Owen Sound: A Four-Year City Business Plan

Home › Appendices › Appendix H

Appendices

Appendix HSources, Evidence and Verification Standards

10,855 words · Mike Seiler · Owen Sound, Ontario

Open in the reader →

Vote on the proposals, hear the audio, read the reviews, search the whole plan.

In this chapter

The stronger the public claim, the stronger the evidence should be

Government makes decisions using information.

Some information is:

Those things are not interchangeable.

A resident saying:

This intersection feels dangerous

is valuable evidence of lived experience.

It is not, by itself, proof of:

A traffic study may measure:

It does not, by itself, tell Council:

A consultant can provide:

A consultant does not:

A survey can describe:

It does not automatically describe:

A government website can provide:

It can still become:

A newspaper can report:

Where the underlying:

is available, the original source is normally stronger for the underlying factual claim.

Artificial intelligence can help:

It does not turn an unsupported statement into:

The operating standard for this plan is:

The stronger the public claim, the stronger the evidence should be.

A second principle follows:

Use the source closest to the fact whenever practical.

A third principle is equally important:

Evidence should inform judgment without pretending judgment has disappeared.

H.1Purpose

This appendix establishes the evidence and verification standards used throughout:

Owen Sound: A Four-Year City Business Plan.

H.2It Applies To

H.3Evidence Is Claim-Specific

There is no single source hierarchy that answers:

H.4Example

A statute may be the strongest source for:

what the law says.

It may be a poor source for:

how residents experience the service.

H.5Another Example

Resident testimony may be excellent evidence for:

what that resident experienced.

It is weaker evidence for:

how every resident experiences the service.

H.6Evidence Hierarchy Is Not Human Hierarchy

Calling one source stronger for a particular factual claim does not mean:

behind another source is less important.

H.7Source Categories

The City's evidence system should distinguish:

Authoritative Legal Sources

Official Administrative Records

Audited and Verified Financial Records

Professional and Technical Evidence

Official External-Government Sources

Primary Source Documents

Academic and Research Evidence

Representative Surveys

Partner-Reported Information

Journalistic Sources

Resident and Stakeholder Evidence

Anecdotal Evidence

Unverified Information

H.8Tier One: Constitution and Statute

For constitutional or statutory claims, begin with:

H.9Current Consolidation

Use the:

when the question concerns current law.

H.10Historical Law

Use the version in force:

H.11Do Not Use Current Law to Describe Past Law Automatically

No.

H.12Do Not Use Old Law to Describe Current Law

No.

H.13Effective Date

Where legal change matters:

Record:

H.14Proposed Legislation

A bill is not automatically:

H.15Passed but Not Yet In Force

Different from:

H.16Regulation

Check separately where statute depends on:

H.17Municipal By-Law

Use the:

H.18Consolidation Warning

Administrative consolidations can be useful.

Where legal precision is critical:

Verify against:

H.19Court Decisions

For legal interpretation:

Use authoritative:

H.20Highest Court Is Not Automatically Only Relevant Court

Binding authority depends on:

H.21Case Summary

Useful for orientation.

H.22Judgment

Stronger for:

H.23Headnote

Helpful.

But not substitute for reading relevant reasons where material.

Useful for:

No.

A qualified legal opinion may be necessary for:

Different.

H.28Privileged Advice

May need to remain:

Can still identify:

where lawful without publishing privileged advice.

H.30Tier Two: Official Municipal Records

For claims about what Owen Sound:

use official municipal records wherever possible.

H.31Examples

H.32Council Agenda Is Not Decision

An item appearing on an agenda does not prove:

H.33Staff Recommendation Is Not Council Decision

Critical.

H.34Draft Is Not Final

Always distinguish.

H.35Resolution Is Evidence of Decision

Subject to:

H.36Minutes

Use approved minutes where decision history matters.

H.37Video

Can supplement.

H.38Video Is Not Preferred Substitute for Formal Decision Record

No.

H.39Official Correspondence

Can establish:

H.40Letter Sent Is Not Result Achieved

No.

H.41Tier Three: Financial Evidence

Financial claims should normally rely on:

H.42Audited Actual

Stronger for completed historical financial claims.

H.43Current-Year Actual

May rely on:

Label where appropriate.

H.44Budget

Evidence of:

Not actual result.

H.45Forecast

Evidence of:

Not actual result.

H.46Campaign Number

Not official financial evidence merely because:

H.47Savings Claim

Requires:

H.48"We Saved"

High evidentiary threshold.

H.49Tier Four: Professional and Technical Evidence

Technical claims may require:

H.50Qualification Matters

A professional opinion is strongest where the person is:

H.51Engineering Report

Can support:

H.52Engineering Report Does Not Decide Political Priority

Council still decides:

H.53Survey Plan

Can establish boundaries more reliably than:

H.54Environmental Assessment

Can establish technical findings within:

H.55Scope Matters

Every technical report should be read for:

H.56Limitation Section

Read it.

H.57Assumptions

Read them.

H.58Date

A technical report can become:

H.59Site Conditions Change

Yes.

H.60Standards Change

Yes.

H.61Professional Opinion Can Differ

When qualified experts disagree:

Do not pretend:

H.62Identify Source of Disagreement

Could involve:

H.63Commissioned Consultant

A consultant paid by City can still provide:

H.64Commissioned Does Not Mean Independent Automatically

Disclose:

H.65Consultant Conflict

Review where material.

H.66Consultant Recommendation

Not Council mandate.

H.67Study Is Evidence

Not:

H.68Tier Five: Official External-Government Information

For claims about:

prefer those institutions' official sources.

H.69Examples

H.70Program Website

Can establish current public description.

H.71Eligibility

Do not assume City qualifies merely because:

H.72Program Announced

Does not mean:

H.73Application Open

Does not mean:

H.74Funding Award

Does not necessarily mean:

H.75Transfer Agreement

Stronger evidence for:

H.76External-Government Statement

Use the institution responsible for the:

H.77Minister Speech

Can indicate:

It may not be the operative:

H.78Press Release

Useful.

But check underlying:

where material.

H.79Tier Six: Primary Source Documents

Primary evidence can include:

H.80Primary Does Not Automatically Mean Accurate

A primary source can contain:

H.81But It Can Establish

what the source itself said or recorded.

H.82Example

A letter proves:

It may not prove:

H.83Contract

Strong source for:

H.84News Story About Contract

Secondary.

H.85Title Record

Strong source for:

subject to professional interpretation where needed.

H.86GIS

Useful navigation.

Not substitute for:

H.87Photograph

Evidence of:

if date and location are reliable.

H.88Photograph Does Not Show Everything

No.

H.89Video

Same.

H.90Metadata

Useful where provenance matters.

H.91Editing

Material edits should be:

if visual evidence is used.

H.92Synthetic Image

Never present as:

H.93AI-Generated Visualization

Label:

Illustration / Conceptual Visualization

where used.

H.94Rendering

Not proof of final:

H.95Tier Seven: Academic and Research Evidence

Academic research can help answer questions about:

H.96Peer Review

Can strengthen reliability.

H.97Peer-Reviewed Does Not Mean Universally True

Research may still have:

H.98One Study

Rarely settles broad policy.

H.99Literature Review

May be stronger for:

H.100Systematic Review

Can be useful.

H.101Meta-Analysis

Can be useful where studies are sufficiently:

H.102Canadian Context

Prefer where issue depends strongly on:

H.103Local Context

Still matters.

H.104Research Transfer

A result from:

may not transfer directly to Owen Sound.

H.105Explain Applicability

Where research underpins major proposal.

H.106Publication Date

Important in fast-changing fields.

H.107Older Research

Can remain valuable.

H.108Foundational Research

May be old and still:

H.109Fast-Changing Technology

Requires recent evidence more often.

