Methodology
How we measure, classify, and qualify our findings
This overview describes the principles behind our published work. Individual audits link to the specific methodology version used.
Evidence standard
Every published status is supported by a written label, a distinguishing shape or icon, and an explanation — never by colour alone. We separate the observed result, the confidence in that result, the number of observations, and the date it was last updated.
What we do not claim
- A timeout is not treated as proof of blocking.
- A DNS anomaly is not treated as proof of intent.
- A single failed probe is not treated as a country-level conclusion.
- Accessibility is not treated as proof of compliance.
- Correlation is not treated as attribution.
Reproducibility
Each release records its schema version, methodology version, data cut-off, and generation time. Where practical, the underlying measurements can be reviewed or re-run so that our conclusions can be independently verified.