The Platform Baseline
What Toronto's 2026 candidates have actually written down — measured, honestly, for the first time. A companion to State of the Election and part of the biweekly record. Baseline edition: August 5, 2026. Province-wide: the Ontario Platform Baseline, every published platform in the province.
1. What we did
On a single day — August 5, 2026 — we took all 223 registered candidates for Toronto's 55 offices (mayor, 25 council wards, 29 school-board seats and counting the French boards' Toronto seats), verified every claimed campaign website by fetching it, searched individually for every candidate who appeared to have none, and captured every published platform, word for word, into a dated evidence record: about 94,000 words of the candidates' own writing. Then we measured each platform on eight dimensions — specificity, fiscal grounding, jurisdictional fit, evidence, implementation, local groundedness, coherence, and time-boundedness — plus how much of the office's real issue space it covers. Twenty percent of the scores were independently re-scored blind as a check on ourselves; the agreement rates are published below (§8).
One rule above all: no position was judged for its direction — anywhere, by construction. A fully specified plan to build bike lanes and a fully specified plan to remove them score identically. This measurement has exactly one thing it can find wanting: the absence of substance. Whether a plan is the right plan is your call, not ours — our job is telling you whether there is a plan to read at all.
2. The headline — the ballot is mostly blank
| What a voter can read | Candidates | Share of 223 |
|---|---|---|
| A real, scoreable platform | 97 | 43% |
| A website with no platform on it (photo, bio, slogan) | 24 | 11% |
| A website our tools can't technically read (held, not judged) | 20 | 9% |
| Nothing — no findable published platform at all (search-verified, one by one) | 82 | 37% |
106 of 223 candidates — 48% of the field — currently offer the public nothing readable about what they would do in office. And among the 97 who genuinely wrote something, the spread is enormous: 26 are little more than named intentions, 29 are thin, 24 are workmanlike, 13 are substantial — and 5 are exceptional (§5). Measured against the full bar a city should be able to expect — strong, comprehensive, actionable, costed, dated — 18 platforms of 223 pass: eight percent.
Why this matters to you: Toronto is hiring 26 executives for a multi-billion-dollar organization and 29 trustees for its school systems, and for half the applicants there is no written application to read. You would not hire a contractor this way — and you don't have to accept it: sixteen days remain for any candidate to publish, and this page re-measures every two weeks. The baseline exists so the city can watch the number move.
3. What the platforms that exist are like
The 97 real platforms tell one consistent story: Toronto's candidates know WHAT they're for — almost none say WHEN, or with WHOSE money.
- The good news, and it's real: the strongest dimension across the corpus is jurisdictional fit (≈3.5/5) — candidates overwhelmingly campaign inside their office's actual powers. The folk theory that municipal candidates promise federal miracles is mostly false in Toronto. Platforms also name real streets, real intersections, real local fights (local groundedness ≈2.9).
- The thin middle: specificity ≈2.5 — named actions, but usually without a mechanism; evidence ≈1.8 — assertion outruns data; implementation ≈1.4 — first-steps gestures at best; fiscal grounding ≈1.4 — money is rarely mentioned with a number attached.
- The rarest object in Toronto politics: a date. Time-boundedness scores ≈0.6/5 across the field — and 0.0 among school-trustee platforms. In July we found that not one Toronto platform in our census committed to a timeline; the deeper measurement confirms it structurally. Almost nobody running for office in this city will tell you what happens in their first hundred days — or any hundred days.
Why this matters to you: a plan without a date and a dollar figure cannot be held to account — there is nothing to check later. The single most useful question any resident can ask any candidate this month is also the simplest: "Which of your commitments will be done by when, and what will it cost?" This page exists partly so that thousands of people ask it at once.
4. Office by office — and the incumbency surprise
- Mayor (37 candidates): 12 scoreable platforms; 16 candidates with nothing findable at all. The strongest platforms in the mayoral field are not where the cameras are (§5).
- Council (132): the healthiest layer — 72 scoreable platforms, and nearly every band from nominal to exceptional is represented. One downtown race carries two independently-assessed exceptional platforms competing head-to-head — the only race in the city where top-tier substance faces top-tier substance.
- School boards (54): the dark zone, confirmed again. 33 of 54 trustee candidates publish nothing at all; the scored remainder averages the thinnest platforms on the ballot; and no trustee platform anywhere in the city contains a single dated commitment. The people seeking control of multi-billion-dollar school systems are the least likely on the whole ballot to write down a plan.
- The incumbency surprise: of the 32 incumbents in the field, only 12 publish a scoreable forward platform — 17 publish no readable forward plan at all (9 with nothing findable, 8 with brochure-class sites). The honest caveat: an incumbent's record IS a public document, and several run on it. But a record is what you did; a platform is what you propose to do next. On the published forward offer, challengers as a class currently out-publish incumbents.