H.110Tier Eight: Professional Standards and Guidance

Standards may come from:

Distinguish.

H.112Standard Incorporated by Law

Different from:

H.113Best Practice

Not automatically:

H.114Guidance

Not automatically:

H.115Language Matters

Use:

recommended

when recommended.

Not:

required

unless required.

H.116Tier Nine: Survey Evidence

Surveys can provide important evidence about:

H.117Survey Quality Depends On

H.118Representative Survey

Must have methodology supporting:

H.119Self-Selected Survey

Useful.

But describes:

H.120Open Poll

Not referendum.

H.121Strong Vote

If verified participation is used:

Still distinguish:

H.122Response Rate

Publish where meaningful.

H.123Sample Size

Publish.

H.124Question Wording

Publish.

H.125Order Effects

Can matter.

H.126Leading Question

Avoid.

H.127Loaded Question

Avoid.

H.128Forced Choice

Use only where appropriate.

H.129Neutral Option

Consider where meaningful.

H.130"Don't Know"

Can be valid response.

H.131Survey Date

Critical.

H.132Public Opinion Changes

Yes.

H.133Survey About Proposed Policy

Should describe policy accurately enough for:

H.134No Campaign Framing in Municipal Survey

H.135Tier Ten: Partner-Reported Evidence

Community organizations, businesses and other partners may report:

H.136Partner Data Can Be Valuable

Yes.

H.137Partner Data Should Be Labelled

Where City has not independently verified:

Partner reported.

H.138Grant Reporting

Should be proportionate.

H.139Large Public Contribution

May warrant stronger verification.

H.140Small Volunteer Grant

Does not require:

H.141Shared Service

May require joint:

H.142County Data

Do not silently rebrand as:

H.143SON Information

Handle according to:

H.144Evidence Hierarchy and Indigenous Knowledge

Indigenous knowledge should not be forced into a simplistic Western ranking system that treats:

as inherently inferior.

H.145Claim-Specific Standard

For legal rights:

Use authoritative legal sources.

For SON's own:

SON's own sources and agreed records may be essential.

H.146Knowledge Holder

Where appropriate:

H.147Confidential Knowledge

Do not publish merely because it informed:

H.148Compensation

Where Elders or knowledge holders provide substantive professional or cultural expertise:

Compensation may be appropriate.

H.149No Extractive Evidence Gathering

Do not request community knowledge merely to:

H.150Tier Eleven: Journalism

Professional journalism can be valuable for:

H.151News Report Is Secondary Evidence

Where primary evidence exists:

Check it.

H.152Breaking News

May contain incomplete information.

H.153Correction

Check for later updates where claim matters.

H.154Opinion Column

Opinion.

H.155News Report

Different.

H.156Editorial

Different.

H.157Sponsored Content

Different.

H.158Local Journalism

Can contain important community information unavailable elsewhere.

H.159Use Carefully

Especially for claims about:

H.160Allegation

Do not convert into:

H.161Charged

Not:

H.162Lawsuit Filed

Not:

H.163Claim Made

Not:

H.164Tier Twelve: Resident and Stakeholder Evidence

Resident testimony is legitimate evidence.

H.165It Can Establish

H.166It May Reveal What Data Miss

Important.

H.167One Story

Can identify:

H.168One Story Does Not Establish Prevalence

No.

H.169Ten Similar Stories

May indicate:

H.170Pattern Still Needs Careful Interpretation

Yes.

H.171Complaint Data

Can support:

H.172No Complaint

Does not prove:

H.173Some Residents Do Not Complain

Yes.

H.174Barriers to Complaining

Can include:

H.175Public Meeting

May overrepresent residents able or motivated to:

H.176Door-to-Door Conversation

Useful.

Not automatically:

H.177Business Roundtable

Useful.

Not automatically:

H.178Youth Session

Same.

H.179Senior Session

Same.

H.180Consultation Evidence

Report:

what participants said

rather than:

what Owen Sound thinks

unless methodology justifies broader claim.

H.181Lived Experience and Technical Evidence

Do not force false competition.

H.182Example

An accessible ramp can comply technically and still be:

Both forms of evidence matter.

H.183Tier Thirteen: Anecdote

Anecdote is:

an individual account or observation not designed to establish a broader statistical conclusion.

H.184Anecdote Is Not Useless

It can generate:

H.185Anecdote Is Not Population Evidence

No.

H.186Political Anecdote

Particularly easy to:

H.187"Everyone Says"

Red flag.

H.188Ask

Who? How many? According to what evidence?

H.189"Nobody Wants"

Same.

H.190"Businesses Are Leaving"

Ask:

H.191"Crime Is Exploding"

Ask:

H.192"Downtown Is Dying"

Ask:

H.193Anecdote to Investigation

Good process:

Hear

Record

Check

Measure

Respond

H.194Tier Fourteen: Social Media

Social media can establish:

what an account posted.

H.195It Does Not Automatically Establish

that the post is true.

H.196Anonymous Post

Lower verification.

H.197Screenshot

May be:

H.198Original Post

Prefer where available.

H.199Deleted Post

Archive evidence may be useful where lawful.

H.200Viral Claim

Does not gain evidentiary strength because:

H.201Repetition Is Not Corroboration

Critical.

H.202Source Laundering

A claim repeated by ten websites may still originate from:

H.203Independent Sources

Ask whether sources genuinely derive from:

H.204Ten Articles Repeating One Press Release

Essentially one underlying:

H.205Corroboration

Means independent evidence supports:

H.206Unverified Information

Information that has not been sufficiently checked should be labelled:

Unverified.

H.207Unverified Does Not Mean False

It means:

H.208Disputed

Use when credible sources materially disagree.

H.209Preliminary

Use when information is credible but:

H.210Verified

Use when evidence has been checked to the standard appropriate for:

H.211Superseded

Use when a newer authoritative source replaces:

H.212Evidence Status Vocabulary

Use:

Verified

Supported

Preliminary

Disputed

Unverified

Superseded

Unknown

H.213Verified

Highest practical confidence for:

not metaphysical certainty.

H.214Supported

Evidence is reasonably strong but may include:

H.215Preliminary

Awaiting:

H.216Disputed

Material credible disagreement.

H.217Unverified

Insufficient checking.

H.218Superseded

No longer current.

H.219Unknown

Evidence currently insufficient.

H.220Claim Confidence

Where useful use:

High

Moderate

Low

Unknown

H.221High Confidence

Multiple strong or authoritative sources, or one source directly authoritative for the claim.

H.222Moderate Confidence

Good support with meaningful limitations.

H.223Low Confidence

Limited, indirect or conflicting support.

H.224Unknown

Insufficient basis.

H.225Confidence Should Match Claim

A legal statute citation may give:

about statutory wording.

It may not give high confidence about:

H.226Claim Types

Every major factual claim can be classified as:

Legal

Financial

Technical

Operational

Statistical

Historical

Causal

Forecast

Perception

Policy Preference

Lived Experience

Best evidence:

H.228Financial Claim

Best evidence:

H.229Technical Claim

Best evidence:

H.230Operational Claim

Best evidence:

H.231Statistical Claim

Best evidence:

H.232Historical Claim

Best evidence:

H.233Causal Claim

Requires stronger evidence than:

H.234Forecast Claim

Requires:

H.235Perception Claim

Best measured through:

H.236Policy Preference

Does not need to pretend to be:

H.237Say

This plan prefers...

where it is a value judgment.

H.238Do Not Invent Study to Justify Every Value

No.

H.239Lived Experience

Evidence of:

Treat respectfully.

H.240Claim Strength

The more consequential the claim:

H.241Low-Consequence Claim

Example:

Residents have asked for more benches.

