Citizens' Assemblies and Sortition — Playbook
Random citizens deciding public questions instead of politicians — the evidence on whether that works better.
What Toronto can build on deliberation that actually sticks, learning from where France broke its own promise.
Voice note: Toronto already paid for the answers — this work reads it all, integrates it, and shows the work, including the uncertainties.
The honest bottom line
A citizens' assembly gathers a randomly selected, demographically representative group of residents, gives them balanced evidence over multiple days or weeks, and asks them to deliberate and recommend — democracy by lottery rather than election. The OECD has catalogued nearly 300 such cases across 34 countries and found the process reliably produces thoughtful, trusted, common-ground recommendations, especially on the values-laden, long-term, trade-off-heavy questions elected politics handles worst. Ireland is the proof it can move the immovable: citizens' assemblies broke decades of political deadlock on same-sex marriage and abortion, with their recommendations carried into successful national referendums. But deliberation is not decision. Assemblies almost always only recommend, and a high-quality recommendation changes nothing if the surrounding politics won't carry it. France's government promised to pass its 2020 climate convention's 149 proposals "without filter," then dropped or diluted most of the boldest ones — participants graded the government 3/10. Canada's own excellent assemblies died the other way: BC's 2004 assembly recommended a new voting system, and the 2005 referendum got 57.7% Yes against a government-set 60% threshold; Ontario's 2007 assembly's recommendation was rejected 63% by voters. It may be a coupling-and-design failure, or simply voters legitimately saying no — this page does not pick one. The Netherlands ran one of the largest citizens' assemblies yet, on climate policy, with the government committing in writing to respond to all 23 proposals within six months. That response has now happened: the Dutch cabinet delivered it on schedule, 29 May 2026, adopting roughly half the recommendations (largely by pointing to plans already underway), still weighing 18, and declining 23 outright. The deadline held; whether "half adopted, mostly via pre-existing plans" counts as genuinely honouring the assembly's work is a fair question this page does not resolve. Neither card below is costed at Toronto scale or has a confirmed funding source; both name what is and isn't settled rather than presenting an aspirational design as decided.
---
a recommendation card — A Standing Toronto Civic Assembly with a Legal Duty-to-Respond
Card id: a recommendation card · Issue: citizens-assemblies-sortition · Backgrounder: our research file for that page · Trust: carried-forward
Problem
The evidence base is consistent that assembly impact is determined by the decision pathway coupled to it, not by deliberation quality, and that one-off assemblies are structurally easy for a government to ignore — France's government pocketed a high-quality 2020 climate convention after promising to pass it "without filter," and ad hoc assemblies restart from scratch each time with no institutional memory. This card addresses only the institutionalization gap — not which specific topics a Toronto assembly should take up, which is a separate design question the backgrounder leaves open.
Action
Establish a standing, recurring Toronto civic assembly — convened on a fixed cycle (e.g., annually or biennially) on a rotating slate of complex, values-laden, citywide questions — paired with a council-adopted charter provision creating a legal duty for Council to publicly respond to each assembly recommendation within a fixed window and either act or explain its refusal, modeled on the Ostbelgien (Belgium) permanent-council/duty-to-respond design and the Dutch National Citizens' Climate Assembly's newly observed six-month response mechanism.
Jurisdiction split
- City does: convening a standing assembly, funding its secretariat and stipends, and adopting a council procedural rule requiring a timed public response to its recommendations are all within existing municipal authority — Toronto does not need provincial or federal permission to create this obligation for itself.
- City demands of Province: none required for the core duty-to-respond mechanism itself, since it binds Council rather than a senior government; where assembly recommendations land on matters outside municipal jurisdiction (housing, transit funding), the City's role becomes advocacy upward rather than an enforceable demand — see a recommendation card for that distinct mechanism.
- City demands of Feds: none identified as required for this specific institutional design.
