RULE: Every objective claim in Candid content carries a named source + date + verbatim quote ≤25 words + confidence label

Rule: Every objective claim in Candid Creative client deliverables, KB entries, and Candid's own public writing carries:

1. A named source (person + institution where applicable)
2. A publication or retrieval date
3. A verbatim quote ≤25 words where possible (the link-rot survival layer)
4. A confidence label from the CANDID REFERENCE: 7-label confidence taxonomy — Verified / Industry-consensus / Single-source / Estimated / Author's view / Contested / Stale

Why:

• Reader trust default in 2026 is distrust (Edelman 2025 Trust Barometer (n=33,000, 28 countries): 7 in 10 believe government/business/journalists deliberately mislead them — 7 in 10 believe leaders mislead)
• BBC §3.2.2, Reuters Handbook, SPJ Code all converge on named-source-preferred (BBC Editorial Guidelines §3.2.2: "All BBC output...must be well sourced, based on sound evidence, thoroughly tested", Reuters Handbook of Journalism: "A named source is always preferable to an unnamed source. Anonymous sources are the weakest sources", SPJ Code of Ethics (2014): "Identify sources clearly. The public is entitled to as much information as possible to judge reliability and motivations")
• Wikipedia's "verifiability not truth" model (Wikipedia: "The threshold for inclusion is verifiability, not truth" — four claim types always need inline citations) is the largest working example
• FTC reasonable-basis doctrine (FTC reasonable-basis doctrine: advertisers must possess "the amount and type of substantiation the ad actually communicates") makes unsourced objective claims legally actionable
• AI engines preferentially cite content with structured citations, named sources, explicit dates (Profound (Aug 2024-Jun 2025, 680M citations): only 11% domain overlap between ChatGPT and Perplexity; 13.7% between AI Overviews and AI Mode)
• The verbatim quote is the highest-leverage link-rot defense (CANDID REFERENCE: 9-step link-rot mitigation plan — archive on capture, verbatim quote, persistent IDs, quarterly check)

How to apply:

• Use CANDID REFERENCE: "what to source" checklist — Must / Should / Doesn't need as the decision matrix for what needs sourcing
• Use CANDID REFERENCE: 8-pattern citation library — inline links, hover footnotes, end-of-section, archive pairs, schema for the format (Candid default: inline parenthetical for research/KB, hover footnotes for long-form, dedicated source page for marketing)
• The test: if a skeptical reader asks "where did that come from?" and you can't answer in one sentence with a name and a date, it needs sourcing
• For LLM-assisted drafting: every claim the LLM produces must be traced to a named source before it ships — the Simhi et al. finding (Simhi et al. (Technion/Oxford/Hebrew U, Feb 2025): "models can hallucinate with high certainty even when they have the correct knowledge") is the reason