Could be supported by:

H.242Higher-Consequence Claim

This bridge is unsafe.

Needs:

H.243Very High-Consequence Claim

The City lacks legal authority.

Needs:

H.244High Financial Claim

This program will save $2 million annually.

Needs:

H.245Causal Economic Claim

This event generated $10 million in new local economic activity.

Requires strong:

H.246Property-Value Claim

This streetscape policy will increase property values.

Should be presented as:

unless evidence supports stronger causal language.

H.247Safety Claim

Cameras made downtown safer.

Requires more than:

H.248Housing Claim

The City created 500 homes.

Requires evidence of:

and careful attribution.

H.249"Could"

Use for:

H.250"May"

Use for:

H.251"Expected"

Forecast.

H.252"Projected"

Model or forecast.

H.253"Estimated"

Approximation.

H.254"Associated With"

Relationship without asserting causation.

H.255"Contributed To"

Partial causal role with support.

H.256"Caused"

High threshold.

H.257"Proved"

Very high threshold.

Use sparingly.

H.258Current-Law Verification

Legal references should include a:

Verified as of

date when current law materially matters.

H.259Current Program Verification

Same for:

H.260Current Ownership Verification

Same for:

H.261Time-Sensitive Facts

Always ask:

Could this have changed?

H.262If Yes

Verify current source.

H.263Old Official Source

Can be accurate historically.

Not necessarily:

H.264Publication Date

Every important source should record:

H.265Access Date

For online information that may change:

record:

H.266Version

Where document has:

record.

H.267Revision Date

Record.

H.268Source Register

Maintain a central:

Evidence and Source Register.

H.269Source Register Fields

Source ID

Title

Issuer / Author

Type

Publication Date

Effective Date where applicable

Version

Access Date

Source Location

Relevant Claim

Confidence

Current / Superseded

Verification Owner

Notes

H.270Claim Register

For the most consequential claims maintain:

Claim Register.

H.271Claim Register Fields

Claim ID

Exact Claim

Claim Type

Initiative / Section

Source IDs

Evidence Status

Confidence

Last Verified

Next Verification Trigger

Responsible Reviewer

H.272Not Every Sentence Needs Claim Register

Proportionality.

H.273Use For

H.274Verification Trigger

Could be:

H.275Claim Drift

A careful statement can become exaggerated over time.

H.276Example

Original:

Study suggests a potential reduction.

Later:

This will reduce costs.

Later:

This saves $1 million.

Prevent.

H.277Preserve Original Strength

Do not upgrade claim without:

H.278Citation

Important factual claims in the final business plan should be tied to:

H.279Citation Style

The final compiled plan should use:

rather than unstable raw links scattered throughout the body.

H.280Source Record

Each citation should allow resident to identify:

Can supplement.

Do not make a raw URL the only evidence record.

H.283Archive

Where lawful and appropriate preserve:

H.284Do Not Archive Confidential Material Publicly

No.

H.285Official Document Title

Record.

H.286Issuing Institution

Record.

H.287Page / Section

Record where practical.

H.288Precise Citation

Better than:

according to government.

H.289Quote

Use exact wording where:

H.290Paraphrase

Usually better for:

H.291Quote Verification

Compare against:

H.292Ellipsis

Do not change:

H.293Partial Quote

Do not remove qualifying words that materially alter:

H.294Context

Read surrounding text.

H.295Statutory Quote

Section context matters.

H.296Court Quote

Facts and legal issue matter.

H.297Consultant Quote

Limitations matter.

H.298Media Quote

Speaker attribution matters.

H.299Translation

If source translated:

Identify where material.

H.300Numbers

Every important public number should have:

H.301"$5 Million"

Ask:

H.302"25%"

Ask:

H.303"500 Homes"

Ask:

H.304"100 Businesses"

Ask:

H.305Evidence Before Headline

Always.

H.306Conflicting Evidence

When credible sources disagree:

Do not quietly select:

H.307Conflict Process

Identify disagreement.

Check source quality.

Check dates.

Check definitions.

Check independence.

Check methodology.

Seek clarification.

Report unresolved conflict where material.

H.308Newer Is Not Automatically Better

But can be more:

H.309Official Is Not Automatically Correct

But can be authoritative for:

H.310Academic Is Not Automatically Better Than Local Data

Depends on:

H.311Resident Experience Is Not Overruled by Average

An average service time can be good while:

Both can be true.

No.

H.314Source Independence

Two sources are not independent if one copied:

H.315Circular Citation

Avoid.

H.316Example

Blog cites newspaper.

Newspaper cites blog.

That is not:

H.317Original Dataset

Find where possible.

H.318Press Release Number

Look for underlying:

H.319Chart

Read:

H.320Infographic

Not enough if methodology absent.

H.321Data Dashboard

Record:

H.322Dynamic Dashboard

Can change.

Preserve relevant:

where material.

H.323Map

Record:

H.324Boundary

Critical.

H.325Downtown Data

Use stable:

H.326Neighbourhood Data

Define neighbourhood.

H.327No Boundary Shopping

Do not change geography to:

H.328Statistical Revision

Official datasets may revise historical values.

H.329Use Latest Revision

Where current reporting requires.

H.330Preserve Original Publication

Where accountability requires showing:

H.331"What We Knew Then"

Different question from:

what is best data now.

H.332Decision Review

Can consider both.

H.333Forecasts

Forecasts should contain:

Source

Assumptions

Base Date

Time Horizon

Scenario

Uncertainty

H.334Scenario

A possible future under:

H.335Scenario Is Not Prediction

No.

H.336Base Case

Not guaranteed.

H.337Best Case

Not plan by default.

H.338Worst Case

Not inevitable.

H.339Sensitivity

Useful.

H.340Forecast Revision

Publish when material.

H.341Forecast Accuracy Review

For major recurring forecasts:

Compare with:

H.342Learn

Yes.

H.343Do Not Punish Honest Uncertainty

Forecasts will be wrong sometimes.

H.344Do Punish Hidden Assumptions

Institutionally, by improving process.

H.345Economic Forecast

Particularly uncertain.

H.346Construction Forecast

Particularly sensitive to:

H.347Population Forecast

Model.

H.348Housing Forecast

Model.

H.349Tax Forecast

Model.

H.350No Forecast as Promise

Unless commitment is actually to:

not control external result.

H.351Causation

The City should be conservative when claiming:

H.352Before-and-After

Does not automatically prove:

H.353Other Factors

Could include:

H.354Control Group

Sometimes available.

Often not.

H.355Difference-in-Differences

Can be useful in advanced evaluation.

Not required for:

H.356Practical Municipal Standard

Use:

honestly.

H.357Pilot Evidence

A pilot should define before launch:

Problem

Hypothesis

Baseline

Measures

Duration

Success Criteria

Stop Criteria

H.358Do Not Declare Pilot Success Because People Liked Launch

No.

H.359Novelty Effect

Can distort early:

H.360Seasonality

Can distort.

H.361Pilot Sample

May be too small for:

H.362Pilot Success

Can justify:

rather than permanent expansion automatically.

H.363Pilot Failure

Useful evidence.

H.364Publish

Where appropriate.

H.365No Failed-Pilot Burial

H.366Policy Evaluation

Ask:

Was it implemented?

Did intended users use it?

Did service change?

Did outcome change?

What did it cost?

What unintended effects appeared?

H.367Process Evaluation

Different from:

H.368Both Matter

H.369Resident Testimony in Evaluation

Include where relevant.

H.370Front-Line Staff

Include.

H.371Partner Evidence

Include.

H.372Financial Evidence

Include.

H.373Technical Evidence

Include.

H.374Triangulation

Use multiple types of evidence when:

H.375Triangulation Does Not Mean Vote Counting Sources

Three weak sources do not automatically beat:

H.376Evidence Weight

Depends on:

H.377Source Quality Test

Ask:

Is it primary?