Cost
Low-to-mid single-digit millions CAD per convening cycle, anchored to two named comparators: BC's 2004 Citizens' Assembly, funded at approximately $5.5 million for 158 members over a year, and the Dutch 2025 National Citizens' Climate Assembly, which sent 70,000 invitations, from which 4,070 residents indicated willingness to participate, with 175 members ultimately selected from that pool — a materially larger recruitment funnel than BC's, useful as an upper-bound scale comparator, and one where the middle (willing-respondent) step matters for reading the ratio correctly rather than treating this as a flat 70,000-to-175 selection rate. A standing body would add ongoing secretariat/infrastructure cost between cycles, offset over time by not rebuilding recruitment and design capacity from scratch each convening.
Funding path
A dedicated line in the City's existing civic-engagement or Clerk's-office budget, sized against the BC/Dutch comparators above; no new provincial or federal transfer is required for the core mechanism, since the funding obligation is entirely within municipal control. This is the most fiscally self-contained option in this page’s evidence base for exactly that reason — it does not depend on another government's cooperation to exist, unlike a recommendation card below.
Who benefits, and how
Toronto residents broadly, via a standing channel for demographically-representative deliberation on citywide questions that elected politics handles poorly (long-term, values-laden, trade-off-heavy issues); specifically, residents routinely absent from ordinary consultation (lower-income, renters, racialized communities, those who never attend a council meeting) benefit from stratified, paid recruitment design, conditional on that design actually being funded.
Who bears the cost, and how
City taxpayers city-wide, via the dedicated civic-engagement budget line; opportunity cost of that budget capacity not being spent on an alternative consultation method. No externally-imposed cost, since the mechanism requires no other government's funding.
Who benefits from the status quo
No beneficiary identified. The backgrounder's Cui Bono table (added 2026-07-14) came up empty after checking this library's internal records in full and a supplementary live check for facilitation/consulting-vendor or referendum-avoidance beneficiaries — none met the Prime Rule's ESTABLISHED/REPORTED sourcing bar. This is a valid outcome, not a gap: a governance-process-design problem (assemblies being advisory and easy to ignore) does not have the same "who profits from the status quo" shape as an extraction-style problem with a named contracted vendor or beneficial owner.
Financial ROI
Not separately quantified for a standing Toronto body specifically; the master briefing frames the cost of any single assembly as "trivial against the stakes" of the billion-dollar budgets or generational policy questions it can inform, but does not model a specific fiscal-return figure for institutionalization itself.
Economic ROI
No source quantifies this. A live-discovery search this review (economic-impact/local-spending/employment studies of citizens' assemblies specifically) found extensive literature on deliberative-democracy governance outcomes (OECD, DemocracyNext, academic reviews) but no third-party economic-impact study quantifying local growth, induced spending, or employment effects from convening a standing assembly. This is a governance-capacity mechanism, not a spending program with a plausible IMPLAN-style multiplier of its own.
Social ROI
Directionally supported by the OECD's ~300-case evidence that deliberation reliably produces trusted, legitimate recommendations on values-laden questions, and by the accountability-deficit critique's own implication that institutionalization (rather than one-off use) is what allows an assembly's legitimacy gains to compound over multiple cycles rather than being spent once and forgotten.
Environmental ROI
Modest and real, not headline-scale — in-person convening (participant/staff travel to a venue, printed materials, venue operation) over an annual or biennial cycle, materially smaller than a capital or infrastructure program's footprint. No source quantifies a specific citizens'-assembly carbon/waste footprint; a live-discovery search found extensive climate-assembly literature documenting assemblies as a policy tool for emissions reduction, but no study modeling the convening's own footprint — a different measurement than what the assembly recommends. Confidence: low.
