How it works, in plain terms

Neighbors read one statement at a time and choose agree, disagree, or pass. Anyone in the Triangle can vote from a phone. In-person sessions add small-table conversations that go deeper than a vote can.

Organizers curate the seed statements, host the in-person tables, write up findings, and hand results to local officials. A human reviews every published summary.

Software records each vote as it is cast, groups similar answers to surface patterns, and renders live counts. Software never edits, invents, or paraphrases what a person said. Draft summaries are always reviewed by a person before publication.

Statements in the vote are ordered by how divided neighbors are on them — closest to a 50/50 split come first. Fatigue is real; if it hits, we want to have already captured where you stand on the questions that actually distinguish positions.

The “Where people who disagree still agree” section splits participants into two halves based on how they voted on the single most divisive statement, then lists statements that draw high support in both halves. This is a simple approximation of the bridge measure Pol.is and vTaiwan use to surface consensus across camps. With a larger sample we will move to the full vote-vector clustering.

When software generates a summary of what neighbors said, participants get to vote on whether that summary represents them fairly. Disclosure is passive; auditing is not. The affordance is built in — every published synthesis carries a “Does this represent what you said?” control, and the recorded agreement / disagreement is reported alongside the synthesis itself.

Every result reports the number of votes, the number of distinct voters, and the number of verified participants — separately. Demographic breakdowns are published with the non-response rate for that field.

Anyone can download the raw, de-identified vote data as CSV from the results page. The statement seed set, session materials, and analysis code will be published as they stabilize. Requests for the full audit trail: hello@carolinacommons.org.

Until the round is officially launched, aggregate views include a set of clearly-marked synthetic participation, so the site can be evaluated as a working product. Every synthetic row carries a flag and is removed in a single operation before launch. Real votes are never mixed in with the synthetic sample.