Is it authoritative?

Is it current?

Is methodology known?

Is it independent?

Is it complete?

Is there conflict?

Is it fit for this claim?

H.378Bias

Every source can have:

H.379Government Source

May have institutional:

H.380Business Source

May have commercial:

H.381Advocacy Source

May have policy:

H.382Campaign Source

Has political:

H.383That Does Not Automatically Make It False

No.

H.384Disclose Interest Where Material

H.385Funding Source

For studies:

Can matter.

H.386Sponsor

Can matter.

H.387Methodology Matters More Than Label Alone

Yes.

H.388Conflict-of-Interest Evidence Review

When City decision could benefit:

use stronger:

H.389RealMap

If municipally considered:

Evidence must be independent of:

H.390map.ca

Same.

H.391Founder Claims

May identify:

They are not sufficient proof of:

H.392Independent Testing

Required where material.

H.393Private Business Plan

Should not silently become:

H.394Campaign Research

Same.

H.395Municipal Adoption

Requires:

H.396Campaign-Public Firewall

A campaign can:

The City must:

H.397Municipal Report Should Not Cite Campaign Platform as Proof

Except to establish:

H.398Campaign Platform Is Primary Source for Campaign Promise

Exactly.

H.399Campaign Claim About Current Law

Needs:

H.400Campaign Claim About City Finance

Needs:

H.401Campaign Claim About Public Opinion

Needs:

H.402Campaign Claim About Personal Intention

Candidate is authoritative source for:

H.403Election Integrity

Do not selectively raise evidence standard against:

while lowering it for:

H.404Same Standard

Always.

H.405AI-Assisted Research

AI can be used to:

H.406AI Is Not Primary Source

Never cite:

AI said

as proof of:

H.407AI Can Hallucinate

Therefore:

H.408AI Can Misquote

Verify quotation.

H.409AI Can Confuse Dates

Verify.

H.410AI Can Merge Different Jurisdictions

Verify.

H.411AI Can Cite Nonexistent Sources

Never accept citation without:

H.412AI Summary

Useful.

But if decision depends on precise text:

Read:

High-risk.

Human legal review where:

H.414Financial AI Use

Reconcile to:

H.415Technical AI Use

Qualified professional remains responsible for:

H.416Survey AI Use

Can assist coding.

But methodology must remain:

H.417Confidentiality

Do not put confidential City information into external AI systems unless:

H.418Personal Information

Same.

H.419Indigenous Knowledge

Same, with additional:

requirements.

H.420Procurement

AI vendor should not receive City data by default because:

H.421Approved Systems

Use.

H.422AI-Generated Text

Should be reviewed by person responsible for:

H.423Accountability Does Not Transfer to AI

No.

H.424AI Research Log

For major research tasks, useful fields may include:

Question

Sources located

Human reviewer

Claims verified

Date

H.425Not Every AI Prompt Needs Archive

Proportionality.

H.426High-Stakes Research

Stronger record.

H.427Source Security

Downloaded documents should be stored where appropriate.

H.428File Name

Use meaningful identifier.

H.429Version Control

Important.

H.430Official Copy

Preserve where lawful.

H.431Hash

May be useful for high-integrity digital evidence.

Not necessary for:

H.432Archive Date

Record.

H.433Website Screenshot

Can establish what page displayed at:

H.434Screenshot Is Not Better Than Downloadable Official Document

Use original where possible.

Source record should still allow:

H.436Permanent Identifiers

Prefer where available.

H.437Legislative Section Number

Useful.

H.438Court Citation

Useful.

H.439Report Number

Useful.

H.440Council Report Number

Useful.

H.441By-Law Number

Useful.

H.442Contract ID

May be useful internally.

H.443Source Accessibility

Public evidence should be accessible where practical.

H.444Paywalled Source

Can be used.

But seek publicly accessible alternative where:

H.445Residents Should Not Need Expensive Subscription to Verify Basic Public Claim

Where avoidable.

H.446Accessibility of Source Documents

Use accessible documents where possible.

H.447Scanned PDF

Can be difficult.

H.448Provide Accessible Summary

Without altering:

H.449Language

Plain-language explanation should not replace technical source.

H.450Both

Provide where useful.

H.451Historical Evidence

Historical claims require special care.

H.452Present-Day Name

May differ from:

H.453Historical Terminology

Can be offensive or outdated.

Quote only where:

H.454Context

Provide.

H.455Local History

Use:

H.456Oral History

Can be important primary historical evidence.

H.457Oral History Is Perspective

Like other primary sources, it should be understood within:

H.458Indigenous Oral History

Do not automatically subordinate to:

Still requires appropriate:

H.460Historical Relationship Question

May require:

H.461Treaty History

Use:

H.462No Simplified "Settled Fact" Where History Is Contested

Explain.

H.463Current Territorial Claim

Requires current verification.

H.464Do Not Recycle Old Political Language

No.

H.465Harbour History

Historical ownership should not be confused with:

H.466Road History

Same.

H.467Financial History

Use comparable accounting where possible.

H.468Methodology Change

Explain.

H.469Long Historical Trend

Check whether definition remained:

H.470Evidence in Council Reports

Major Council reports should distinguish:

What We Know

What We Estimate

What We Do Not Know

What We Recommend

H.471This Is Powerful

Because it separates:

H.472Recommendation

Can be strong even with uncertainty.

H.473But Uncertainty Should Be Visible

H.474Decision Under Uncertainty

Sometimes necessary.

H.475Precaution

Can justify action where potential harm is serious.

H.476Precaution Is Not Permission to Invent Facts

No.

H.477Reversibility

When evidence is weak:

Prefer reversible:

where suitable.

H.478Irreversible Decision

Requires stronger evidence.

H.479Examples

H.480Reversible Decision

May tolerate:

H.481Evidence Threshold by Decision Type

DecisionTypical Evidence Standard
Routine operational adjustmentAdministrative evidence
Low-cost reversible pilotBaseline plus defined measures
Major procurementBusiness case, cost, procurement evidence
Significant capitalTechnical, financial and lifecycle evidence
Land sale / acquisitionTitle, valuation, strategic review, legal review
Rights-sensitive policyLegal and rights analysis
Major surveillance / data systemLegal, privacy, security, necessity evidence
Major intergovernmental claimCurrent authoritative government sources
Major historical assertionAppropriate primary and scholarly sources
Significant causal claimStrong analytical evidence

H.482Proportionality

Do not commission:

to decide:

without good reason.

H.483But Do Not Understudy

A cheap study is not worth it if the decision could create:

H.484Evidence Cost Is Part of Decision Cost

Yes.

H.485Due Diligence

Can save:

H.486Study Trap

More study is not always:

H.487Study Completion Gate

At each new study request ask:

What question remains?

Why does it matter?

What decision changes depending on answer?

Has prior work already answered it?

H.488No Duplicate Study

Unless prior work is:

H.489Study Inventory

Maintain major previous:

H.490Before New Consultant

Search:

H.491Standing Work List

Use existing evidence.

H.492Decision Deferred by Study

Track.

H.493Analysis Paralysis

Real risk.

H.494Evidence Is to Support Decision

Not avoid:

H.495Public Consultation as Evidence

Consultation can answer:

H.496It Cannot Answer Alone

H.497Technical Evidence Cannot Answer Alone

H.498Council Integrates

That is governance.

H.499Evidence and Values

A transparent report should be willing to say:

The evidence does not dictate one answer. Council prefers Option B because it places greater weight on X than Y.

H.500That Is More Honest Than Pretending

The data made us do it.

H.501Evidence-Based Does Not Mean Value-Free

Important.

