The Commons

For builders & agents

Everything on this site is rendered HTML, built for reading in a browser. The substance behind it lives in source form, one public repository: github.com/unitetolove/commons — sealed research markdown, cleaned datasets, frameworks, and the written policy pages, mirrored one-way out of the working estate (how this was made has the fuller story).

If you are an AI agent, scraper, or researcher and want the cleanest text — no HTML chrome, no rendering layer between you and the source — start with the repo directly, not this page's own prose. Read the repo's README for the current shelf, or clone https://github.com/unitetolove/commons and read the markdown in place. This site also publishes a llms.txt and llms-full.txt (the llms.txt convention, llmstxt.org) as a machine-oriented index of both this site and the repo.

For the day-to-day planning and review documents behind this project — not just the sealed research and finished pages — see the working notes shelf.

What's in the repo

Start here

• docs/methodology.md — How claims are sourced and verified
• docs/ai-disclosure.md — How AI assistance is used and disclosed
• docs/corrections-policy.md — How corrections are handled
• docs/funding-transparency.md — Funding transparency
• docs/how-we-share.md — What publishes here and when

The rule this repo follows

Nothing is "shared except…"; everything is "unshared unless listed." A manifest gates every file that leaves the working estate; a screening pass blocks filesystem paths, private patterns, and internal-process vocabulary before anything is pushed. What's here is safe to read, fork, and cite. Corrections and improvements: see the corrections log, or email directly.

Open data

The cleaned civic datasets behind this site — downloadable as plain CSV files, documented field by field, with the original publisher and known caveats stated for every one. Reuse them, check them, build on them.

This page replaces the earlier “coming soon” placeholder. Seven datasets are live below, each one CSV file. Six of them share the same structured local-statistics shape — populations, municipal finances, rental markets, homelessness counts, collisions, and municipal council resolutions — that back the claims made elsewhere on this site: every row in those six is keyed to a geography (Statistics Canada’s Standard Geographical Classification, where the dataset uses one) and a period, so the same figure can be checked against its original publisher independently of anything we say about it here. Files were assembled for this page on 2026-08-05; each of those six datasets’ own retrieval date is stated in its section below and, for five of the six files, repeated on every row (retrieved_date column). The seventh, added 2026-08-16, is different in kind — an entity registry, not a statistical series — see #7 below.

Not included here: this page covers seven datasets only. A separate internal working folder also contains quote-verification logs used to fact-check narrative pages elsewhere on the site (matching quoted text back to its source paragraph). Those logs are an editorial QA tool for our own writing, not a civic dataset in the sense of the seven below, and aren’t published here for that reason.

1. Ontario population by census division, 2001–2025

What it is: Annual population estimates for every county- or region-level jurisdiction in Ontario (Census Division, or “CD”) — the number StatCan itself currently stands behind for a place’s population, not the once-every-five-years Census count.

• Geographic coverage: all 49 Ontario Census Divisions.
• Temporal coverage: 2001–2025, one estimate per year (July 1 reference date).
• Vintage / as of: retrieved 2026-07-12. StatCan’s own release window for the 2025 estimates was confirmed via a StatCan Daily release dated 2026-01-14.
• Source & provenance: Statistics Canada, table 17-10-0152-01, “Population estimates, July 1, by census division, 2021 boundaries,” pulled via StatCan’s own Web Data Service API and filtered to Ontario. Two rows spot-checked against a third-party demographic reference (citypopulation.de) matched exactly.
• Known caveats: CD (county/region) level only — no per-municipality breakdown in this file. An earlier candidate table (17-10-0139-01) was found to be inactive/superseded before this one was selected; that correction is recorded, not hidden.
• Download: ontario_cd_population_2001-2025.csv — 1,225 rows × 9 columns.

2. Municipal finances — Ontario Financial Information Return, 2024

What it is: Revenue figures for Ontario municipalities’ 2024 fiscal year, drawn from the Financial Information Return (FIR) every Ontario municipality is legally required to file annually.

• Geographic coverage: 402 Ontario municipalities and upper-tier counties/regions (both CD and CSD level — two-tier areas, like a county and the municipalities inside it, each file their own FIR and both appear separately).
• Temporal coverage: fiscal year 2024 only — the most recently complete filing year at the time of retrieval (2025 filings were still arriving; too incomplete to include).
• Vintage / as of: retrieved 2026-07-12.
• Source & provenance: Ontario Ministry of Municipal Affairs and Housing, EFIS (Financial Information Return) open-data bulk export. fir_total_revenue is the FIR’s own Schedule 10 “Total Revenues” line. fir_own_source_revenue is that same figure minus Schedule 12’s four senior-government-transfer lines — the FIR has no line item literally called “own source revenue”; this is the standard net-of-transfers calculation used in municipal finance studies, applied consistently to every row.
• Known caveats: the City of Hamilton is absent from this file (and from the province’s own 2023 and 2025 exports, checked directly). Every other single-tier city we checked is present. A documented, independent city report ties Hamilton’s delayed financial statements to a 2024 cyberattack, which plausibly explains a stalled FIR filing — not proven as the cause, but a real, externally-corroborated gap rather than a silent hole. Beyond Hamilton, only a subset of these 402 municipalities has a written, line-by-line verification note on file; the rest share the same bulk-export pull and the same calculation method above, applied uniformly.
• Download: fin_fir_2024.csv — 804 rows × 9 columns.

