Lifecycle composition options
CA-R3 record. Updated: 2026-10-05. Candidate ways the CA-R2 lifecycle could be satisfied by composing Cicada with existing systems. No composition is chosen here — selection belongs to CA-A1 + operator decision.
Composition A — Cicada lifecycle + OpenDLP + external deliberation tools
Cicada owns population/frame/spec/audit + post-selection records; OpenDLP runs the respondent→selection→replacement workflow; Loomio (or similar) covers async member deliberation; mass input via Polis/Decidim.
- CA-R2 coverage: good at selection front; gap at response/implementation tracking; agenda/evidence thin.
- Coupling: moderate — one GPL-3.0 engine under OpenDLP (Apache-2.0); SF roadmap dependence.
- Fragmentation: moderate — member state lives in Cicada + Loomio + OpenDLP.
- Auditability: strong potential if Cicada binds spec/run/witness; weakest at cross-system joins.
- Offline: invitation + CONSUL-style channels; member side web-first.
- Burden: 2–3 platforms + Cicada.
Composition B — Cicada lifecycle + Decidim participation/accountability + external selection
Decidim carries the public participation envelope (proposals, meetings, results tracking); selection engine consumed directly (sortition-algorithms) or via OpenDLP.
- CA-R2 coverage: best at agenda-adjacent public input + result tracking; selection still external.
- Coupling: heavy — AGPL monolith; Cicada-hosted modifications carry copyleft obligations.
- Fragmentation: high — Decidim is a platform-with-opinions; reconciling its objects to assembly records is the integration cost.
- Auditability: Decidim records exist; response-object semantics still Cicada’s.
- Offline: Decidim has none native; CONSUL is the offline-equipped alternative.
- Burden: highest operational weight of the options.
Composition C — mostly-integrated external stack, thin Cicada coordination
Cicada = coordination/provenance ledger recording references into external systems that do the actual work.
- CA-R2 coverage: weak at anything nobody does (spec artefact, witness bundle, response chain).
- Coupling: extreme — every lifecycle fact lives in a third party.
- Fragmentation: worst — the connective layer the corpus found absent is exactly what this outsources.
- Auditability: poor — provenance crosses system boundaries by assertion only.
- Portability: worst — exit means rebuilding records.
- Verdict: undermines the CA-R2 record contract; listed for completeness, not recommended.
Composition D — Cicada-native lifecycle records + optional adapters to external tools
Cicada implements the full record model (population→response tracking) as first-class data; external tools attach via seams as channels (a Polis conversation, a Loomio group, an OpenDLP run) rather than system-of-record.
- CA-R2 coverage: complete-by-construction at the record layer; deliberation surfaces remain best-of-breed.
- Coupling: lowest — adapters optional, removable.
- Fragmentation: lowest — one canonical record, external copies are views.
- Auditability: strongest — provenance never leaves canonical store.
- Burden: largest build surface; but aligns with CA-R2’s finding that the record contract is the thing nobody else provides.
- Offline: easiest place to honour offline-parity (records are ours).
Composition E — minimal Cicada record/audit layer; process mostly offline/manual
Cicada records specs, runs, rosters, decisions, responses as signed artefacts; institutions execute via paper/email/whatever tools they already use.
- CA-R2 coverage: covers the audit spine only; participation UX left to institutions.
- Coupling: none.
- Burden: smallest build; largest institutional labour requirement (the corpus’s status quo).
- Fit: coherent as a v0 — the “audit ledger first” posture; consistent with evidence that practitioner labour currently fills the connective layer.
Comparative summary
| A | B | C | D | E | |
|---|---|---|---|---|---|
| CA-R2 coverage | med | med | low | high | low-med |
| Build surface | med | med | low | high | lowest |
| Vendor coupling | med | high | extreme | low | none |
| Data fragmentation | med | high | worst | low | n/a |
| Auditability | med | med | poor | high | med |
| Offline viability | med | low | low | med | n/a |
| Portability | med | low | worst | high | high |
| Ops burden | med | high | low | med | low |
CA-A1 questions this frames
- Is Cicada a system of record with channels (D) or a coordinator of platforms (A/B)?
- Does v1 need public participation surfaces at all, or only the member lifecycle?
- Where does the witness ceremony live — and is it digital, physical, or hybrid?