Multi-level and nested assembly models
CA-R2 requirements record. Updated: 2026-10-05. Compares models for organising assemblies across geographic/institutional levels. Most of this comparison is theoretical — the corpus contains no direct real-world implementation of nested sortition structures. Evidence basis is research/theory/nested-and-multi-level-assemblies.md (SRC-0109, SRC-0111, SRC-0112, SRC-0125, SRC-0126) plus the single-level real cases. No empirical validation is claimed where none exists.
The core recorded trade-off (theory, SRC-0111/SRC-0125): equal probability of selection argues for fresh independent draws at each level; linkage and accountability argue for delegation upward.
Model 1 — Independent sortition at each level
Neighbourhood → local → regional → national assemblies are each selected independently by fresh stratified draws.
- Legitimacy: each body can claim its own descriptive legitimacy at its own scale — no borrowed authority.
- Representativeness: strongest per-level statistical claim; each level’s population is sampled directly.
- Continuity: weakest — levels are unconnected; the same person almost never serves twice.
- Accountability: each level answers at its own level; no upward chain to inspect.
- Concentration of influence: lowest — no delegation class accumulates power.
- Professionalisation risk: lowest.
- Institutional memory: lowest; every level is institutionally alone.
- Scalability: high in structure (embarrassingly parallel); costly in operation — every level needs its own invitation/selection/support pipeline.
- Complexity: lowest conceptually; highest aggregate administrative load.
- Privacy/data: population frames needed at every level — the largest aggregate data footprint.
- Audit: each draw independently auditable — simple provenance, many ceremonies.
- Suitability: cleanest where levels are constitutionally independent (e.g. genuinely federal or devolved structures).
- Evidence status: theoretical — no corpus case implements it.
Model 2 — Delegation upward
Lower-level assemblies select or delegate members to sit in higher-level bodies.
- Legitimacy: indirect — higher bodies inherit legitimacy from the selecting assemblies (analogous to indirect election, SRC-0125).
- Representativeness: weakest statistical claim — delegation compresses; the national body describes the assemblies, not the population.
- Continuity: strongest — the same people carry experience upward.
- Accountability: delegation chains create identifiable conduits but also principal-agent gaps.
- Concentration of influence: highest — delegates accumulate agenda knowledge and visibility across levels.
- Professionalisation risk: highest — delegates become a semi-professional political class.
- Institutional memory: highest among the models.
- Scalability: cheap — no new draws; but each lower assembly must run a credible internal selection.
- Complexity: low administration; high legitimacy-accounting complexity (who chose whom, when, by what rule).
- Privacy/data: smallest new-data footprint (no new frames); delegate relationships become sensitive records.
- Audit: must record delegation choices, not just draws — a distinct audit class.
- Suitability: tight-knit structures where upward linkage matters more than fresh descriptive legitimacy.
- Evidence status: theoretical for sortition systems; the delegation mechanism is well-evidenced in elected/federal structures.
Model 3 — Mixed delegation + fresh sortition
Some higher-level seats are filled by delegation (continuity), the rest by fresh stratified draws (representativeness).
- Legitimacy: hybrid — partially direct (fresh draw), partially delegated.
- Representativeness: moderated — fresh-draw seats preserve population linkage.
- Continuity: partial — a continuity fraction carries memory without dominating.
- Accountability: mixed channels.
- Concentration of influence: bounded by the delegation fraction — the fraction itself is a governance choice.
- Professionalisation risk: present but diluted.
- Institutional memory: delegated fraction carries it.
- Scalability: moderate — needs fresh draws plus delegation machinery.
- Complexity: highest design complexity — two membership bases in one body.
- Privacy/data: both footprints apply.
- Audit: two distinct provenance classes (draws + delegations).
- Suitability: where a system wants both linkage and freshness — the corpus theory suggests this is the most defensible general-purpose compromise (SRC-0125).
- Evidence status: theoretical — analogous to the Brussels hybrid-elected+citizen mixing but not implemented for level-nesting in the corpus.
Model 4 — Permanent upper citizens’ council + temporary issue assemblies
A standing citizens’ council exists at the upper level; it convenes, briefs or oversees temporary single-issue assemblies beneath it (the Ostbelgien relation, generalised — SRC-0021).
- Legitimacy: upper council’s permanence + bounded mandate; lower bodies stay bounded.
- Representativeness: upper council is stratified; issue assemblies drawn per topic allow topic-fit strata.
- Continuity: council provides it; issue assemblies needn’t.
- Accountability: council is inspectable continuously; issue assemblies via their outputs and council oversight.
- Concentration of influence: concentrates in the permanent council — mitigated by rotation (Ostbelgien thirds).
- Professionalisation risk: concentrated in council members and secretariat — managed by rotation but never zero.
- Institutional memory: highest of the sortition-native models — the council is the memory.
- Scalability: good — one permanent institution plus repeatable bounded assemblies.
- Complexity: moderate; the council-issue relationship (convene, brief, oversee, agenda-set) needs explicit definition.
- Privacy/data: one rolling frame + per-issue draws.
- Audit: council rotation draws + each issue-assembly draw + council oversight decisions.
- Suitability: the best-evidenced model in the corpus — Ostbelgien implements it at regional scale.
- Evidence status: implemented (Ostbelgien, regional scale); generalisation to multi-level nesting is theoretical.
Model 5 — Issue-based networks rather than strict geography
Bodies organised around policy domains (climate, housing, care) drawing members across levels rather than per territory.
- Legitimacy: issue-framed; risks looking like stakeholder fora unless sortition discipline is maintained.
- Representativeness: defined against the affected population, not a territory — the “eligible population” itself becomes contested (who counts as affected?).
- Continuity: issue networks persist while issues do.
- Accountability: diffuses to whichever authorities own the issue — often multiple or none.
- Concentration of influence: risk of issue-network capture by engaged stakeholders.
- Professionalisation risk: topic expertise accrues — members may professionalise per issue.
- Institutional memory: issue-level.
- Scalability: flexible; boundaries are semantic not administrative.
- Complexity: high definitional complexity; low administrative nesting.
- Privacy/data: frames defined by issue-relevance are harder to source than territorial registers.
- Audit: eligibility-rule provenance becomes the critical record.
- Suitability: cross-cutting problems (climate is the corpus example — Brussels climate assembly is issue-scoped though territorial, SRC-0132).
- Evidence status: theoretical as a nesting model; issue-scoped single assemblies are implemented.
Comparative summary
| Model | Representativeness | Continuity | Power concentration | Complexity | Evidence |
|---|---|---|---|---|---|
| 1 independent draws | strong per level | none | lowest | low design / high ops | theoretical |
| 2 delegation upward | weakest | strongest | highest | low ops / high legitimacy | theoretical for sortition |
| 3 mixed delegation + draw | moderate | partial | bounded | highest design | theoretical |
| 4 permanent council + issue assemblies | good | high in council | concentrated, rotated | moderate | implemented (Ostbelgien) |
| 5 issue networks | contested | issue-level | stakeholder capture risk | high definitional | partially (issue-scoped cases) |
Cicada requirements
- Represent nesting relationships between assemblies explicitly — including none (Model 1).
- Represent both delegation events and fresh draws as distinct, auditable selection-type events.
- Never silently compose descriptive-representativeness claims upward — a delegated body is not population-representative.
- Record which model a given structure instantiates as a governance declaration, not a derived property.