Why this matters to you: the places where scrutiny is weakest — trustee races, safe wards, uncontested name recognition — are exactly where published substance is thinnest. That is not a coincidence; it is what happens when nobody asks. Asking is free.
5. The bright spots — five platforms worth studying
Named for praise only, under our standing rule: strong published substance, confirmed by direct reading, receipts attached. This measures the written offer — it is not an endorsement, a viability judgment, or agreement with any position. We serve every candidate identically, and any candidate can join this list by publishing at this standard.
Gabe Blanc — council, University–Rosedale
Four fully-worked, numbered proposals — a public-seating strategy, a data-centre power-rate discount, streetcar priority, "more homes, lower rents" — each its own policy page with a mechanism and reasoning, not a slogan.
Joe Cadeau — council, Toronto Centre
Accountability planks built on named projects and named numbers ("233 Carlton ballooning from $3.7 million to over $13.5 million") with specific process commitments attached to each.
Chris Moise — council, Toronto Centre
A delivered-results record converted into a specific forward program — and, with Cadeau, one half of the city's only head-to-head contest between two top-band platforms.
Cory Deville — mayor
A staged, dollar-figured policy suite — vertical-farming, rental affordability — far beyond the mayoral field's norm, from a candidate with almost no press coverage.
Georgios Kalkounis — mayor
The single highest-scoring platform in the entire corpus: a 23-page structured plan with costing, sequencing and delivery detail — from a first-time candidate you have probably never read about.
Notice what just happened: two of the five best-written plans in Toronto's 2026 election belong to mayoral candidates the coverage has entirely passed over. Substance is where the spotlight isn't — which is an argument for reading, not headlines, as the way a city chooses.
6. If you're a candidate: how to move your row
Every number on this page can change before August 21 — and after it, every two weeks. What the strongest platforms in your own city do, in plain words: they say what (specific, verifiable commitments), how (the mechanism and first steps), how much (a number and where it comes from), by when (dates — the rarest and cheapest upgrade on this list), within your office's real powers (voters notice, and so does our rubric), and why here (this ward's actual live debates). Publish it anywhere public and we will find it, read all of it, and update your row — or tell us directly via the corrections page and we'll re-measure within the cycle. Thinking about running at all? The papers take an afternoon; here's everything you need. Every candidate in every race gets identical treatment from us: same rubric, same corrections door, same welcome.
7. If you're a voter: how to use this
Three uses, in rising order of power. Read your race: the State of the Election page maps every contest; your ward's candidates' own platforms are the primary documents — read them before anyone's coverage of them, ours included. Ask the question: "by when, and what will it cost?" — in person, at the door, by email; candidates answer what enough people ask. Watch the number: this page re-measures every two weeks until the election. If Toronto's candidates publish, the 48% falls and we will report it falling, by name in praise and in aggregate everywhere else. Growth is the headline we are hoping to write.
8. How we measured — and how to check us
The instrument, in brief: every platform was scored 0–5 on eight anchored dimensions — specificity · fiscal grounding · jurisdictional fit · evidence & reasoning · implementation detail · local groundedness · coherence · time-boundedness — plus issue-breadth measured against the office's own issue space (a trustee is never measured against transit; a councillor never against special education). Every score of 3 or higher must carry a supporting quote from the platform itself; every score of 1 or lower must state what was absent. Composite bands: nominal · thin · workmanlike · substantial · exceptional.
The checks on us: a blind, stratified 20% sample was independently re-scored by a second assessor — class agreement 23 of 24; every dimension ≥95% agreement within one point; mean composite divergence 0.25 on a 5-point scale. Scoring is AI-assisted under a versioned, human-designed rubric with human adjudication — disclosed here as everywhere (how this was made). Aggregate statistics publish; per-candidate scores do not — publishing a comparative ranking of named candidates is a line we do not cross by policy, and the only names on this page appear in praise.
Limitations, stated by us first: we measure the web-public record only — print literature and doorstep conversations are invisible to us; twenty sites defeated our capture tools and are held as unknowns, never counted absent; nine platforms were scored on slightly older captures where live sites resisted reading (flagged, dated); the field itself grows until August 21. And the deepest one: a great platform is not proof of a great officeholder. It is the one part of the offer a voter can actually read — and the part a city is entitled to demand more of. That demand is this project.
The number to beat is 48%
Half of Toronto's candidates have published nothing you can read. Sixteen days remain before the field is final, and eleven weeks before the vote. If you're a candidate: publish — we will find it and say so. If you're a voter: ask "by when, and what will it cost?" And whoever you are: register for the conversation — the thirty-second act this whole project is built around, open to every person in this city whether you can vote or not.
Found a mistake?
That's the system working. Every figure on this page regenerates from a receipt, every quoted phrase traces to a candidate's own published text, and every correction is published, dated, and left visible. Candidates: this includes your row — tell us what changed and we'll re-measure. The corrections page takes thirty seconds.