Methodology
Source Classification
Sources are classified by type, not by a single trust score:
- Official primary source (documentation, model cards, security advisories, pricing pages)
- Peer-reviewed or formal research
- Independent benchmark or test
- Reputable technical journalism
- Specialist analysis
- Community report
- Anonymous or unverifiable claim
Evidence Dimensions
Each claim is evaluated across multiple dimensions, never reduced to a single score:
- Primary-source support
- Methodology quality
- Reproducibility
- Independent corroboration
- Source independence
- Conflict of interest
- Timeliness
- Version specificity
- Claim severity (extraordinary claims require extraordinary evidence)
- Contradictory evidence
- Correction history
Public Rating Labels
Confirmed — Supported by multiple independent primary sources
Strongly supported — Strong primary-source support with some corroboration
Vendor-reported — Claimed by the vendor; independent verification pending or not available
Unverified — Claim has not yet been independently checked
Disputed — Credible contradictory evidence exists
Corrected — Previously published information has been updated
Outdated — Information is no longer current
Corrections Policy
Corrections are public and permanent. When we correct a claim, rating, or recommendation, the original statement remains visible alongside the correction, the date, the reason, and the evidence that prompted the change. We do not silently rewrite history.
View the corrections ledger →
Monetisation and Independence
Advertisers, sponsors, and affiliates cannot purchase higher ratings, better rankings, or suppressed negative findings. Rankings and evidence ratings are structurally separated from commercial relationships.
Read the full editorial independence policy →
This methodology will be expanded as the platform develops. Last reviewed: 12 July 2026.