In 94.5% of the Trust’s minutes, the record does not say how anyone voted

Read this before the numbers

We want the Islands Trust abolished, so assume we will publish whatever helps us. This page began as the opposite study — a measure of how divided Trust Council has become — and that measure had to be withdrawn. 4 claims are withdrawn below, 4 of them squarely against our own case, including the one this page was built to make.

A tally is printed because a vote was contested. So among motions that record anything at all, the divided share is biased upward by construction. That is why there is no rate here: what follows is a count of documents and a distribution of margins on known-contested votes, never a base rate.

Two framings of the headline, both true and neither rounded: only 2.1% of minutes documents carry a tally at all, and 94.5% carry neither a tally nor a statement that a vote was unanimous. The second is the figure for “the record does not say how anyone voted”, because “carried unanimously” does say how everyone voted. We publish the smaller-sounding number beside the larger one so neither can be quoted alone.

What the record says, and does not

Method. Counted over documents.extracted_text where the title matches '%minute%'. There is no 'minutes' document type in the archive — these are stored as islands_trust_document (7,197) and ltc_agenda (278) — so an earlier query keyed on type returned zero rows for every question. A document counts as carrying a tally if the pattern appears ANYWHERE in it, so these are upper bounds on documents and say nothing about coverage of motions within a document. The true per-motion coverage is therefore worse, not better. Source. Islands Trust document archive, mirrored and full-text extracted (measured 2026-09-12). Corpus. 2,301 documents titled as minutes, 2,289 with extracted text, 2008-2026. Predicates: tally-with-label (in favou?r|opposed)\s*[-:–—]\s*[0-9]; tally count-first [0-9]+[ \t]+(in favou?r|opposed); named sides [0-9]+[ \t]+trustees?[ \t]+(in favou?r|opposed)\s*:

• “recorded vote” — 0 documents. The mechanism that exists in Trust Council's own bylaw. Not once in the archive.
• “recorded division” — 0 documents. The other standard term for the same thing.
• “roll call” — 9 documents. Eight of the nine are Trust Council. A roll call names every trustee on both sides — the strongest record the archive holds.
• “voted against” — 13 documents.
• “abstain” — 24 documents. An abstention is deemed an affirmative vote under CC s.123(4), so recording it changes nothing about the outcome.
• “dissent” — 8 documents.

Method. Case-insensitive substring counts over the same corpus. 'division' alone is excluded as unusable: 569 documents match it, almost all from organisational names such as Parks Canada's 'Protected Areas Establishment Branch' rather than any vote. Source. Islands Trust document archive (measured 2026-09-12). Corpus. 2,289 minutes documents with extracted text

Trust Council records. The committees that decide land use do not.

The distinction matters more than the totals. Trust Council — which sets policy and the budget — records a split in 22 of its 84 minutes documents and names both sides in 2. The local trust committees, which decide what may be built on your land, record a split in 12 of 1,616 documents between them, and name both sides 0 times.

5 bodies record a split never: Salt Spring, Gabriola, Thetis, Saturna, Islands Trust Conservancy. The largest is Salt Spring, the biggest local trust area in the Trust, across 225 minutes documents.

Method. The body is taken from the document title, which is how the archive names them ('Gabriola LTC Regular Meeting Minutes'). metadata->>'area' is populated on only 17 of 7,505 documents and cannot be used. Title classification leaves 38 documents unclassified and produces small duplicate buckets ('Denman Island' 4, 'Sal Spring' 1 — a typo in the source title); those are excluded here rather than merged, and 259 documents from other committees are omitted as not attributable to one body. Source. Islands Trust document archive (measured 2026-09-12). Corpus. 2,289 minutes documents with extracted text, classified by title

Why there is no trend line on this page

'Unanimously' appears in 24 of 42 minutes documents in 2011, 3 of 141 in 2013, and between 0 and 8 a year thereafter against 130-200 documents a year. After 2011 a motion with no tally and no named dissent is indistinguishable between unanimous and bitterly split.

Tallies with a label appear in 6 documents in 2011 and 7 in 2013, then ZERO across 2016-2021, 2 in 2023, and none after.

Both sides are named in 2 documents in the entire archive, both in 2026.

A per-year divided-vote rate over this corpus would measure the minute-taker's habits rather than the Trust's conduct. It is not published, and FALLEN[0] records its withdrawal.

No tally of any kind appears in 7 of the 18 years in the archive: 2008, 2009, 2016, 2017, 2019, 2020, 2021.