3. Rental market — CMHC, October 2025

What it is: Rental vacancy rates and average rents from CMHC’s Rental Market Survey, the standard federal source for this figure.

• Geographic coverage: 21 geographies. Some are a single municipality (e.g. Toronto, Guelph); others are CMHC’s own metro-area (“CMA/CA”) geographies used as the closest available stand-in where CMHC does not publish a municipality-specific figure — see caveats.
• Temporal coverage: October 2025 survey (a single period).
• Vintage / as of: retrieved 2026-07-12.
• Source & provenance: Canada Mortgage and Housing Corporation (CMHC), Housing Market Information Portal, Rental Market Survey Data Tables.
• Known caveats: 10 of the 21 geographies in this file are explicit proxy rows, flagged in the proxy_note column, because CMHC has no native geography matching the area in question (e.g. Peel, York, Halton and Durham Regions all share the same “Toronto CMA” figure in this file, since that metro area spans all of them — the figure describes the metro area, not any one region specifically). The quality_flag column carries CMHC’s own reliability rating for each figure: a = good, b = moderate, c = low (small sample, typically a small municipality).
• Download: housing_cmhc_2025-10.csv — 42 rows × 11 columns.

4. Homelessness point-in-time counts, 2024

What it is: Single-night enumeration counts of people experiencing homelessness, organized by Ontario’s Service Manager areas (the regional bodies responsible for housing and homelessness services).

• Geographic coverage: 15 of Ontario’s roughly 47 Service Manager areas — the highest-population ones.
• Temporal coverage: the Fall 2024 count cycle; exact count dates vary by area (mostly October–November 2024) and are given in each row’s period and notes fields.
• Vintage / as of: retrieved 2026-07-12 and 2026-07-13.
• Source & provenance: unlike the other datasets on this page, there is no single provincial or federal publisher for this data — we checked directly and no dedicated Ontario open-data catalogue entry for point-in-time counts exists. Each row is instead sourced to that specific Service Manager’s own published count report (a city or regional government page or council report). See each row’s notes field for source-specific detail.
• Known caveats: this is the least uniformly sourced dataset on this page — counting methods and what is included (e.g. whether asylum claimants are counted) are not standardized between areas, and each row’s notes field states area-specific limits, for example: the Windsor-Essex figure is not council-endorsed per that city’s own council minutes; the London-Middlesex figure covers the City of London only, with the county’s rural portion unconfirmed; the Peel Region figure includes asylum claimants, with the non-asylum-claimant figure mentioned in the source but not independently confirmed. One row (Peel Region) required a minor punctuation fix on our end so the file parses cleanly as CSV — the note text itself contained an unescaped comma; no data value was changed.
• Download: housing_pit_2024.csv — 15 rows × 10 columns.

5. Police-reported collisions

What it is: Annual counts of police-reported injury and fatal motor-vehicle collisions, drawn directly from each police service’s own open-data feed.

• Geographic coverage: two municipalities only — Toronto and Ottawa. This is not a province-wide dataset.
• Temporal coverage: Toronto 2020–2023; Ottawa 2019–2024 (see caveats for the gaps inside those ranges).
• Vintage / as of: retrieved 2026-07-12 and 2026-07-13.
• Source & provenance: Toronto Police Service open-data portal (the TOTAL_KSI dataset) and City of Ottawa open-data portal (Traffic Collision Data), both queried directly from their live data feeds.
• Known caveats: Toronto’s 2019 is excluded — the source feed’s collision-identifier field is entirely blank for that year, which would otherwise silently produce a false count of zero rather than a real one, so it was left out rather than reported wrong. Ottawa’s 2023 genuinely has zero matching rows in the live source feed itself — a real gap in that dataset (most likely a records-system transition), not a fetch error on our end, and left in the file rather than smoothed over.
• Download: police_collisions.csv — 20 rows × 9 columns.

6. Municipal council asks to the province and federal government

What it is: A catalog of Ontario municipal council resolutions and motions that formally ask the provincial or federal government to act, across five issue areas: housing supply/affordability, homelessness/encampments, community safety/policing, mobility/congestion/transit, and property tax/municipal finance. Each row describes an institutional act (a council resolution), never an individual politician’s personal position.

• Geographic coverage: 46 Ontario municipalities.
• Temporal coverage: resolutions from 2021-06-16 onward; a handful of rows have no specific date in the original source and are marked accordingly.
• Vintage / as of: this file has no single retrieval date — each row instead carries its own source_url as its provenance anchor, fetched individually.
• Source & provenance: the Association of Municipalities of Ontario (AMO) resolution/advocacy trackers, the Federation of Canadian Municipalities (FCM) resolutions database, and individual municipal council-minutes/agenda archives, named per row in the source_url column.
• Known caveats — flagged visibly: unlike the other five datasets on this page, this one is not documented in our data schema record; its methodology instead lives in a separate internal working note, not a formal schema. The field meanings below come from that note and the file’s own header, not from the same schema document the other five datasets cite. Of the 81 rows, 9 are directly fetched and confirmed (tier=1) and 72 are corroborated via search rather than a direct fetch of the primary document (tier=1-search-verified) — see the tier column on each row.
• Download: municipal_asks.csv — 81 rows × 8 columns.

