Our Verification Methodology
Lyfye Facts uses a rigorous, evidence-based approach to verify claims. Here's how our verification engine works.
Developed by Lyfye R&D
Source Hierarchy
Not all sources are created equal. We classify sources into tiers based on their authority and reliability:
Tier 1: Authoritative Sources
These sources have the highest weight in our verdicts. Tier 1 evidence can override conflicting Tier 2/3 sources.
- Official government sources (.gov, .mil)
- Academic institutions (.edu)
- International organizations (UN, WHO, World Bank)
- Government statistical agencies (Census, BLS, CDC)
Tier 2: Reputable Sources
Trusted journalism and research organizations. Strong supporting evidence when Tier 1 is unavailable.
- Major wire services (Reuters, AP News, BBC)
- Fact-checking organizations (Snopes, PolitiFact, FactCheck.org)
- Peer-reviewed journals (Nature, Science)
- Research institutions (Pew, RAND, Brookings)
Tier 3: Contextual Sources
General web sources that provide context but cannot determine verdicts on their own.
- •Wikipedia (encyclopedic context)
- •News articles from regional outlets
- •Industry publications and blogs
Excluded: Social Media
Social media platforms (Twitter/X, Facebook, Reddit, TikTok) are excluded from our evidence set. User-generated content is not considered reliable for verification purposes.
Multi-Query Search Strategy
For each claim, we run 6 specialized searches to maximize coverage:
Verdict Classifications
Every verification resolves to one of these six evidence verdicts. The tone you read is cold and the conviction comes from the sources, never from louder language.
The evidence supports the claim as stated. This checks out.
The evidence contradicts the claim. This is false.
The literal words may be defensible, but the claim is constructed to mislead by cherry-picking or stripping context. Receipts (what they said, what they left out, how it misleads) are required for this verdict.
The claim is materially true but incomplete in a way that changes the picture. True, but you're not getting the whole story.
The available sources cannot back this claim up either way. Nobody can back this up yet.
Credible sources of comparable authority genuinely conflict, and the engine documents the split. The real sources don't agree.
OPINION is a future non-verdict outcome (for value, opinion, or prediction claims) and is not one of the six evidence verdicts.
Limitations & Transparency
We do not claim omniscience. Our verifications are based on publicly available evidence at the time of analysis. New information may emerge that changes conclusions.
Confidence is capped. If no Tier 1 sources are found, our confidence is automatically limited to 65% maximum.
Domain diversity is enforced. We limit results to 2 per domain (3 for Tier 1) to prevent any single source from dominating the analysis.
We cite our sources. Every verification includes full citations with links so you can verify our findings independently.