Where we cannot see
Voidly publishes a censorship score for 121 of the 130 countries it covers. The other 9 do not have enough measurement coverage in the last 7 days to score, so no score is published for them.
This page exists because the alternative is worse. Until 2 August 2026 those countries were scored 0 and labelled free — so this site published North Korea as having “minimal” censorship, on the strength of zero measurements. An absence of data is not an absence of censorship, and the gap is not random: it is widest exactly where measurement is most dangerous.
How the threshold works
A country is scored only when it has at least 100 measurements in the rolling 7-day window. Below that the anomaly rate is too noisy to mean anything — a single unlucky measurement in a country with three of them produces a 33% anomaly rate, which would outrank China.
In the API these countries carry score: null and level: "insufficient_coverage", never 0. If you are consuming the API programmatically, branch on scored rather than treating a missing score as zero.
The 9 countries we cannot score
Sorted by risk tier, then by how little data there is. Scores generated 2026-08-02; this list moves as the window slides.
| Country | Risk tier | Measurements (7d) | All-time |
|---|---|---|---|
| EritreaER | T1SevereSevere | 0 | — |
| North KoreaKP | T1SevereSevere | 0 | — |
| TurkmenistanTM | T1SevereSevere | 0 | 307 |
| HTHT | T3MediumMedium | 0 | 3 |
| NicaraguaNI | T3MediumMedium | 0 | 53,895 |
| LaosLA | T3MediumMedium | 16 | 12,517 |
| RwandaRW | T3MediumMedium | 36 | 11,141 |
| MozambiqueMZ | T3MediumMedium | 41 | 7,055 |
| BahamasBS | T5FreeFree | 0 | — |
Why the gap is shaped like this
Almost all network-censorship measurement depends on someone inside the country running a probe. That is a personal risk that scales with how repressive the network is, so volunteer coverage thins out precisely where the measurement would matter most. Every project in this field has the same problem; Voidly is not unusual in having it, only in publishing it.
6 of these countries produced no measurements at all in the window. The remaining 3 produced some, but too few to distinguish signal from noise.
Remote measurement — techniques that observe a network from outside it, without anyone inside running anything — is the partial answer, and Voidly already ingests one such source (CensoredPlanet). Extending that coverage is the most valuable open problem this project has.
Using this
- Do not read a gap as an all-clear. If you are writing about a country on this list, the honest sentence is “no recent independent measurements are available”, not “no censorship was detected”.
- Machine-readable:
api.voidly.ai/data/censorship-index.jsoncarriesscored,insufficientCoverageCountandscoreMinMeasurements. - Run a probe. The fastest way to close a gap is a measurement from inside it — see the probe network. Read the risks first; in some of these countries running one is not safe.
- Methodology: how every number here is computed.
130 countries covered · 121 scored · 9 insufficient coverage · CC BY 4.0