How the One Big Conversation week works
No one convenes this. It’s the week every room already talking — a dinner table, a congregation, a block circle, a campus club — points at the same instrument, and one synthesis publishes what the city actually said.
Nobody convenes it — and that’s the design, not a dodge
The One Big Conversation has no convenor, no stage, and no headline event with one organization’s name on it. It is a week when the rooms already talking point at the same instrument: every kit room, every congregation, every block circle, every campus evening feeds the same harvest and the same city-scale conversation. We publish the window, the tool, and the kits — every gathering inside the window belongs to whoever holds it, in their own name. If anything about this page ever reads like there’s a host organization behind it, that’s a bug, and it gets fixed on sight.
How it’s built — independent voices first, deliberation second
- Independent inputs first. The starting statements come from the summer’s harvest — visions and complaints, in participants’ own words — plus what assembly-kit rooms already recorded. Voting opens only once that starting set is fixed and published with its own sourcing. People can keep adding statements throughout, through a reviewed queue. Everyone votes alone, on their own screen, before and independent of any room’s discussion — the same silent-cards rule the assembly kits already use, run at city scale.
- Deliberation second. During the window, rooms can run their own kit sessions on top of the emerging map: pull up the live conversation, discuss it, then vote and add statements as individuals. Rooms deliberate; only individuals vote. No room ever votes as a bloc.
- Dissent stays on the record. The published synthesis carries a “Where we divide” section: agreement rates on the statements that split opinion, both sides told in participants’ own words, minority statements printed in full. Consensus is never manufactured by leaving something out.
The calendar
- August 15–23 — a small pilot, live now: a real trial run of this exact process on one question, testing the mechanics before the city-scale week. See it running.
- Late September — the starting statement set for the citywide conversation is frozen and published, with its sourcing, ahead of the window opening.
- October 7–14 — the window. The city-scale conversation runs live alongside kit rooms everywhere, with a public counter.
- October 15–18 — synthesis, done in the open: the export is pulled, checked, and turned into a human-edited record with receipts.
- Monday, October 19 — the State of the Conversation publishes, and the same plan goes to every candidate on every ballot with one identical question.
- Friday, October 23 — candidate answers close.
- Saturday, October 24 — the adoption record publishes: every answer, verbatim, equal space, alphabetical; silence listed as “no response by [date],” never characterized.
- Monday, October 26 — election day, and the mid-sprint scoreboard.
How the record stays honest
Every line in the published synthesis has to trace back to something real, or it doesn’t print. A statement carries its own agreement rate, overall and by group. A harvest line resolves back to the words that built it. A room’s own three-part record — submitted and printed in that room’s own name — counts as a source too. No traceable source, no line in the record.
Agreement rates are the finding; opinion-group maps are only illustrations. We will not publish a line like “Toronto has three camps” — that is never what the data actually says.
AI assists with the clustering pass; a named human editor signs the synthesis; and the method — what went in, how it was grouped — publishes alongside the result, so anyone can check whether their own statement landed where they’d expect it to.
And if you contributed a line to the vision work this summer, you will be able to find your own sentence’s place in the published plan. That receipt was promised at the start, and this is where it’s kept.
Counting without watching anyone
Everything here is counted in aggregate, never as a tracked individual. What’s counted: conversations held, harvest items collected, votes cast, assembly records submitted. An online conversation counts the same way — a thread on any platform, submitted through the same room flow, with at least two participants and one harvest item, reported by its own host, and reported as its own line rather than folded silently into the rest.
Beyond that, a wider reach — a million or more people discussing this, two million the stretch — is only ever published as a labeled target, never a claim: instrumented from our own pages only (no ad trackers, ever), public share counts platforms already expose, and an honestly kept media log. If this page ever says a million conversations, that is an invitation and a stretch goal, in those words, not a number we are claiming already happened.
The public counter opens at zero when the window opens — a real moment, not a warmed-up number — and any threshold gets disclosed after the fact, never staged in advance.
What this hands the candidates
What comes out of this window isn’t only a plan — it’s a question, sent to every candidate in every race, the same day, the same way:
“The people’s plan for Toronto was built from [N] conversations across the city. Will you commit to implementing it — and which parts? Your answer publishes verbatim, alongside every other candidate’s, in equal space.”
[N] resolves from the public counter at send time — never typed in by hand.
Every candidate receives identical text, the same day, through the same channel. Every answer publishes verbatim, equal space, alphabetical. Silence is recorded as “no response by [date]” and nothing more — never characterized.
Publishing that record is ours to do. Anything that reads as “vote for the ones who answered” is a different, harder question, held on its own gate until it clears — adoption, not endorsement.
The risks, named honestly
A city-scale conversation tool needs a real crowd to mean anything — a thin week produces a thin map. The mitigation is that the window rides the kits and the rooms already meeting, not a standalone ad campaign, and the pilot above is the real test of whether that conversion actually works.
Any open conversation tool can be gamed by a coordinated push. We check for that with more than one signal — a verified track and a separate open track, reported apart, plus a person watching for anomalies, not an algorithm alone.
And the deepest risk in this whole genre, visible elsewhere: a citizens’ assembly or a mapping exercise that ends at a report nobody acts on. Ours does not end at a document — it closes at the adoption record, at the ballot, and at a year the coalition actually signs up to run. The synthesis has two named recipients: the candidates, who receive one equal question, and the thousand rooms that built it, who get the record handed back.