The funnel — the modern era, 2011–2026
"Procedural gates" = screening, deadlines, abandonment. Settlements and withdrawals produce no public decision and are not visible here. Counted by matter.
How matters end
The largest single outcome is nobody answering a letter. None of the first three rows involved any assessment of the evidence — and the last row is unknown, not "live": those matters may have settled, been withdrawn, or continue. The published record cannot say.
Representation is the clearest pattern
Two caveats travel with every number: counsel select cases they judge viable, so representation partly marks case strength; and nine represented matters is a small sample. The direction holds at every gate. Province-wide, 80% of applicants file unrepresented (Tribunal Watch Ontario).
Both sides of the same table
76% of applicants (28 of 37 matters) never had any representative. Of the 19 matters where respondent counsel is recorded, 13 (68%) were defended by lawyers of one police-sector management firm, Johnstone & Cowling LLP — 21 of 27 such decisions. Counted by matter, the firm's counsel defended 13 matters; no applicant-side representative appears in more than 2.
Firm attribution: four counsel names verified on the firm's current published roster; one further counsel sharing the firm name is attributed to it; historical rosters may differ.
The one remedy — and the fight to take it back
Tribunal finds racial profiling. $10,000 awarded. (2015 HRTO 1712)
Tribunal cancels its own finding on the basis of a settlement signed while the decision was on reserve. (2019 HRTO 565)
Divisional Court sets the cancellation aside. The finding is restored. (2021 ONSC 414)
Court of Appeal dismisses the police board's appeal. (2022 ONCA 823)
Supreme Court refuses leave — with costs. The finding and the award stand. (SCC file 40587)
Tribunals do not annotate their published decisions when courts reverse them. Reading the tribunal record alone, this story ends backwards in 2019 — which is why every outcome we publish is joined to court records first.
Two details from the court chain deserve their own lines. The settlement the Tribunal relied on to cancel its finding contained a non-disparagement clause — Durham's entry in the settlement-silencing pattern this project tracks across services (its Thunder Bay mirror is the Burns arc). And at the Court of Appeal, the court granted the remedy itself rather than remitting — because tribunal delay had made sending it back untenable. The province's appellate court, working around the Tribunal's queue, inside the province's only successful police human-rights case.
What people alleged
Reprisal tied at the top is a finding about the complaints system itself: people allege that complaining produced consequences. Grounds not restated in 6 short abandonment decisions — recorded as missing, never as zero.
The early era, 2009–2010: when hearings were fast
The eight earliest decisions — five applicants, ingested from full text in August 2026 — describe a different tribunal. Spears filed in 2009 and had a full merits hearing decided the same December — months, not years, with witnesses examined and credibility weighed (dismissed on the merits). Beyette, a transitional file carried over from the old Commission system, likewise received a full merits hearing (dismissed: the officer's alleged comments not proven). S.M. was dismissed as untimely. Marshall — filed with the old Commission in June 2007 and carried into the new tribunal — reached a merits dismissal in December 2013: six and a half years across two systems. A.F. proceeded through hearing directions in 2011, and then the published record goes silent.
Early-era files include transitional T-/TR- numbers (matters carried from the pre-2008 Commission system). Full-record totals: 60 decisions · 40 matters · 6 merits hearings · 1 remedy. (Marshall and A.F. were already tracked in the relational corpus — their transitional files carried post-2011 decisions — so the early era adds three matters: Spears, Beyette and S.M. A.F.'s two further s.34 files appear only inside a consolidation ruling and are counted as files, not matters.)
Employment versus services
By social area (matter level, multi-flagged): goods, services and facilities — 22 matters; employment — 5; housing — 4; contracts — 2; vocational association — 1; not stated in the decision — 10. This is overwhelmingly a record of civilians and policing services — the mirror image of Thunder Bay, where the service's own members are a major share of applicants. A civilian's complaint about a stop and an officer's complaint about their workplace are different legal worlds; any comparison between services must hold this split constant.
Evidence: the missing camera
0 appearances of body-worn or CCTV footage weighed in evidence across fifteen years. What the Tribunal weighed instead: police notebooks, occurrence and arrest reports, 911 audio, dispatch records, CPIC/PARIS query logs, texts, medical records, mobile-phone video, transcripts, expert witnesses. In the one successful case, the decisive evidence was the contradiction between what officers wrote at the time and what they said at the hearing.
Second looks
Six reconsideration requests appear in this record. Six were denied. Four were decided by the same adjudicator who wrote the original decision — including one where the request alleged bias. Nothing here breaches the Rules: Rule 26 does not require reassignment. That is the point. After a denial, the remaining route is judicial review in Divisional Court — functionally out of reach for most people in this record.
Time: every matter, filing to last decision
Each bar is one matter, from filing (exact where stated in a decision; otherwise the file-number year) to its last published decision. Dots are decisions. Bars are coloured by how the matter ended.
Two clocks run in this record, in opposite directions. Matters that survive take ever longer — the one remedy took ten years from filing to the end of the court fight; Marshall took six and a half to a merits answer. But matters that die now die fast: for matters filed since 2020, the median time from filing to final dismissal is about a year, because they end at paper gates — abandonment and screening — not hearings. Delay did not slow everything down equally. It moved the endings earlier and the answers later.
Filing dates: exact for 22 of 37 corpus matters (stated in decisions); file-number year otherwise; transitional T-/TR- files shown from the 2008 transition or their stated Commission filing date. "Last decision" is not resolution for matters without a final outcome in the record.
The credibility frame
Most matters here turned, at some stage, on whose account was accepted. The corpus's honest lesson is that this is not principally a contest of memory: it is an asymmetry of record-making power. One side generates notebooks, occurrence reports, dispatch logs and query records as it goes — the other side has whatever it thought to write down. In the one successful case, the applicant won because the institution's own records contradicted its testimony. That is why the Records Guide exists, and why one caption-level detail matters: inside the A.F. transitional file sits a hidden co-applicant, S.F., visible only in the full text — captions hide people, which is one more reason counting from captions fails.
What this record cannot show
Province-wide context: the index lists 60 decisions naming Durham (2009–2026); the 8 decisions from 2009–2010 are queued for ingestion and this page's window statement has been corrected accordingly. The 8 decisions from 2009–2010 have now been obtained and ingested (see the early-era section above and corrections). Still queued: Yogarajah v. Tong, 2024 HRTO 393, whose text and outcome we have not yet obtained and do not characterize. See corrections.