Method. Grouped by the year in published_at, which for some documents is the posting date rather than the meeting date. 2010 has no minutes documents in the archive. Counts are of DOCUMENTS matching a pattern, not of motions. Source. Islands Trust document archive (measured 2026-09-12). Corpus. 2,289 minutes documents with extracted text, grouped by published_at year

Is this lawful? Yes — and that is the point

Everything the law requires, the Trust does. The duties arrive through the Islands Trust Regulation, B.C. Reg. 119/90 (O.C. 524/90), which applies them to the trust council, the executive committee, local trust committees and the Trust Fund Board.

• Community Charter Part 4, Division 3 (ss.89-93) — Open meetings. A meeting must be open to the public except as that Division provides.
• Community Charter s.95(2)-(5) — Public access to records. Inspection at the office in regular hours; copies on payment of the fee; subject to FIPPA non-disclosure.
• Community Charter s.97, excluding (1)(d) and (h) — Records available for public inspection. s.97(1)(b) covers 'all minutes of council meetings, other than a meeting or part of a meeting that is closed to the public'. s.97(2): the duty is met if the record is available within 7 days of being requested.
• Community Charter s.123(1)-(5) and (7) — Voting rules. Decided by a majority of members present; each member present must vote; a member who does not indicate how they vote is DEEMED TO HAVE VOTED IN THE AFFIRMATIVE; a tie defeats the motion. Subsection (6) is not applied.
• Local Government Act s.223(1)-(2) — Minutes of meetings. Minutes must be legibly recorded, certified as correct, and signed by the chair or presiding member. It says nothing about how votes are recorded, and nothing about public access.

The finding. Nothing in the Islands Trust Act, the applied Community Charter provisions, the applied Local Government Act provisions, or the regulation requires the minutes to record HOW EACH TRUSTEE VOTED. The minutes must exist, be certified, and be available. What they must contain does not include the direction of anyone's vote.

Community Charter s.97 sits in Part 4 DIVISION 4 ('Public Notice and Access to Records', ss.94-97), not the applied Division 3 ('Open Meetings', ss.89-93). The regulation reaches it by section number, so checking Division membership instead of the regulation's own list returns a false negative.

The mechanism exists. It was never used.

• Trust Council · s.8.13 Recording of Votes
  “A trustee may require his or her vote on any question to be recorded by the Secretary. The Secretary in such case must record in the meeting minutes the name of the trustee and which way the trustee voted.”
  Real, available on demand, and used in ZERO of 2,289 minutes documents. 'Recorded vote' does not appear once in the archive. (TC — Meeting Procedures Bylaw No. 101 (Islands Trust Policy Manual, Ch. 2 s.2 subs.4))
• Trust Council · s.8.14 Recording Abstentions
  “When a member abstains from voting on any question, he or she shall be regarded as having voted in the affirmative and at the request of the member abstaining, or the Chair, the abstention must be recorded in the minutes of the meeting.”
  Mirrors Community Charter s.123(4). An abstention changes nothing about the outcome; recording it only records the reluctance. (TC — Meeting Procedures Bylaw No. 101)
• Local trust committees · s.10 — not operative
  “When a member so requests, the names of those who vote for and those who vote against a question must be recorded in the minutes of the meeting.”
  NOT OPERATIVE. Attachment 3 states in terms: 'This attachment does not form part of the Bylaw but is inserted for convenience only.' And the provision appears in none of the 14 adopted island procedure bylaws. (Policy 4.1.3 Model Local Trust Committee Meetings Procedures Bylaw, Attachment 3 'Rules of Order')

Method. Read from the bylaw text in our own mirrored archive, on whitespace-normalized extracted text. Normalization matters: testing the raw PDF text returned false negatives because phrases break across lines, which produced two wrong conclusions before it was corrected. Source. Islands Trust policy manual and procedure bylaws, mirrored archive (measured 2026-09-12). Corpus. TC Bylaw No. 101; TC LTC Procedure Bylaw No. 197; Policy 4.1.3; and all 14 adopted island procedure bylaws

What the island bylaws do require of their minutes is identical across all fourteen: “The minutes shall record every resolution of the Committee, and the mover, including every resolution closing a meeting to the public, the reading and adoption of every bylaw, and every declaration made in relation to a conflict of interest.” Five things are enumerated. The mover is named by rule. How anyone voted is not on the list.

