Source Hierarchy

The four-tier evidence classification

Every piece of evidence in a Veridi assessment is classified by source quality. The tier determines the maximum confidence that evidence can support, regardless of how many sources at that tier agree.

Tier 1: Primary Sources

Government databases, peer-reviewed research, court records, official statistics, raw authoritative data.

Examples: Census data, FDA clinical trial results, court filings, IPCC assessment reports, central bank interest rate announcements.

Confidence ceiling: 95% (with multiple Tier 1 sources in agreement).

Tier 2: Authoritative Secondary Sources

Wire services (AP, Reuters, AFP), established fact-checkers (IFCN-affiliated), credentialed domain experts speaking within their expertise, institutional reports from recognized bodies.

Examples: AP investigation, PolitiFact assessment, a published epidemiologist discussing disease transmission, World Bank economic analysis.

Confidence ceiling: 80%.

Tier 3: General News Media

Mainstream news reporting, newspaper articles, broadcast news, established digital news outlets.

Examples: Major newspaper coverage, national broadcast news reports, established online news sites.

Confidence ceiling: 65%.

Tier 4: Lowest Reliability

Social media posts, anonymous sources, partisan outlets, opinion pieces presented as news, self-published content.

Examples: Twitter/X threads, anonymous blog posts, advocacy organization press releases, forum posts.

Confidence ceiling: 50%.


How the ceilings work

The confidence ceiling is structural and non-negotiable. If the best evidence for a claim comes from Tier 3 sources - even ten of them, all in agreement - confidence cannot exceed 65%.

This design choice reflects a specific epistemological position: corroboration from low-quality sources does not equal high-quality evidence. Ten news articles that all cite the same press release are one source, not ten. Multiple social media posts repeating the same claim create the appearance of consensus without the substance.

The ceilings can combine:

Sourcing levelConfidence ceiling
Multiple Tier 1 sources in agreement95%
Tier 1 + Tier 2 corroboration90%
Tier 2 sources only80%
Tier 3 with corroboration65%
Tier 4 only50%
No sourcing / assertion only25%

Source independence

The methodology requires verifying that sources are genuinely independent: different reporting, different ownership, different access to the underlying information, different timestamps. This addresses confidence laundering, where a single unreliable origin is amplified through multiple outlets that create the false appearance of independent confirmation.

When institutions degrade

The source hierarchy assumes that a Tier 1 source is producing reliable output. When that assumption breaks - due to political interference, defunding, or institutional capture - the methodology applies the Institutional Reliability Index to adjust the effective tier.

A government agency assessed at Level 2 or Level 3 degradation may have its output reclassified from Tier 1 to Tier 2 or Tier 3 on specific topics, with comparison anchors (international equivalents, independent monitoring systems) consulted as alternative primary sources.

This adjustment is per-agency and per-function. An agency may remain Tier 1 for routine administrative data while being downgraded on politically sensitive topics where interference has been documented.

Detailed IRI documentation