Software capability landscape
Status: CA-R1 research record. Updated: 2026-10-05. Evidence labels per RESEARCH-METHOD.md. Matrix values: native / partial / plugin-module / external-dependency / not found in reviewed documentation / unclear.
Central question: which parts of the assembly lifecycle already have credible software support, and which are under-served or fragmented? Answers are tied to institutional processes evidenced in research/case-studies/, especially UK-SORTITION-ASSEMBLY-LANDSCAPE.md.
1. What the systems are
| System | Classification | Licence | Maintenance signal | Per-system record |
|---|---|---|---|---|
| OpenDLP | Civic-lottery platform + stratified-selection engine | Apache-2.0 | Young (repo 2025-07), actively pushed (SRC-0073) | opendlp.md |
| Panelot | Stratified-selection engine (hosted; leximin-optimal) | Unverified licence; described as open-source in publications | Service live; repo not located (SRC-0078–SRC-0080) | panelot.md |
| sortition-algorithms (SF) | Stratified-selection library | Open source (repo) | Active; implements the Nature algorithm family (SRC-0077) | covered in opendlp.md/panelot.md |
| Decidim | Multi-purpose civic participation platform | AGPL-3.0 | Very active; large deployments (SRC-0082) | decidim.md |
| CONSUL DEMOCRACY | Mass civic participation / PB platform | AGPL-3.0 | Active; 250+ deployments claimed (SRC-0086, SRC-0087) | consul-democracy.md |
| Loomio | Asynchronous group deliberation/decision tool | AGPL-3.0 | Active (SRC-0089) | loomio.md |
| Polis | Consensus / opinion-clustering platform | AGPL-3.0 | Active; public API (SRC-0091–SRC-0094) | polis.md |
| Stanford ODP | Online deliberation/video + automated moderation | Not found; hosted research infrastructure | Limited alpha (SRC-0095, SRC-0096) | stanford-online-deliberation-platform.md |
| Dembrane Echo | Conversation capture/analysis/reporting | Other/unasserted | Active, young (SRC-0098) | dembrane-echo.md |
2. Lifecycle capability matrix
| Lifecycle stage | OpenDLP | Panelot | Decidim | CONSUL | Loomio | Polis | Stanford ODP | Dembrane Echo |
|---|---|---|---|---|---|---|---|---|
| Eligible population definition | partial (operator-defined targets) | partial (quota file) | not found | partial (verified-user rules) | not found | not found | not found | not found |
| Sampling frame / invitation population | external dependency (address source outside tool) | not found | not found | not found | not found | not found | not found | not found |
| Invitation generation & distribution | partial/external (docs cover invitation ops; postal handled outside) | not found | not found | not found | not found | not found | not found | not found |
| Respondent registration | native (registration page, respondent data) | not found | partial (registration + verification) | native (registered/verified users) | partial (invites) | partial (open link) | unclear | not found |
| Stratified selection | native | native | not found for people (proposal draws only) | not found | not found | not found | not found | not found |
| Reserve / replacement management | native (documented replacement runs) | not found | not found | not found | not found | not found | not found | not found |
| Selection audit / reproducibility | partial (open-source algorithm; seed/witness artefacts not evidenced) | native (computed lottery + public draw) | native for proposal draws (die-roll seed + witnesses) | not found | not found | not found | not found | not found |
| Identity / eligibility verification | partial (operator-managed data) | partial (anonymised IDs) | plugin/module (verification workflows; e.g. CSAM at Brussels) | native (census verification) | not found | not found | unclear | not found |
| Participant onboarding | partial (“keeping warm” docs) | not found | partial | partial | partial | not found | partial (briefing mode) | not found |
| Agenda / topic generation | not found | not found | partial (admin-configured content; proposals feed) | partial (proposals feed process) | partial (group-driven) | partial (seeded statements + open input) | native (convener agenda builder) | partial (theme synthesis) |
| Evidence / submissions | not found | not found | native (proposals, documents, meetings, debates) | native (proposals w/ documents) | partial | partial (statements) | native (balanced briefing materials) | partial |
| Facilitator workflow | not found | not found | partial (admin/moderation) | partial (moderation) | partial (facilitator settings) | partial (moderation queue) | native (automated moderator) | native (capture/reporting for facilitators) |
| Synchronous deliberation | not found | not found | partial (meetings registration) | not found | not found | not found | native (video + queues) | partial (records it) |
| Asynchronous deliberation | not found | not found | native (debates, proposals, comments) | native (debates) | native (threads + decisions) | native (async statements/votes) | not found | not found |
| Proposals / amendments | not found | not found | native | native | partial (proposals; amendments via process) | partial (statements only) | not found | not found |
| Consensus / ranking / voting | not found | not found | native (supports, votes, budgets) | native (supports, votes, PB) | native (proposal/consent/consensus/score/rank/STV) | native (agree/disagree/pass — analytics, not binding) | partial (pre/post questionnaires per Deliberative Polling) | not found |
| Minority / dissent capture | not found | not found | partial (comments, vote splits) | partial | partial (reasons + gradients) | native (opinion groups preserve minority structure) | partial | partial |
| Recommendation publication | not found | not found | native (content publishing) | partial | partial (outcome statements) | native (public reports) | partial | native (reports) |
| Receiving-authority response | not found | not found | partial (process answers/statuses) | partial | not found | not found | not found | not found |
| Implementation / follow-up / accountability | not found | not found | native-partial (accountability Results: % progress, milestones, links) | partial (proposal tracking) | not found | not found | not found | not found |
| Mass public input around a smaller assembly | not found | not found | native (process-level public participation) | native | partial | native (scales to very large N) | partial (public DP events) | native (in-person capture) |
| APIs / export / import / interoperability | partial (CSV in; API undocumented) | partial (CSV in/out) | native (GraphQL + CSV) | partial | partial | native (public API + exports) | not found | unclear |
Values reflect reviewed documentation only; “not found” is not a claim of absolute absence.