H.502Decision Record

For major contested decisions record:

Evidence relied upon

Key uncertainties

Options

Values / priorities

Decision

H.503Minority Position

Can cite:

H.504Evidence Should Not Be Personalized

Disagree with:

Do not attack:

H.505Expert Disagreement

Treat professionally.

H.506Resident Disagreement

Same.

H.507Correction Standard

When a material factual error is identified:

Verify.

Correct.

Date.

Preserve record.

Explain effect where material.

H.508Minor Typo

Proportionate correction.

H.509Material Numerical Error

Public correction.

Public correction.

H.511Material Historical Error

Public correction.

H.512Source Misquotation

Correct.

Update source reference where possible.

H.514New Evidence

Not necessarily prior error.

Could be:

H.515Label

Correction

or

Update

correctly.

H.516Retraction

Use where claim should no longer be relied upon.

H.517Correction Log

Connect to:

H.518No Quiet Deletion

For material public claims.

H.519Error Does Not Mean Bad Faith

No.

H.520Refusal to Correct

More serious.

H.521Research Conflict

If researcher has:

interest:

Disclose where material.

H.522Evidence Can Still Be Valid

Conflict does not automatically invalidate:

H.523It Changes Safeguard

Use:

H.524map.ca Conflict

Particularly important.

H.525RealMap Conflict

Particularly important.

H.526Campaign Conflict

Particularly important during municipal adoption.

H.527Procurement Evidence

Vendor marketing is evidence of:

H.528Not Proof of Performance

No.

H.529Reference Customer

Useful.

H.530Independent Test

Stronger.

H.531Demonstration

Useful.

But controlled demo may not reflect:

H.532Pilot

Stronger.

H.533Contractual Warranty

Different from:

H.534Cybersecurity Certification

Useful.

Not proof of:

H.535Accessibility Conformance Claim

Useful.

Verify for important systems.

H.536Canadian Hosting Claim

Verify:

where material.

H.537Open Source Claim

Verify licence and:

H.538Data Portability Claim

Test.

H.539"AI-Powered"

Not evidence of:

H.540Public Communications

The evidence standard applies to:

H.541Short Format

May simplify.

H.542Simplify Without Distorting

H.543Headline

Should be supportable.

H.544Fine Print Cannot Rescue False Headline

No.

H.545Example

Headline:

City Cuts Wait Times 50%

requires the defined wait-time measure actually:

H.546If Only One Service

Say:

Building permit median City review time fell 50%.

H.547Precision Builds Trust

H.548Public Claim Review

High-profile claims should receive:

before release.

H.549Not Political Censorship

It is:

H.550Mayor's Speech

Political interpretation remains allowed.

H.551Official Fact

Should remain:

H.552Campaign Communications

Campaign may express:

H.553But Campaign Factual Claims

Should still be:

H.554Municipal Research Staff

Should not be used to:

for campaign.

H.555Campaign Team

Should not have privileged access to:

H.556Public Records

Available to:

H.557Election-Year Evidence Freeze

No.

New evidence can emerge.

H.558Election-Year Evidence Standard

Same as:

H.559Do Not Suppress Bad News

No.

H.560Do Not Rush Good News Before Verification

No.

H.561Independent Verification

For especially consequential claims consider:

H.562Independence Has Cost

Use where:

H.563Not Every Claim Needs Third Party

No.

H.564Internal Expertise Is Evidence

Yes.

H.565Distrust of Staff Is Not Evidence Policy

No.

H.566Staff Expertise

Should be respected.

H.567Staff Claim

Still documented.

H.568Front-Line Knowledge

Can identify:

not visible in dashboard.

H.569Institutional Memory

Evidence.

But memories can be:

H.570Document Where Important

H.571"We Have Always Done It This Way"

Not evidence of:

H.572"We Tried That Once"

Ask:

H.573Past Failure

Does not prove permanent:

H.574Past Success

Does not prove current:

H.575Replication

Needs current context.

H.576Source Register Should Include Superseded Documents

Do not delete.

H.577Archive Status

Mark:

H.578Historical Decision Review

Needs old source.

H.579Current Decision

Needs current source.

H.580Source Retention

Follow applicable:

Public sourcing should respect:

Do not republish entire protected work unnecessarily.

H.583Public Domain

Can still require:

H.584Open Licence

Follow terms.

H.585Indigenous Cultural Material

Copyright law may not capture all:

obligations.

Respect agreed protocols.

H.586Confidential Business Information

Protect where legally required.

H.587Procurement Confidentiality

Does not mean every procurement fact remains:

Protect.

H.589Personal Information

Protect.

H.590Evidence Transparency Has Limits

Yes.

H.591But Conclusions Still Need Public Basis

Where possible.

H.592Confidential Evidence

A public report can state:

Confidential legal, commercial or security information was reviewed

without disclosing protected detail where lawful.

H.593No Secret Evidence Theatre

Do not use vague:

confidential information proves us right

as political shield.

H.594Independent Oversight

Can help where public disclosure impossible.

H.595Evidence Challenge Process

Residents should have a way to flag:

H.596Challenge Does Not Guarantee Change

Review.

H.597Correction Response

Explain:

H.598Public Source Notes

Can be updated.

H.599Evidence Dispute Register

For major unresolved disputes, useful.

H.600Example Fields

Claim

Source A

Source B

Nature of conflict

Current City position

Next verification

H.601Avoid False Certainty

If unresolved:

Say.

H.602"Best Available Evidence"

Useful phrase when:

H.603But Explain Limit

H.604Evidence Threshold for Action

Not all decisions require:

H.605Low-Risk Reversible

Can act with:

H.606High-Risk Irreversible

Needs:

H.607Emergency

May require action with:

H.608Emergency Action

Document:

H.609After-Action Review

Update.

H.610Precautionary Action

Can be legitimate.

H.611Sunset

Can make uncertain intervention:

H.612Pilot

Same.

H.613Review Date

Same.

H.614Evidence and Procurement

Procurement specifications should be based on:

H.615Vendor Feature

Do not become requirement because:

H.616Technology Hype

Not evidence.

H.617Reference Architecture

Useful.

H.618Proof of Concept

Useful.

H.619Production Test

Stronger.

H.620Evidence and Asset Decisions

Asset sale:

Needs:

H.621Asset acquisition:

Needs:

H.622Infrastructure priority:

Needs:

H.623Evidence and Housing

Separate:

H.624Evidence and Business

Separate:

H.625Evidence and Downtown

Separate:

H.626Evidence and Safety

Separate:

H.627Evidence and Youth

Separate:

H.628Evidence and Accessibility

Separate:

H.629Evidence and Privacy

Separate:

H.630Evidence and Environment

Separate:

H.631Evidence and Partnership

Separate:

H.632Evidence and Satisfaction

Separate:

opinion.

H.633Evidence and Completed Commitments

Original wording is primary source for:

H.634Completion Evidence

Must match:

H.635Evidence and Missed Commitments

Do not change original:

H.636Reason for Miss

Can be supported by:

H.637Attribution Discipline

For every achievement ask:

Who actually did it?

H.638City

Credit appropriately.

H.639County

Credit.

H.640Ontario

Credit.

H.641Canada

Credit.

H.642SON

Do not appropriate.

H.643Community Partner

Credit.

H.644Private Business

Credit.

H.645Resident

Credit where appropriate.

H.646Shared Achievement

Call:

H.647No Credit Theft

Important.

H.648No Blame Shifting

Also.

H.649Evidence Register Publication

A public source appendix should make major sources:

H.650Not Every Internal Source Should Be Public

Privacy, privilege, security and contractual restrictions apply.

H.651Public Source Identifier

Use stable:

H.652Source Note

Could include:

Current as of August 2026

where appropriate.

