Theory — mass participation and mini-publics
Updated: 2026-10-05. Stream J. Connects to research/software/CAPABILITY-LANDSCAPE.md (the tooling gap is documented there; this record is the theory side).
Concept
The relationship between the wider public and a randomly selected mini-public — and how to connect them without importing self-selection into the assembly itself.
Definitions / competing meanings
- Deliberative system: the whole division of labour across sites of talk — mini-publics are one site, not the system (SRC-0111).
- Crowdsourcing + mini-public: mass input generates agenda/evidence; the mini-public deliberates it — “agenda-setting by the many, deliberation by the few” (SRC-0111, SRC-0112).
- Mass opinion mapping: large-scale structured input (Polis: agree/disagree/pass → opinion clusters) — opinion distribution, not deliberated judgement (SRC-0092).
- Self-selection pressure: any open channel biases toward the already-engaged; if the mini-public merely ratifies open input, it imports that bias (SRC-0120, SRC-0127).
Major theoretical arguments
- Normative: Lafont’s dilemma implies the healthy role for mini-publics is stimulating public deliberation, not replacing it — mass-input linkage is theoretically necessary, not optional (SRC-0104).
- Systems response: legitimacy comes from transmission — mass input → deliberated judgement → institutional response → public justification — evaluated as a chain (SRC-0111, SRC-0112).
- Empirical caution: open submissions skew organised and resourced; corpus shows submissions managed through curation/balancing, not raw aggregation (SRC-0038 evidence governance, SRC-0102).
Connecting a mini-public to the wider public (required specific question)
Mechanisms observed or theorised, with their bias profile:
- Open submissions (Ireland, CA UK stakeholder calls): imports self-selection; mitigate via balanced curation + disclosure (SRC-0017, SRC-0038, SRC-0102).
- Structured mass opinion mapping (Polis-style): broader reach, still opt-in; output is opinion clusters for the mini-public to weigh, not commands (SRC-0092).
- Topic selection by public vote (Newham): democratic but majoritarian and self-selecting (SRC-0048).
- Standing agenda council (Ostbelgien): selected citizens, not the mass public, set agendas — avoids self-selection but narrows input (SRC-0021).
- Petitions/thresholds (Paris jury requests): self-selection filtered by threshold + institutional gatekeeping (SRC-0025).
- Surveys/reference polls: population-representative input without opt-in skew — strongest representativeness, weakest deliberative depth (SRC-0127).
Finding: no mechanism removes self-selection from the mass side; the design question is whether bias enters as agenda (correctable by curation/balancing) or as membership (uncorrectable inside the room). Keeping membership by lottery and input by curated channels preserves the distinction (SRC-0111).
Critiques / limitations
- Mass-input legitimacy claims are often inflated; opt-in numbers are not representativeness (SRC-0124 is the sceptical counterweight).
- Connecting the mini-public outward can itself distort: publicity pressures members toward performance (see protected-deliberation trade-off,
deliberation-risks-and-safeguards.md, SRC-0030).
Operationalisation in the corpus
- People’s Assembly for Nature: ~30,000-response national conversation → 100-member assembly — the largest mass→mini-public pipeline in the corpus (SRC-0059).
- Ireland/CA UK: curated submission calls (SRC-0017, SRC-0038); Ostbelgien: citizen-initiated agenda via council (SRC-0021).
Cicada relevance
- Cicada should represent mass input and mini-public membership as distinct participation modes with distinct bias properties — conflating them is the known failure mode (SRC-0111; software side: Polis ≠ assembly, SRC-0092).