3. Important overlaps
- Stratified selection is served twice, at two depths: OpenDLP (full lottery workflow: respondents→selection→replacements) and Panelot/
sortition-algorithms(the mathematical engine: leximin-optimal fairness). OpenDLP itself uses the Sortition Foundation algorithm family (SRC-0076, SRC-0077) — overlap is engine-vs-platform, not duplication. - Mass participation is served three ways with different semantics: Decidim/CONSUL (institutional proposals-votes-budgets inside an authority’s portal) versus Polis (open-ended opinion mapping). They are not interchangeable: portal proposals are content; Polis output is an opinion distribution (SRC-0087, SRC-0092).
- Voting is everywhere but means different things: Consul/Decidim institutional votes; Loomio group decision methods (incl. STV, consent, consensus templates); Polis statement-level agree/disagree/pass feeding clusters; Decidim sortitions — a proposal draw (SRC-0083, SRC-0088, SRC-0090, SRC-0094). Case-study assemblies used secure ballots (CA UK, SRC-0038) — a separate function none of these tools is documented to provide end-to-end for a mini-public.
- Deliberation is split: Stanford ODP covers synchronous facilitated video; Loomio covers asynchronous discussion+decision; neither connects to a selection upstream or an accountability downstream.
- Dembrane Echo occupies the capture/synthesis niche evidenced in the corpus (House of the People, SRC-0067) — a class distinct from all seven named systems.
4. Apparent gaps (carefully qualified)
Within this bounded review, no reviewed system was found covering:
- Sampling-frame acquisition and invitation distribution — always external (Royal Mail PAF letters, National Register draws); software starts at the respondent pool. Confirmed as institutional/ops work in OpenDLP docs and case studies (SRC-0038, SRC-0074).
- End-to-end member lifecycle: recruitment → selection → onboarding → rotation → completion appears only partially in OpenDLP (selection+replacement) and nowhere else.
- Receiving-authority response and implementation tracking as first-class objects: only Decidim’s accountability component approaches it (manually maintained Results) (SRC-0084). Every case-study response duty (Scotland’s statutory 6-month response, Newham’s standing commitment) runs outside any platform.
- Audit-grade selection ceremonies: Decidim’s die-roll+witnesses pattern exists but for proposals; participant-selection audit artefacts (published seeds, witness records, reproducible re-runs) are documented in practice (Panelot live lottery; CA UK open code) but not as a reusable platform feature (partial).
- Offline/non-digital participation: CONSUL’s ballot-boxes/signature-sheets are the only designed offline channel found (SRC-0088); no reviewed tool supports postal/non-internet participation in the selected-assembly lifecycle.
5. The missing connective layer
The documented lifecycle —
population → civic lottery → selected assembly → deliberation →
recommendations → receiving authority → formal response → implementation tracking
— is not supported end-to-end by any reviewed system (Analysis). Instead:
- selection lives in OpenDLP/Panelot;
- deliberation lives in Loomio/Stanford ODP (+ offline facilitation);
- public input lives in Decidim/CONSUL/Polis;
- response/follow-up lives in Decidim-accountability at best, mostly nowhere.
Real assemblies in the corpus assemble this chain with institutional process + glue (Involve liaising members, councils publishing responses manually) — i.e., the connective layer today is practitioner labour, not software (SRC-0038, SRC-0049).
This does not conclude Cicada should build that layer. It records that the gap exists and is currently filled institutionally.
6. Preliminary implications for Cicada
Research synthesis only — not architecture, not adoption.
- Well-served by mature software: stratified selection algorithms (open-source, peer-reviewed); mass public input (Decidim/CONSUL/Polis — three credible, deployable options); small-group asynchronous decision-making (Loomio); synchronous moderated deliberation (Stanford ODP, subject to access).
- Split across unrelated tools: invitation/registration → selection → onboarding; deliberation → recommendation → publication; response → implementation tracking. Chains would require interoperability work or institutional process.
- No strong support found: end-to-end assembly lifecycle management; receiving-authority response obligations; implementation tracking tied to recommendations; audit-ceremony features for selection; offline participation inside the selected-assembly lifecycle.
- Interoperability rather than reimplementation: selection engines (CSV I/O), Decidim GraphQL API, Polis public API — the mature systems have seams worth studying at CA-R3 rather than rebuilding.
- Depends more on institutional method than software: stratification target-setting (who chooses strata — see CA UK governance, SRC-0038), evidence curation (Expert Leads/Advisory/Academic panels), replacement policy, response drafting — these are governance functions that software can record but not own.
- Questions for CA-R3: (1) licence verification for Panelot service and Dembrane Echo; (2) OpenDLP audit/seed/witness model vs Cicada’s determinism requirements (CW-D015); (3) whether Decidim accountability component semantics match “formal response obligation” concepts; (4) whether any tool supports reserve-list rotation and member-term models (Newham carry-over); (5) API depth of CONSUL/Loomio; (6) whether a Cicada-specific lifecycle layer is the actual missing artefact — to be decided, not presumed.