H.653Current-Law Appendix

Should be refreshed before:

H.654Current-Program Claims

Refresh.

H.655Current-Office-Holder Claims

Avoid where unnecessary.

H.656Institutional Role

Often more durable than:

H.657Example

Prefer:

Minister responsible for...

only if name not material.

H.658Name When Material

Verify current.

H.659Change Detection

Source register should flag:

H.660Review Cadence

Possible:

Before Every Major Decision

Quarterly

Annual

Before Final Publication

On Legal Change

H.661First 30 Days

Create:

H.662First 60 Days

Audit major claims already used in:

H.663First 100 Days

Establish:

H.664Year One

Prioritize verification of:

H.665Year Two

Strengthen:

H.666Year Three

Review:

H.667Year Four

Publish:

Four-Year Evidence Integrity Review.

H.668Evidence Integrity Review

Should answer:

Which major claims were corrected?

Which were superseded?

Which remained disputed?

Which forecasts were wrong?

Which early hypotheses were confirmed?

Which were rejected?

Which sources proved unreliable?

Which evidence gaps remain?

Which decisions were changed by new evidence?

H.669Name a Major Claim Corrected

Where applicable.

H.670Name a Major Forecast That Missed

Where applicable.

H.671Name a Major Hypothesis Rejected

Where applicable.

H.672Name a Major Uncertainty Resolved

H.673Name a Decision Stopped Because Evidence Changed

Where applicable.

H.674Name a Decision Accelerated Because Evidence Strengthened

Where applicable.

H.675Name a Resident Concern Later Confirmed by Data

Where applicable.

Where applicable.

H.677Name a Technical Report That Changed City Priority

Where applicable.

H.678Name a Source Error Corrected Publicly

H.679Name the Largest Remaining Evidence Gap

H.680Name the Most Important Time-Sensitive Claim for Next Council to Reverify

H.681Handoff

The next Council should inherit:

Source Register

Claim Register

Verification dates

Current-law list

Major disputed claims

Major forecast assumptions

Methodology notes

Correction history

Archived sources

H.682Research Should Survive Turnover

Yes.

H.683No "Ask the Former Mayor"

Institutional evidence should not depend on:

H.684Anti-Gaming Rule One

Do not cite a source that does not support the claim.

H.685Rule Two

Do not quote a source out of context.

H.686Rule Three

Do not omit a material qualification because it weakens:

H.687Rule Four

Do not use an outdated statute to describe:

H.688Rule Five

Do not use current statute to describe historical law without checking:

H.689Rule Six

Do not describe a bill as:

H.690Rule Seven

Do not describe a program announcement as:

without verification.

H.691Rule Eight

Do not describe an application as:

H.692Rule Nine

Do not describe awarded funding as:

unless received.

H.693Rule Ten

Do not describe a consultant recommendation as:

H.694Rule Eleven

Do not describe a staff recommendation as:

H.695Rule Twelve

Do not describe an agenda item as:

H.696Rule Thirteen

Do not describe a draft as:

H.697Rule Fourteen

Do not describe a primary source's assertion as:

merely because it is primary.

H.698Rule Fifteen

Do not describe journalism reporting an allegation as:

H.699Rule Sixteen

Do not describe social-media repetition as:

H.700Rule Seventeen

Do not count ten sources copied from one original as:

H.701Rule Eighteen

Do not describe a press release as stronger than the underlying:

where available.

H.702Rule Nineteen

Do not describe a self-selected survey as:

H.703Rule Twenty

Do not use a margin of error where survey methodology does not support:

H.704Rule Twenty-One

Do not use anecdote as population statistic.

H.705Rule Twenty-Two

Do not dismiss lived experience because it is not:

H.706Rule Twenty-Three

Do not use one technical metric to erase contradictory lived experience without:

H.707Rule Twenty-Four

Do not use public opinion to determine technical safety facts.

H.708Rule Twenty-Five

Do not use technical expertise alone to decide political values.

H.709Rule Twenty-Six

Do not describe correlation as:

H.710Rule Twenty-Seven

Do not upgrade:

into:

without new evidence.

H.711Rule Twenty-Eight

Do not upgrade forecast into:

H.712Rule Twenty-Nine

Do not present best-case scenario as:

H.713Rule Thirty

Do not hide forecast assumptions.

H.714Rule Thirty-One

Do not selectively publish only favourable consultant findings.

H.715Rule Thirty-Two

Do not commission another study solely because existing study gives:

H.716Rule Thirty-Three

Do not refuse new study solely because old study gives:

when old work is obsolete.

H.717Rule Thirty-Four

Do not commission expensive research where a reversible low-cost pilot can answer:

H.718Rule Thirty-Five

Do not pilot high-risk irreversible action merely to avoid:

H.719Rule Thirty-Six

Do not use AI as authoritative source.

H.720Rule Thirty-Seven

Do not publish AI-generated citations without verification.

H.721Rule Thirty-Eight

Do not upload confidential, personal or culturally sensitive information to unapproved AI tools.

H.722Rule Thirty-Nine

Do not present AI-generated conceptual images as:

H.723Rule Forty

Do not call vendor marketing:

H.724Rule Forty-One

Do not assume security certification means:

H.725Rule Forty-Two

Do not assume accessibility vendor claim means:

H.726Rule Forty-Three

Do not assume "Canadian hosted" answers every:

question.

H.727Rule Forty-Four

Do not use campaign research as independent municipal due diligence.

H.728Rule Forty-Five

Do not use private founder materials as sole evidence for municipal adoption of founder-associated technology.

H.729Rule Forty-Six

Do not reduce Indigenous knowledge to an evidence ranking designed for technical reports.

H.730Rule Forty-Seven

Do not publish Indigenous knowledge without appropriate:

H.731Rule Forty-Eight

Do not use only government sources for claims about:

H.732Rule Forty-Nine

Do not use only resident anecdotes for claims about:

H.733Rule Fifty

Do not hide material evidence conflict.

H.734Rule Fifty-One

Do not present disputed evidence as:

H.735Rule Fifty-Two

Do not present low-confidence evidence with:

H.736Rule Fifty-Three

Do not keep superseded sources marked:

H.737Rule Fifty-Four

Do not delete superseded evidence needed for:

H.738Rule Fifty-Five

Do not quietly change official source after public claim is challenged.

H.739Rule Fifty-Six

Do not refuse to correct a factual error because:

H.740Rule Fifty-Seven

Do not call an update:

if previous statement was accurate at the time.

H.741Rule Fifty-Eight

Do not call a correction merely:

to avoid acknowledging error.

H.742Rule Fifty-Nine

Do not hide behind confidential evidence to make broad public claims that cannot be meaningfully scrutinized.

H.743Rule Sixty

Do not publish confidential evidence merely to win:

H.744Evidence Integrity Test

Before publishing a major claim ask:

What exactly are we claiming?

H.745Source Test

What source is closest to the fact?

H.746Authority Test

Is the source authoritative for this specific claim?

H.747Currency Test

Could the information have changed?

H.748Date Test

When was it last verified?

H.749Independence Test

Are multiple sources genuinely independent?

H.750Method Test

How was the result produced?

H.751Scope Test

What did the source actually examine?

H.752Limitation Test

What does it not prove?

H.753Conflict Test

Is there credible contradictory evidence?

H.754Confidence Test

How sure should we be?

H.755Language Test

Are we using stronger words than the evidence supports?

H.756Causation Test

Are we claiming cause when we only know association?

H.757Forecast Test

Are assumptions visible?

H.758Resident Test

Does quantitative evidence miss an important lived experience?

H.759Technical Test

Does lived experience require professional or technical verification before a high-risk conclusion?

H.760Indigenous Knowledge Test

Are we respecting source, consent, context and governance?

