The coverage gaps
Where our probes actually are
Voidly operates 16 probe nodes globally — but they're mostly in non-censoring countries. This page is the honest disclosure of where we DO have inside-country presence and where we DON'T, vs where censorship actually happens.
One more thing the count itself cannot tell you. The API splits the fleet into active_nodes (core) and everything else as community, but part of that community bucket is Voidly-run Fly.io machines that register through the same public path a volunteer's node does — nothing in the response separates them. Read the core figure as a floor on what Voidly operates and the community figure as a ceiling on genuine third-party participation. We publish what the endpoint can prove, not an internal tally it cannot corroborate.
Updated every 30 minutes from /v1/probe/network · CC BY 4.0
The honest version
Our probe nodes are mostly hosted in places where it's easy to host probes — the US, UK, Germany, Japan, Singapore, India, Brazil, South Africa. Those vantage points are great for “is this domain reachable from outside” tests (which is how CensoredPlanet works), but they don't tell you what someone inside Iran or China actually sees.
For inside-country measurements, the Voidly Atlas relies on OONI (volunteer-run probes inside countries) and CensoredPlanet Satellite (remote DNS measurement technique that works for ~50 countries). The data we publish IS honest — but the “our probes verify censorship” claim is partially-true at best.
The fix is community probe operators inside censored regions. See /probes for the install path. Anyone with a Raspberry Pi, a VPS, or a spare laptop in one of the gap countries below can directly improve our ability to verify censorship from the inside.
Wanted: probe operators in these countries
High-censorship countries where we have no inside-country probe presence. Hosting a probe here would directly improve our coverage of the events that matter most.
2,401 incidents in the corpus come from these gap countries — currently verified via OONI volunteer probes + CensoredPlanet remote measurement.
Coverage of top-incident countries
The 10 countries with the most incidents in the Voidly corpus, sorted by incident count. The right column shows whether we have inside-country probe presence in each.
| Country | Incidents | Probe |
|---|---|---|
| Russia RU | 310 | external only |
| Venezuela VE | 309 | external only |
| Iran IR | 299 | external only |
| Egypt EG | 249 | external only |
| Pakistan PK | 244 | external only |
| Azerbaijan AZ | 241 | external only |
| Myanmar MM | 241 | external only |
| Belarus BY | 170 | external only |
| AE AE | 169 | external only |
| Bangladesh BD | 169 | external only |
Scroll to see every column.
Where we DO have probes (16 active)
Host a probe (Pi, VPS, spare laptop)
The probe is a single-purpose process — rotates through the Citizen Lab test list and reports to api.voidly.ai/v1/probe/results over HMAC. Runs as a dedicated non-root user, uses pinned TLS, never exposes your location to other probe operators. Check current probe coverage before citing country or domain counts.
Related
- /probes — Voidly probe network status + install instructions
- /methodology — full data pipeline including external-source fallbacks
- /atlas/methods — how blocking is detected (DNS, SNI, TCP RST, etc.)