Evidence
the inherited master briefing's §The strongest case AGAINST· carried-forward · one-off fragility, accountability-deficit trade-offthe inherited master briefing's §Real-world precedents· carried-forward · Ostbelgien/Paris standing-assembly modelthe inherited master briefing's §Policy recommendations, item 1· carried-forward · duty-to-respond design rationalethe inherited master briefing's §Costs & financing· carried-forward · BC $5.5M anchor, institutionalization cost-amortization argument- NEW (2026-07-13) · KNOCA, Dutch Citizens' Climate Assembly report · six-month response mechanism, 70,000-invitation → 4,070-willing → 175-member selection funnel
the inherited master briefing's §Equity & distribution· carried-forward · representativeness conditional on paid, supported design
Confidence & uncertainties
Medium-high confidence that institutionalization plus a duty-to-respond addresses the specific, well-evidenced one-off/ignorability failure mode. Lower confidence on the exact convening cadence, topic-selection process, and total standing-body cost, none of which is resolved in this page’s sources. ⚠️ UPDATE (live-checked 2026-07-13): the Dutch government's response arrived on schedule, 29 May 2026 — the deadline itself held — but the substance was mixed: roughly half the recommendations adopted, largely by pointing to pre-existing plans rather than new commitments, 18 still under consideration, 23 declined. That mixed result is itself informative for this card: the duty-to-respond mechanism reliably produces a response, but does not by itself guarantee the response is substantively generous — a distinction this card's own design (a duty to respond, not a duty to adopt) already anticipates.
Status
DRAFT — blocked on: a Toronto-specific standing-body cost estimate, fairness and legal review. (No longer blocked on confirming the Dutch government's response — it delivered 29 May 2026, per a 2026-07-13 live-check.)
---
a recommendation card — An Assembly-Triggered Referral Mechanism for Deadlocked Council Questions
Card id: a recommendation card · Issue: citizens-assemblies-sortition · Backgrounder: our research file for that page · Trust: carried-forward
Problem
Toronto's council system, like elected politics generally, is described in this page’s sources as poorly suited to genuinely deadlocked, values-laden, low-consensus questions specifically because members face re-election pressure and party-adjacent incentives that reward short-term, safe positions over long-horizon trade-offs. Ireland's experience shows sortition can break exactly this kind of deadlock by depoliticizing an issue, but the master briefing is explicit that Ireland's referendum-culture, sovereign-government context "may not transfer" directly to a municipality without referendum machinery. This card addresses only the referral trigger and its coupling design — not which policy questions should be referred, which is a case-by-case council decision.
Action
Create a formal council procedure allowing a defined supermajority of Council (or a citizen-petition threshold) to refer a deadlocked, values-laden question to an ad hoc citizens' assembly convened under a recommendation card's standing infrastructure, with the referring motion required to state in advance what Council commits to do with the result — rather than convening an assembly first and deciding the decision pathway afterward, the design failure this page’s evidence identifies in France.
Jurisdiction split
- City does: creating the referral procedure itself, funding and convening the ad hoc assembly, and committing Council to its own pre-stated duty-to-respond are all within existing municipal authority.
- City demands of Province: where the referred question implicates a matter genuinely outside municipal authority (housing supply, transit funding), the assembly's output becomes an advocacy input to the Province rather than a locally executable decision, and the City's role is to carry that recommendation upward with "the moral weight of a representative citizen body," not to promise an outcome it cannot deliver.
- City demands of Feds: same upward-advocacy framing as above, for matters within federal jurisdiction specifically.
Cost
Comparable to a single ad hoc assembly cycle under a recommendation card's cost anchors (BC's ~$5.5 million single-assembly comparator is the more directly applicable anchor here than the larger Dutch national-scale figure, since this is a per-referral, not standing-body, cost). No additional standing infrastructure cost beyond what a recommendation card already funds, assuming both cards are adopted together; if a recommendation card is adopted without a recommendation card's standing infrastructure, the full one-off setup cost applies each time — the less efficient path.
Funding path
Drawn from the same civic-engagement budget line as a recommendation card if both are adopted together; if adopted alone, a per-referral council appropriation, contingent on Council actually approving the referral in the first place (a supermajority or petition-threshold gate is itself a funding gate, since no referral proceeds without it).
Who benefits, and how
Residents on the specific side of a genuinely deadlocked, values-laden question, via a depoliticized deliberative process that the Irish precedent shows can produce a recommendation with cross-partisan legitimacy that elected debate alone could not generate — with the explicit caveat that Ireland's referendum-culture advantage does not automatically transfer, so the benefit depends on this card's own pre-committed decision pathway actually holding.
Who bears the cost, and how
City taxpayers via the same civic-engagement budget mechanism as a recommendation card; opportunity cost of Council time and staff capacity spent designing and running the referral process for each invoked case, which is why a defined trigger threshold (rather than an open-ended referral option) is part of the action above.