H.761Conflict-of-Interest Test

Does anyone providing evidence stand to benefit from the decision?

H.762AI Test

Has a human verified the source and claim rather than trusting generated text?

H.763Public Test

Could a reasonable resident trace the major claim to its evidence?

H.764Reverse Test

Would we accept this same evidence standard if the conclusion favoured a political opponent?

H.765Correction Test

If this claim proves wrong, are we prepared to correct it publicly?

H.766Decision Test

Is the evidence strong enough for the reversibility, cost and risk of the decision being made?

H.767Evidence Standard for Council

A major Council report should be able to say:

Here is what we know.

Here is how we know it.

Here is what remains uncertain.

Here are the competing interpretations.

Here is the recommended decision.

H.768That Is Evidence-Informed Government

Not:

H.769The Sources, Evidence and Verification Commitment

Owen Sound should commit to:

Use the strongest practical source for the claim being made.

Prefer sources closest to the underlying fact.

Recognize that evidence quality is claim-specific rather than governed by one simplistic hierarchy.

Distinguish authoritative legal sources, administrative records, financial records, technical evidence, research, surveys, journalism, resident experience, anecdote and unverified information.

Use current consolidated constitutional, statutory and regulatory sources for current-law claims.

Use the version of law actually in force at the historical time when making historical legal claims.

Distinguish enacted law, regulations, proposed legislation and provisions not yet in force.

Use municipal by-laws, resolutions and approved minutes to establish municipal decisions.

Never treat a staff recommendation as a Council decision.

Never treat an agenda item as an approved policy.

Never treat a draft report as final.

Use Finance and audited records for significant financial claims.

Never use campaign rhetoric as a substitute for official financial evidence.

Require Appendix E verification before publicly describing a financial benefit as a saving.

Use qualified professional evidence for significant technical claims.

Read the scope, limitations, assumptions and date of technical reports before relying on their conclusions.

Recognize that professional expertise informs but does not replace democratic policy judgment.

Acknowledge legitimate expert disagreement rather than concealing it.

Use official Grey County, Ontario, federal and regulatory sources for claims about those institutions wherever practical.

Do not assume the existence of a government program establishes Owen Sound's eligibility.

Distinguish program announcements, open applications, awards, agreements and funds actually received.

Use original agreements, contracts, title documents, permits, invoices, datasets and records where they are the best evidence.

Recognize that primary sources establish what was said or recorded but can still contain error or perspective.

Use photographs and video carefully with appropriate date, location and provenance.

Never present synthetic or AI-generated imagery as documentary evidence.

Use academic research to inform policy while recognizing context, limitations, disagreement and transferability.

Prefer Canadian and local evidence where institutional or geographic context materially affects relevance.

Distinguish mandatory professional standards from voluntary best practices and guidance.

Use representative surveys only where methodology supports population-level inference.

Label self-selected polls and consultations accurately.

Publish question wording, sample size, dates and methodology for important public-opinion measures.

Use partner-reported evidence where useful but label it when the City has not independently verified it.

Keep grant and partnership reporting proportionate to the amount of public money and risk involved.

Do not treat Indigenous knowledge as inherently inferior because it does not fit a Western technical evidence hierarchy.

Use legal sources for legal rights and use Saugeen Ojibway Nation's own sources, knowledge and agreed records for SON's history, perspective and priorities where appropriate.

Protect confidential Indigenous knowledge and respect consent over use.

Avoid extractive collection of Indigenous knowledge merely to populate municipal reports.

Use journalism as an important secondary source while seeking underlying primary evidence for material claims where available.

Distinguish allegations from findings, charges from convictions and lawsuits from judgments.

Treat resident testimony as legitimate evidence of lived experience.

Do not generalize one resident's experience into a population statistic without supporting evidence.

Use recurring resident stories as signals worthy of investigation rather than dismissing them because they are not formal datasets.

Recognize that residents who do not complain may still face barriers.

Do not describe consultation participants as the whole City unless the method justifies that conclusion.

Use anecdote to generate questions rather than manufacture statistics.

Treat social media as evidence of what was posted, not automatic proof of what was claimed.

Do not treat repetition across websites or social media as independent corroboration.

Trace repeated claims back to their underlying sources.

Use the evidence-status vocabulary Verified, Supported, Preliminary, Disputed, Unverified, Superseded and Unknown.

Use confidence levels only where they improve understanding.

Match confidence language to the strength and relevance of the evidence.

Classify major claims as legal, financial, technical, operational, statistical, historical, causal, forecast, perception, policy preference or lived experience where doing so helps select the correct evidence.

Do not pretend value judgments are empirical facts.

Use stronger evidence for higher-consequence claims.

Require particularly strong support for legal, safety, major financial, major causal and irreversible asset claims.

Use cautious language such as could, may, expected, projected, estimated, associated with, contributed to and caused according to the actual strength of evidence.

Use caused and proved only where evidence justifies them.

Record verification dates for current law, programs, ownership, funding and other time-sensitive facts.

Ask whether information could have changed before repeating it as current.

Maintain a central Evidence and Source Register.

Maintain a Claim Register for the most consequential or time-sensitive factual assertions.

Use proportionality so ordinary statements do not generate unnecessary bureaucracy.

Assign verification triggers to claims likely to change.

Prevent claim drift from cautious evidence to exaggerated political certainty.

Do not strengthen a claim without strengthening the evidence.

Use consistent citations in the final compiled plan.

Prefer stable footnotes, endnotes or structured source references over raw URLs scattered through policy text.

Record source title, issuer, date and relevant location so the evidence remains identifiable if a link changes.

Archive important public sources where lawful and practical.

Never archive or expose confidential material publicly merely for citation convenience.

Verify quotations against the original text.

Do not remove qualifications from quotations in ways that change meaning.

Read enough surrounding context to understand what the source actually says.

Give important numbers a source, period and unit.

Never publish a percentage without knowing the denominator.

Never use a dollar figure without understanding whether it is gross, net, annual, capital, lifecycle or City share where that distinction matters.

Resolve conflicting evidence by checking source authority, dates, definitions, independence and methodology.

Do not hide material credible disagreement.

Recognize that newer evidence may be more current without automatically being better evidence for every purpose.

Recognize that official information may be authoritative for an institution's action while still requiring independent analysis for broader conclusions.

Respect lived experience even where average quantitative results appear positive.

Use current datasets and preserve the information available at the time of major past decisions when conducting accountability reviews.

Distinguish forecasts, scenarios and predictions.

Publish assumptions for significant forecasts.

Never present best-case scenarios as expected results without justification.

Review major forecasts against actual results so the City learns from forecasting error.

Use caused, contributed to and coincided with carefully when discussing outcomes.

Do not infer causation merely from before-and-after change.

Define pilot hypotheses, baselines, success measures and stop conditions before launch.

Do not declare pilots successful merely because launch was popular.

Treat failed pilots as useful evidence rather than burying them.

Evaluate whether programs were implemented, used, effective, affordable and associated with unintended consequences.

Use triangulation where technical, financial, operational and lived-experience evidence answer different parts of the same question.

Do not count sources as votes.

Evaluate relevance, authority, methodology, currency and independence.

Disclose material interests or funding relationships behind evidence.

Do not automatically reject evidence because its author has an interest, but strengthen independent verification where conflict exists.

Apply heightened independent evidence standards to founder-associated map.ca and RealMap proposals.

Never use founder enthusiasm as proof of municipal need, fair value, security, privacy or accessibility.

Do not allow a private commercial business plan to silently become a municipal business case.

Keep campaign proposals and municipal verification institutionally separate.

Use the campaign platform as evidence of what was promised, not proof that the policy itself is legally, financially or technically sound.

Apply the same evidence standards to incumbent and opposition claims.