Method. Clause 13 is identical across the bylaws; verified word for word against Denman No. 244, Hornby No. 170 and Salt Spring No. 529. Five things are enumerated. The MOVER is named by rule. How anyone voted is not on the list. Tested on normalized text, all 14 return no match for 'roll call', 'division', 'polled', 'names of those who vote', 'how ... voted' or 'affirmative' — so no mechanism exists anywhere in them. Source. Islands Trust local trust committee meeting procedure bylaws, mirrored archive (measured 2026-09-12). Corpus. 14 adopted procedure bylaws, whitespace-normalized extracted text

Clause 14, identical across the island bylaws: “Any question of meeting procedure that is not provided for in this Bylaw, the Islands Trust Act, the Local Government Act, the Community Charter, or regulations under either of those statutes, shall be resolved in accordance with the most current edition of Robert's Rules of Order.” ⚠️ Whether Robert's Rules would grant a division or roll call on demand is NOT established here and must not be asserted without reading the edition in force. The verified claim is narrower: the bylaws themselves provide nothing.

The two documents that do name both sides

These are exhibits, not a series. Three votes are three votes, and this page says so.

• TC-2026-017 · Reduce the Contracted Temporary Staffing services line item in the draft 2026/27 budget by $5,000
  12 in favour · 12 opposed · 1 abstained · 1 absent (not a vote)
  A tie. Under Community Charter s.123(5) as applied, an equal vote defeats the motion.
  ⚠️ ATTRIBUTION TRAP. The roll call is printed under TC-2026-018, the procedural motion 'that Islands Trust Council request a roll call'. But the minutes say it reports 'how they voted on the previous motion' — the budget motion. Attaching the tally to the block it sits in would record a tied budget vote as a vote about asking for a roll call. Attribution follows the minutes' own cue phrase, never position.
• Unlinked · Reduce the total 2026/27 Islands Trust Conservancy budget by $130,000
  6 in favour · 15 opposed · 5 abstained · 1 absent (not a vote)
  Defeated.
  Left unlinked deliberately. The cue is 'The question on the following motion was then called by roll call', and the motion it resolves (TC-2026-019) carries no outcome line of its own. An unlinked vote is worth more than a wrongly linked one.
• Unlinked · Islands Trust Council Bylaw 204, the Financial Plan Bylaw 2026-2027, be read a first time
  16 in favour · 8 opposed · 2 absent (not a vote)
  Carried.
  Here the tally is printed BEFORE the motion text and its outcome — a third attribution direction in the same document.

Method. Extracted by parseRollCalls() in server/minutesVotes.ts, which reads the whole document rather than a motion block: these roll calls follow the outcome word, where a block-scoped reader finds zero of them (measured: 0 named sides across all 289 fixture motions). Names are read only from a segment anchored to a tally label, because the same shape also matches presenter lists and agenda numbering. Absences are recorded separately and never enter a side. Guarded by 8 assertions in tests/minutesRollCalls.test.ts. Source. Islands Trust Council Quarterly Meeting Minutes, 10-12 March 2026 (adopted) (measured 2026-09-12). Corpus. The 2 documents in 2,289 that name both sides of a vote

Opacity in the minutes is not an absence of rights

The Trust is a public body. FIPPA's 'local public body' includes 'local government body', which expressly includes 'the trust council, the executive committee, a local trust committee and the Islands Trust Conservancy, as these are defined in the Islands Trust Act'.

OIPC Order F21-40 (Elizabeth Barker, Director of Adjudication): the Trust withheld Salt Spring bylaw-enforcement records in their entirety under FIPPA s.15(1), and the adjudicator found its law-enforcement harm submissions 'consisted of assertions that were unsupported by evidence'. Read the order.

Opacity in the minutes is NOT an absence of access rights. A reader who concludes that nothing can be obtained has been misled. Records are requestable; they are simply not published.

