uniteTOlove.ca
FROZEN EDITION — July 2026. This page is part of the v1 Toronto homelessness research library, a complete, sealed project frozen at its July 2026 figures (92% of its 770 claims individually verified against primary sources). It is a product of its July-2026 campaign context, published as a citable historical artifact. Where our newer research disagrees, a dated margin note points at the living page — the frozen text is never silently rewritten. Why there are two editions.

LEGAL / DEFAMATION STRESS-TEST — FINDINGS

Regenerative Toronto | July 2, 2026

Document ID: MAIN-AUDIT-002 Executed against: MAIN_REVIEW_004_Legal_Stress_Test_Prompt.md (created this pass — the deep version of REVIEW_001_External_Review_Prompt §7, which was one sentence). This stress-test prepares material for the Tier 0 lawyer reviews; it does not replace them. Every conclusion below is a triage assessment for counsel, not legal advice. Method: pattern-priority sweeps (P1–P6 per the prompt) across all public research and DERIV documents: named individuals × dishonesty imputations, named organizations × impropriety framing, unhedged causation, litigation/appeal framing, creative assets, republication discipline, plus direct verification of the standing trafficking boundary and existing legal hardening.

HEADLINE ASSESSMENT

The library's legal discipline is genuinely strong — materially stronger than its citation bookkeeping was before AUDIT-001. The highest-risk patterns a plaintiff-side reader hunts for came back clean or already-hardened: zero instances of a named individual adjacent to an imputation of dishonesty; zero instances of a named operator adjacent to impropriety framing; zero unhedged causation-of-death claims; exemplary sub judice discipline on every live matter checked. The risk that remains is concentrated in one pending editorial decision that now has a legal dimension ("poverty industry") and in process items for counsel, not in publishable defamation exposure sitting in the text today.

FINDINGS BY SEVERITY

BLOCKING — unchanged, confirmed still correct

The existing Tier 0 gates (lawyer review of DERIV-007 and HMK-007) remain the only true publish-blockers, and this stress-test found nothing that should be added to blocking status. It did confirm the gates are doing real work: HMK-007's Waterloo treatment carries the under-appeal caveat in bold at every citation site across four documents, with explicit "not settled law / never cite without this caveat" instructions — the appeal-status discipline held through all subsequent edits.

FIX — one item, elevated into the existing pending decision rather than resolved

"Poverty industry" now has a legal read attached to the editorial call (HMK-001 lines 57, 301, 1155). The term targets a system — no identifiable plaintiff on its face, and identification is an element of defamation. But a plaintiff-side argument exists: HMK-001 elsewhere names specific operators and their executive compensation, and an innuendo claim would argue readers understand the named operators to be the poverty industry. Two of the three instances are also comment in factual register ("has calcified into a poverty industry" — line 57 states a characterization as a fact). If the framing is kept, the fair-comment defence is materially strengthened by two changes: (1) anchor the term at first use to the documented facts it comments on — the per-diem incentive structure and Gaetz's independently published academic critique ("shelters are rewarded for keeping people homeless"), which gives the characterization scholarly provenance the campaign didn't invent; (2) keep it strictly system-level — never within the same passage as a named operator. If the framing is dropped (per HMK-037's deliberate avoidance), the inconsistency resolves itself. Either way this remains the user's call — the legal read is now part of the decision's inputs, recorded in TRACK_001_Master_Issues.

NOTE — defensible as written; brief counsel on each

  1. Executive compensation (HMK-005). Already model practice: an explicit legal note ("All data is from public CRA filings and public Sunshine List disclosure. No allegations of wrongdoing. All claims are about taxpayer value, not fraud") plus a footer recommending legal review before publication. Truth defence covers the figures (public statutory disclosure); the residual risk is juxtaposition innuendo, which the existing note directly addresses. Recommendation: that one-line legal note must travel with the figures into any DERIV or public asset that reuses them — it currently lives only in HMK-005.
  2. HMK-028's "stolen wages." Quoted, attributed to worker demands, about legislation (Bill 124) subsequently ruled unconstitutional. Recognizable as comment, factual foundation stated, public interest — fair comment carries it comfortably.
  3. Ford-specific conduct claims ("publicly attacked the judge," "froze OW since 2018"). Statements about a Premier's public official acts, sourced to public record; the widest fair-comment latitude in the library applies. The judge-attack claim additionally carries the corroborating detail that counsel formally asked the AG to affirm judicial independence.
  4. DERIV-008's Morley litigation passage. Correctly states the court-record facts (an advocacy organization is suing the City and the councillor) and explicitly refuses the unverified causal story ("currently two independently-sourced facts about the same site"). One counsel note: the aspirational "attributable accountability story" should be handled carefully if ever built — Morley is the defendant in that suit, not its initiator, and an accountability framing that blurs that direction would be a confused imputation.
  5. Litigation-as-allegation discipline generally: the $50M class action is "proposed class action" everywhere it appears; the Ombudsman findings are attributed to the Ombudsman; the Waterloo decision is never cited as settled law. Republication discipline (P6) is intact.

CLEAN — swept and verified, recorded so counsel doesn't re-chase

STRUCTURAL OBSERVATIONS FOR COUNSEL

  1. The anonymity/responsible-communication tension cuts both ways — and the campaign's infrastructure is the answer. Grant v. Torstar's diligence factors are usually hard for an anonymous publisher. But this library has something most named publishers don't: a documented, timestamped diligence record — a corrections log (105+ passes), an automated regression suite, a claim ledger with per-claim verification paths, and a tracked open-issues register that includes items against the campaign's own interest. Recommendation: the lawyer should be shown the diligence infrastructure itself, not just the documents. It is the responsible-communication defence, in artifact form.
  2. Anti-SLAPP posture (CJA s.137.1) is strong. Everything found is public-interest expression grounded in government records — the profile of expression the regime exists to protect. Counsel should still map the notice/limitation mechanics of the Libel and Slander Act for whatever publication channel launches first, since "broadcast" definitions affect deadlines.
  3. The DERIV inheritance problem is the campaign's largest structural legal risk — not any single claim. Both AUDIT-001 propagation failures and this stress-test's compensation-note finding show the same shape: protections written in HMK documents (caveats, legal notes, precision hedges) do not automatically travel into the DERIV assets that actually get published. Before the public-facing flagship is drafted, a propagation rule is needed: any figure, litigation reference, or compensation claim entering a public asset carries its HMK-side caveat with it, verbatim or summarized — enforced by test where patternable.

MAIN-AUDIT-002 | Version 1.0 | July 2, 2026 | Regenerative Toronto Fixed: 0 (nothing required unilateral fixing). Elevated: 1 (poverty-industry legal read → pending decision). Counsel notes: 5. Verified clean: 6 pattern families. New standing recommendation: the DERIV caveat-inheritance rule.

Frozen v1 edition (July 2026) · published 2026-08-17 · corrections to the living library are welcome — tell us where we’re wrong.