7. Toronto civil-society organization registry

What it is: A directory-style registry of Toronto-area civil-society organizations — neighbourhood and tenant associations, shelters and settlement agencies, faith and cultural umbrellas, fraternal orders and service clubs, hubs and funders, and dozens of other categories — behind this site’s Organization Directory. This is a different kind of file from the six above: an entity list, one row per organization, not a statistical time series.

• Geographic coverage: Toronto and the wider Greater Toronto Area, unevenly — most rows are Toronto proper. Province- or nationwide bodies with a Toronto presence are included and flagged as such in the area column rather than left out.
• Temporal coverage: not applicable in the sense the six datasets above use it. Each row instead carries its own last_verified date where a verification pass has actually touched it; rows added but not yet individually re-checked carry no date.
• Vintage / as of: 2026-08-18.
• Source & provenance: this is our own original field research, not a re-publication of an existing directory. Rows were assembled from roughly 55 documented sources — city and provincial open-data portals, community-hub and grant-recipient lists, umbrella-organization member rosters, and direct web verification — catalogued in this project’s own source inventory. Each organization is minted once into a single central registry with a permanent org_id, through a single-writer process (no parallel or duplicate lists), so the same organization is never counted twice under two different names.
• Known caveats: coverage is partial and uneven by category. We track how complete each category is against an estimated citywide total and keep closing gaps as we find them — some categories (established fraternal lodges, city-run community hubs) are close to fully counted; others remain order-of-magnitude estimates with real, acknowledged gaps still open. area is free text, not a structured ward or riding join — values range from a ward name to a full qualifying sentence. Behind the register there is now a separate structured geography table that resolves 14% of all recorded organizations to a specific Toronto ward (most placements instead land at city or GTA-region level), and about 16% of the rows in this file are placed that way, so the free-text column is still the honest thing to hand you. Verification dates vary widely: most rows are marked verified-live with a recent date, and the rest are seed-quality (added but not yet individually re-checked) or already known to be dormant, unreachable, or defunct — each labelled by its own verification_status rather than smoothed over. This download is not the whole register, and here is exactly what is missing. Behind these 2,863 published rows sit a further 17,210 organizations we have recorded but are not publishing: 16,976 found through a weaker verification path and not yet individually re-checked, 142 we could not reach, 37 we believe have closed, 32 named but unconfirmed, 17 apparently dormant, and 6 confirmed closed with no successor. That withheld number first grew sevenfold in a single day on 16 August 2026, when we pulled eleven of the City of Toronto’s own open-data registers — every licensed child-care centre, every EarlyON site, every shelter location, every music venue, every community-grant recipient, and a 2006 survey of places of worship — and added 5,679 organizations to the register at once; it has kept growing since, through ward-by-ward search passes across Toronto’s 25 council wards and further register pulls. Almost none of the withheld rows are run by the City itself, so a register listing tells us the organization was listed, not that it is open today. Until we check each one directly, it stays recorded and unpublished. We hold those back deliberately. Publishing “this organization has shut down” about a group that is in fact still running would be a claim we could never take back once this file has been copied elsewhere, and our confidence in those rows is not yet high enough to make it. So this file carries only what we have confirmed directly, and matches the Organization Directory page row for row — the two can never disagree about what we claim to know. This is a snapshot of a living city — organizations open, close, move, and rename continuously, and this file will always be somewhat behind that reality.
• Licence: the same as every dataset on this page — CC BY 4.0 (see License below). This is our own original research, so there is no separate upstream licence to reconcile.
• Download: tribes_orgs.csv — 2,863 rows × 8 columns.
• Query it without downloading: open in Datasette Lite — it runs entirely in your browser (WebAssembly), so there is no server and no account, and it is safe to click.

License

These datasets are licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0). Share and adapt this material for any purpose, including commercially, with attribution. Several source datasets carry their own upstream terms this license does not override — Statistics Canada, CMHC, the Ontario Financial Information Return, and municipal/police open-data portals each keep their own licence for the original data they publish; attribute both this project and the named upstream source per each dataset's own caveats above.

These datasets are version-controlled in the open at github.com/unitetolove/commons, alongside a data dictionary mapping each file here to its source, row count, and columns (data/README.md). Found a problem with a dataset? Open an issue there using the Dataset problem template.

The frozen research editions — homelessness

Two complete, sealed research components publish their machine-readable spines alongside their pages: the frozen v1 edition (July 2026) — a 770-claim ledger with per-claim source URLs and verification paths, its source registry, and its data series — and the CSRL v2 trust layer — a 3,071-row claim ledger (1,243 claims verified against primary sources, the rest disclosed as pending) and its source registry. Frozen editions: figures stay as published, with dated margin notes where our living research has since moved.