argue for a bicameral legislature combining an elected chamber with one chosen by sortition, claiming that the pairing would enhance the virtues of each while dampening their vices, and identifying sampling, training and deliberative procedure as the main design challenges. Abizadeh (2021) offers a third justification for bicameralism — an institutional compromise between the agency and accountability that elections provide and the equality and impartiality that sortition provides — and applies it to propose that Canada's Senate be reconstituted as a randomly selected assembly. Bouricius (2013), drawing on the Athenian system of multiple allotted bodies, proposes dividing law-making among several allotted bodies (agenda-setting, proposal development, review, final decision) to avoid concentrating power in any one. Van Reybrouck's Against Elections (2016) popularised the case in public debate, and Landemore's Open Democracy (2020) argues for "open mini-publics" as a core institution of a more inclusive democracy.

The atlas's case file on a legislature by lot collects these proposals. None has been adopted by a national legislature. The closest real institutions are the Ostbelgien Citizens' Council, which sets agendas but cannot legislate, and the Brussels mixed committees, which deliberate inside parliament but only recommend. The design questions the theorists raise — accountability of members who cannot be voted out, protection against capture by staff, lobbyists or facilitators, and the right size and term for a legislative body chosen by lot — have therefore not been tested at scale.

What the evidence does not show

• It does not show that permanent bodies produce more policy change than one-off assemblies. No study compares uptake across the two formats with comparable measures.
• It does not show that rotating membership solves capture. Rotation limits the entrenchment of members, but permanent bodies concentrate continuity in secretariats, facilitators and sponsoring officials, whose influence is rarely examined.
• It does not show stable representativeness over time. Attrition data are missing for every permanent body reviewed.
• It does not establish legitimacy beyond the sponsor's account. Paris and Milan publish descriptions, not evaluations; Ostbelgien's reviews ask about "inspiration", not implementation.
• It does not show that a sortition chamber would work. The theoretical literature is rich; the empirical base is advisory bodies serving populations from fewer than 100,000 to about two million.

Implications for institution design

1. Secure cross-party support before creation. Ostbelgien's unanimous decree contrasts with Madrid's single-coalition regulation. A body that depends on one majority should expect to end with it.
2. Put follow-up obligations in law or standing rules. Reasoned responses within a fixed period (six months in Brussels; joint committee responses in Ostbelgien) are the most transferable feature. Pledges by individual mayors, as in Gdańsk, are fragile.
3. Measure and publish representativeness and attrition. Every cohort should report invitations sent, response rates by group, quotas used, selection probabilities, attendance and replacements. These data cost little to publish and are the basis of any claim to represent the public.
4. Use fair selection algorithms and over-sampling. Algorithms such as LEXIMIN, attitudinal quotas where relevant, and reserve lists reduce, though they cannot remove, the effects of self-selection.
5. Pay members. Payment and care support are the most practical tools for including people who would otherwise decline.
6. Protect independence from staff and sponsors. Publish agendas, witness lists and facilitation guidelines; give the citizen body a role in choosing its own experts; rotate facilitators.
7. Test, do not leap. A legislature by lot would be a large constitutional change. Agenda-setting councils, mixed committees and petition-triggered assemblies offer smaller steps whose effects can be measured first.

Open questions

• Does permanence increase uptake, or only the predictability of the process?
• How long can members serve before they behave like a new political class — or simply stop attending?
• Can mixed bodies of citizens and politicians avoid domination by the politicians?
• What public awareness does a permanent body need to gain legitimacy in its own right?
• Would a sortition chamber with real veto power attract organised lobbying of its members, and how could that be policed?
• How should a body chosen by lot be held accountable when its members cannot be re-elected or removed by voters?

Sources

1. OECD (2020). Innovative Citizen Participation and New Democratic Institutions: Catching the Deliberative Wave. Paris: OECD Publishing. Chapter 2 (models), chapter 4 (selection, remuneration), chapter 6 (institutionalisation; Boxes 6.2, 6.3, 6.4, 6.12, 6.13).
2. Niessen, C. & Reuchamps, M. (2022). Institutionalising citizen deliberation in parliament: the permanent citizens' dialogue in the German-speaking Community of Belgium. Parliamentary Affairs 75(1), 135–153.
3. Bürgerdialog in Ostbelgien. Stand der Dinge.
4. Bürgerdialog in Ostbelgien. Losverfahren and Gremien und Beteiligte.
5. Parliament of the Brussels-Capital Region (8 February 2024). Commissions délibératives (press release on six deliberative committees).
6. Reuchamps, M. (2020). Belgium's experiment in permanent forms of deliberative democracy. ConstitutionNet, International IDEA (composition, citizens' suggestions and six-month follow-up rule of the Brussels committees).
7. Ville de Paris. Tout savoir sur l'Assemblée citoyenne de Paris.
8. Comune di Milano / Poliedra–Politecnico di Milano. Assemblea Permanente dei Cittadini sul Clima (brochure).
9. Gerwin, M. Designing the process of delivering recommendations by the citizens' assembly. newDemocracy Foundation research note.
10. Climate Assembly UK (2020). The path to net zero: About Climate Assembly UK.
11. Jacquet, V. (2017). Explaining non-participation in deliberative mini-publics. European Journal of Political Research 56(3), 640–659.
12. Flanigan, B., Gölz, P., Gupta, A., Hennig, B. & Procaccia, A. D. (2021). Fair algorithms for selecting citizens' assemblies. Nature 596, 548–552.
13. Pilet, J.-B., Bol, D., Vittori, D. & Paulis, E. (2023). Public support for deliberative citizens' assemblies selected through sortition: evidence from 15 countries. European Journal of Political Research 62(3), 873–902.
14. Giraudet, L.-G. et al. (2022). "Co-construction" in deliberative democracy: lessons from the French Citizens' Convention for Climate. Humanities and Social Sciences Communications 9, 207.
15. Gastil, J. & Wright, E. O. (2018). Legislature by lot: envisioning sortition within a bicameral system. Politics & Society 46(3), 303–330.
16. Abizadeh, A. (2021). Representation, bicameralism, political equality, and sortition: reconstituting the second chamber as a randomly selected assembly. Perspectives on Politics 19(3), 791–806.
17. Bouricius, T. G. (2013). Democracy through multi-body sortition: Athenian lessons for the modern day. Journal of Public Deliberation 9(1).
18. Van Reybrouck, D. (2016). Against Elections: The Case for Democracy. Translated by Liz Waters. London: The Bodley Head. ISBN 9781847924223.
19. Landemore, H. (2020). Open Democracy: Reinventing Popular Rule for the Twenty-First Century. Princeton: Princeton University Press.