Use AI to locate, organize, compare and summarize sources, not as the source of truth.

Never cite AI as proof of law, financial fact, engineering fact or other authoritative claim.

Verify every material AI-generated citation, quotation, date and factual assertion against the underlying source.

Do not allow AI summaries to replace reading the underlying source where exact wording or technical detail matters.

Require qualified human review for high-stakes legal, financial, technical and privacy decisions.

Never place confidential, personal, privileged or culturally sensitive information into unapproved AI tools.

Keep accountability with the human public institution rather than transferring it to an AI system.

Maintain stronger research records for high-stakes AI-assisted work.

Preserve important source documents and version information in accordance with records requirements.

Use screenshots only as evidence of what was displayed at a point in time and prefer original authoritative documents where available.

Make major public evidence accessible without requiring residents to purchase expensive subscriptions where an equally strong public source exists.

Provide plain-language explanations without replacing the underlying technical evidence.

Handle historical evidence with attention to period, terminology, context and multiple legitimate knowledge sources.

Do not automatically privilege colonial written records over Indigenous oral history for questions of community history and perspective.

Use proper legal evidence for legal conclusions arising from historical matters.

Verify current territorial, treaty, harbour and ownership claims before repeating historical formulations as current fact.

Require major Council reports to separate What We Know, What We Estimate, What We Do Not Know and What We Recommend.

Allow Council to make decisions under uncertainty when necessary while making the uncertainty visible.

Prefer reversible pilots where evidence is weak and the decision can safely be tested.

Require stronger evidence before irreversible decisions such as major land sales, substantial borrowing, demolition, permanent closures, major surveillance systems or significant asset acquisitions.

Use evidence proportional to the cost and risk of the decision.

Do not spend more researching a small reversible problem than the problem reasonably warrants.

Do not underinvest in due diligence where the City could inherit a large long-term liability.

Before commissioning a new study, identify the unanswered question and what decision depends on it.

Search existing institutional studies before paying to repeat work.

Do not commission a new study solely because the previous answer was politically inconvenient.

Do not refuse to update an obsolete study because the old answer is politically convenient.

Use evidence to support decisions rather than postpone decisions indefinitely.

Use public consultation to understand experience, preference and values, not as a substitute for engineering, law or accounting.

Use professional evidence to understand technical issues without pretending professionals alone should decide political values.

Record major decisions so residents can see both the evidence and the value judgment applied to it.

Correct material factual errors openly.

Distinguish corrections from legitimate updates based on newly changed information.

Retract claims that should no longer be relied upon.

Maintain the Correction Log.

Never allow political embarrassment to prevent factual correction.

Use independent verification when private conflicts make internal evidence insufficient.

Treat vendor marketing as vendor claims rather than independent performance evidence.

Test important vendor claims relating to security, accessibility, data portability, Canadian hosting and operational performance.

Do not treat labels such as AI-powered, secure, accessible, Canadian or open source as proof of public value.

Apply evidence standards to press releases, speeches, dashboards and social media as well as formal Council reports.

Simplify public communication without overstating what the evidence supports.

Do not use fine print to rescue a misleading headline.

Keep municipal research resources separate from campaign opposition research.

Do not suppress unfavourable verified evidence during an election year.

Do not rush favourable information into publication before verification because an election is approaching.

Use independent external review selectively where the consequence and conflict justify the cost.

Respect internal professional expertise rather than assuming outside consultants are inherently more credible.

Preserve front-line institutional knowledge while verifying memory against records where important.

Never accept We Have Always Done It This Way as proof that a process remains necessary.

Never accept We Tried That Once as proof that an idea can never work without examining what was actually tried.

Keep superseded sources for historical accountability while marking them clearly as superseded.

Respect copyright, licensing, privacy, legal privilege, confidential business information and Indigenous cultural protocols when preserving or publishing evidence.

Do not use confidentiality as a vague shield for public claims that cannot be meaningfully explained.

Provide residents a route to report broken citations, outdated sources, mathematical errors and factual inaccuracies.

Maintain a clear process for reviewing and correcting evidence challenges.

Use an Evidence Dispute Register for major unresolved factual disputes where it improves accountability.

Use the phrase Best Available Evidence where appropriate while explaining material limitations.

Match the evidence threshold to the cost, risk and reversibility of the decision.

Document what was known and unknown when emergency action must occur before complete evidence is available.

Use sunset clauses and review dates where uncertainty warrants them.

Base procurement specifications on public need rather than favourite vendor features.

Require title and valuation evidence for material land transactions.

Require condition and liability evidence for significant asset acquisitions.

Require condition, confidence, criticality and lifecycle evidence for infrastructure priorities.

Maintain the housing evidence chain from inquiry to occupancy.

Maintain business evidence that distinguishes enquiries, licences, openings, closures, relocations and expansions.

Maintain downtown evidence that distinguishes perception, measured vacancy, public-realm conditions and safety information.

Maintain safety evidence that distinguishes harm, calls, response, enforcement activity and perception.

Maintain youth evidence that distinguishes opportunities, participation, completion and next steps.

Maintain accessibility evidence that distinguishes review, accommodation and barriers actually removed.

Maintain privacy evidence that distinguishes assessment, controls, incidents and actual remediation.

Maintain environmental evidence that distinguishes activities from outcomes.

Maintain partnership evidence that distinguishes meetings, agreements, money, deliverables and resident results.

Maintain satisfaction evidence that distinguishes representative surveys, self-selected feedback and service-user experience.

Preserve original wording as the primary evidence for campaign and public commitments.

Require completion evidence to match the substance of what was actually promised.

Give credit to the institution or community that actually delivered the result.

Do not appropriate the accomplishments of Grey County, Ontario, Canada, Saugeen Ojibway Nation, community partners, residents or private businesses.

Do not shift blame to those institutions for failures the City controlled.

Maintain a public source appendix for the major claims in the Four-Year City Business Plan.

Use stable document identifiers rather than relying solely on web links.

Reverify current-law, program, ownership and funding claims before final publication.

Flag time-sensitive claims for future review.

Use the first 30 days to establish the Source Register, Claim Register and current-law review list.

Use the first 60 days to audit high-risk factual claims already being carried from planning or campaign materials into government.

Use the first 100 days to establish citation, correction, evidence-status and AI-assisted research standards.

Use Year One to establish strong legal, financial, asset and service baselines.

Use Year Two to strengthen survey, partner and outcome evidence.

Use Year Three to revisit forecasts and evidence underlying long-term commitments.

Use Year Four to publish a Four-Year Evidence Integrity Review.

Name important evidence failures as openly as successful verification.

Give the next Council the sources, claims, assumptions, unresolved disputes and verification dates needed to continue governing from evidence rather than institutional memory.

Apply the final evidence test to every major claim: What exactly are we saying, what is the strongest appropriate source, how current is it, what does it actually prove, what does it not prove, and would we apply the same evidence standard if the conclusion were politically inconvenient?

The evidence standard can therefore be reduced to a few rules:

Go to the original source.

Check the date.

Use the source that actually has authority over the fact.

Separate fact from forecast, opinion and experience.

Respect lived experience without turning anecdote into statistics.

Respect technical evidence without turning experts into elected government.

Show uncertainty.

Correct mistakes.

Verify AI output.

Use stronger evidence for stronger claims.

Keep the source record.

The purpose is not to produce a government that refuses to act until it possesses:

Perfect information rarely exists.

The purpose is to produce a government that knows the difference between:

Then Council can make the decision.

And residents can judge whether the decision was:

The stronger the public claim, the stronger the evidence should be. Use the source closest to the fact. State what remains uncertain. Never manufacture certainty for political convenience.

← Appendix G: Scorecard Definitions and Measurement StandardsAppendix I: Legal Review and Decision Standards →