Who benefits from the status quo
No beneficiary identified. Same finding as a recommendation card above: the backgrounder's Cui Bono table (added 2026-07-14) came up empty for this page after a full check of this library's internal records and a supplementary live check — no ESTABLISHED or REPORTED beneficiary of the status quo was located for this specific referral-mechanism gap either.
Financial ROI
Not quantified for a referral mechanism specifically; the general cost-versus-stakes argument (a few million to legitimately deliberate a high-stakes question is "a rounding error" against the cost of a badly-made or contested decision) is the closest available anchor, and it is a qualitative argument, not a modeled figure.
Economic ROI
No source quantifies this — same live-discovery result as a recommendation card: no third-party economic-impact study quantifying local growth, induced spending, or employment from convening an ad hoc referral-triggered assembly was located. A per-referral mechanism is smaller in scale than a recommendation card's standing body, so even a borrowed multiplier from an unrelated sector would be doubly speculative here.
Social ROI
The clearest social-ROI case in this page’s evidence is depoliticization itself — Ireland's assemblies gave politicians public cover to act on same-sex marriage and abortion after decades of avoidance, and gave voters a credible, cross-partisan signal distinct from a normal campaign-driven referendum. This card's confidence in that benefit is tempered by the same document's own warning that Ireland's specific conditions (referendum culture, sovereign authority, historical moment) are not guaranteed to reproduce the same effect in a municipal, non-referendum context.
Environmental ROI
Modest, likely smaller than a recommendation card's — a single ad hoc convening (one referral cycle) rather than a recurring standing body, so the per-instance travel/venue/materials footprint is real but bounded to whenever a referral is actually triggered. No study quantifies a single deliberative convening's own environmental footprint, as distinct from the climate-assembly policy-outcome literature, which measures the assembly's recommendations, not its own footprint.
Evidence
the inherited master briefing's §The strongest case FOR, item 2 / item 4· carried-forward · Irish depoliticization mechanism and its exceptionalism caveat; theoretical case for sortition on deadlocked questionsthe inherited master briefing's §The strongest case AGAINST· carried-forward · France's broken "without filter" promise as the design failure this card's advance-commitment requirement is built to avoidthe inherited master briefing's §Toronto-specific factors, §Policy recommendations item 7· carried-forward · jurisdiction-gap / "deliberate upward" framing- backgrounder "Costs and financing" · carried-forward · BC $5.5M single-assembly cost anchor
Confidence & uncertainties
Medium confidence in the referral-trigger design itself (it directly targets the documented France failure mode by requiring the decision pathway to be pre-stated, not decided after the fact). Lower confidence on the right supermajority/petition threshold, which is not addressed anywhere in this page’s sources and is a genuine open design question this card does not resolve. The Irish-transferability caveat applies at full force here and is not softened by this card's own design, since a municipal referral mechanism still lacks Ireland's referendum culture and sovereign authority.
Status
DRAFT — blocked on: a defined referral-threshold design (supermajority size or petition-signature count), confirmation of what local plebiscite authority (if any) Toronto actually holds under Ontario municipal law, fairness and legal review.
---
Production record
Drafting record
Version: v2.0 (playbook conversion) · Original date: 2026-07-13 · Status: DRAFT · What this page draws on: cards draw only on the backgrounder's carried-forward and NEW-cited claims; no new factual claims are introduced in card prose beyond what the backgrounder already establishes, per the L6 template's firewall discipline. FIX-3 note (W1b cards audit, 2026-08-06, carried forward): a recommendation card's Action shortened to the template's one-action bar (was 105 words). No substance changed.
Playbook conversion (2026-08-11, Lane L3a): opened with "The honest bottom line" adapted from archive/dayone/citizens-assemblies-sortition.md (a recorded standing decision retired day-one memo, kept as history in archive/); ROI sections tightened, repeated "not yet estimable / genuine gap" boilerplate collapsed to one honest line each, matching that page's recommendation cards's playbook shape. No a formally registered claim tokens present in this file; all carried-forward/NEW citations preserved unchanged.