What we got wrong

• Withdrawn 2026-09-12 · against our own case
  A per-year divided-vote rate showing Trust Council growing more divided over time — the measure that would have carried 'A Trust Divided'.
  Not computable, and the attempt is what produced this page. Across 2,289 minutes documents only 47 carry any tally, both sides are named in 2, and 'recorded vote' appears 0 times. Worse, the convention is era-shaped: 'unanimously' appears in 24 of 42 documents in 2011 and 3 of 141 by 2013, and tallies are absent entirely from 2016 to 2021. After 2011 a motion with no tally is indistinguishable between unanimous and bitterly split, so any rate would measure the minute-taker rather than the Trust — the identical defect that invalidated the named-dissent series. This also falsified our own protocol's open item, which asserted tallies 'exist in 2011, 2019-2024 and 2025-26 ... across every era'. They do not.
• Withdrawn 2026-09-12 · against our own case
  Trust Council votes are overwhelmingly unanimous, or overwhelmingly divided — either framing, as a general characterisation.
  Neither is measurable, and a tally is printed BECAUSE a vote was contested, so the divided share among motions that have any basis is biased upward by construction. What survives is a distribution of margins on known-contested votes, plus three named roll calls as exhibits. Three votes are three votes, and the page says so.
• Withdrawn 2026-09-12 · against our own case
  The municipal transparency floor does not reach the Islands Trust — no open-meeting duty, no minutes duty, no public-inspection duty.
  Wrong, and it was our own reading of the Islands Trust Act that produced it. ss.13(1) and 26 apply only Community Charter ss.122, 135, 136, 139, 140, 163 and LGA s.225(1), so the Act alone does suggest that conclusion. But B.C. Reg. 119/90 applies Division 3 of Part 4 (open meetings), s.95(2)-(5), s.97 excluding (1)(d) and (h), s.123(1)-(5) and (7), and LGA s.223(1)-(2) to every trust body. Open meetings, minutes and public inspection all bind. The claim that would have been stronger for the campaign is the false one, and it is withdrawn.
• Withdrawn 2026-09-12 · against our own case
  The Trust breaks its own bylaw by not recording votes against, because the model LTC bylaw requires it.
  Withdrawn twice over, and the sequence is instructive. Policy 4.1.3 s.10 does say the names of those voting for and against 'must be recorded' — but conditioned on 'when a member so requests', so it is opt-in, not mandatory. It then turned out to sit in Attachment 3, which states it 'does not form part of the Bylaw but is inserted for convenience only', and to appear in none of the 14 adopted island bylaws. So there is no breach: there is no rule to breach. The strongest available version of this accusation is that no adopted LTC bylaw provides the mechanism at all.

What we cannot compute yet

• Per-MOTION coverage — what share of individual motions in the archive record a split. Every figure here counts DOCUMENTS matching a pattern, which is an upper bound. Producing a per-motion figure means parsing all 2,289 documents with parseMotions(), which is built but has not been run at archive scale. The direction is known: per-motion coverage is worse than per-document coverage, never better.
• Whether Robert's Rules of Order would give an LTC trustee a recorded vote on demand. Clause 14 of every island bylaw falls back to 'the most current edition' of Robert's Rules, which we have not read and do not hold. Asserting what it grants would be exactly the kind of unsourced claim this project retires.
• How any individual enforcement or in-camera matter was decided. There is no decisions register. Enforcement files are handled in camera and never referenced by number in the public record — see the enforcement entry in shared/trustPowers.ts, where the counts exist and the outcomes do not.
• Island-level vote cohesion of any kind. The local trust committees record no split at all. Salt Spring's 225 minutes documents contain none; neither do Gabriola's 186, Thetis's 95, Saturna's 67, or the Conservancy's 55. There is nothing to compute from.
• Whether trustees vote in stable blocs. Would need many named votes. The archive holds 2 documents naming both sides. Three roll calls support exhibits, not a bloc analysis, and a coalition score from n=3 would be invention.

Method. Each blocker names the specific rule, absence or unread source that causes it, so a reader can tell a limit of the record from a limit of our effort. Source. docs/vote-transparency-in-the-minutes.md (measured 2026-09-12). Corpus. Various — each blocker names what it fails at.

What this page will not do

• We will not say the Islands Trust is breaking the law by not recording how trustees vote. It is not. Every transparency duty that is written down is met. 'Unlawful' and 'lawful but chosen' are different accusations and only the second is true.
• We will not publish a divided-vote rate, per year or per island. The record cannot support one, and a rate built on this corpus would measure the minute-taker.
• We will not treat an absence as a vote. A trustee recorded absent is never published as having opposed anything; absences are held in their own field and guarded by a test.
• We will not name a trustee as voting a particular way unless the record names them. Inference about how someone probably voted stays out of public pages permanently.
• We will not merge this page with our own /transparency page. That page is about this project's ownership, funding, provenance labels and corrections. Conflating the two would let a reader think our disclosures and the Trust's conduct are the same subject.
• We will not describe the absence of a recorded vote as proof a vote was contested, or as proof it was unanimous. It is proof of neither, and that is the finding.