We stand on a lineage.

Real-time opinion mapping. Participants write short statements, vote agree/disagree/pass on others, and a UMAP/PCA projection surfaces opinion groups and "consensus" statements (statements that cross groups).

The core insight that clustering opinion into groups reveals cross-group consensus that surveys miss. Our LRD structure — strong consensus, conditional, divergence, gaps — is downstream of the Pol.is "group-informed consensus" pattern.

Pol.is captures short written statements. Civic Filament captures a 3–7 turn conversation where the AI mediator elicits underlying values and needs. Higher fidelity input, but the trade-off is more expensive per participant.

Multi-stage deliberation combining Pol.is, in-person meetings, expert consultation, and government commitment to act on rough consensus. Ran on national policy (Uber, alcohol sales, cyberbullying, telemedicine).

The proof point that structured deliberation can influence real legislation. The "rough consensus" posture (don't optimize for unanimity, name divergence explicitly) shaped how we designed the four-section LRD.

vTaiwan is a process orchestrated by humans across multiple tools. Civic Filament is a single-app protocol. We are not a substitute for the vTaiwan model — we're a component that could plug into it.

Full participatory-democracy platform: participatory budgeting, proposals, debates, assemblies, initiatives. Ruby on Rails, GPL-licensed, ~500 deployments as of the last public census.

The reminder that "participation" is a bigger stack than "consensus." Decidim handles the entire proposal → deliberation → decision → implementation lifecycle. We handle one node in that stack (elicit + synthesize) very well.

Decidim is process infrastructure — the software equivalent of a city clerk. Civic Filament is intelligence infrastructure. The two are complementary and would compose.

SaaS community-engagement platform: idea submission, surveys, participatory budgeting, native input, an AI Copilot summarizing responses.

The commercial validation that municipalities will pay for engagement software — and the shape of the buyer (city clerk, engagement officer, communications lead).

CitizenLab is a general-purpose engagement platform with AI summarization bolted on. Civic Filament is an LLM-native protocol where the interview itself is the AI, and the output is a versioned consensus document, not a summary of comments.

Pro/con deliberation on individual statements. Users place their opinion on a 0–100 slider and structure it with pros and cons.

The reminder that even one dimension of nuance (a slider vs. a binary vote) meaningfully changes the quality of the signal.

Still position-based (agree ↔ disagree on a stated proposal). Civic Filament works one level below — extracting the values that generate the position, so a shared value can be found even when the positions diverge.

Small-group decision-making for co-ops, unions, and collectives. Threaded discussion + structured proposals with block/agree/abstain/stand-aside.

The pattern that groups need structured decision moments, not just endless discussion. Loomio's "block" option is a rich version of divergence-as-first-class-signal.

Loomio scales to dozens; Civic Filament targets thousands. Different problem shape.