What OONI's Internet Censorship Data Can (and Can't) Tell You

OONI data can show you whether specific websites and apps are blocked, throttled, or otherwise interfered with on the networks where volunteers ran tests — and it can show this across more than 200 countries going back to 2012. What it cannot do is give you a single, definitive verdict on "how censored" a country is. Each measurement reflects one test on one network at one moment, so the data is strongest when you read many measurements together and weakest when you treat a single result as proof.

What OONI measures

OONI (Open Observatory of Network Interference) is a global community project that measures internet censorship. Its two main tools serve different purposes:

• OONI Probe — the client you install to run tests on your own network. It tests the blocking of websites and apps and measures network speed and performance.
• OONI Explorer — the public interface for browsing the resulting dataset, described by OONI as the world's largest open dataset on internet censorship, with millions of measurements collected from more than 200 countries since 2012.

The kinds of interference the data can surface include:

The IP-based blocking case is worth understanding because it produces misleading-looking results. OONI's own research on LALIGA football streaming in Spain documents how blocking aimed at one target can cause collateral damage to unrelated services that happen to share infrastructure. If you see a cluster of unrelated failures, shared hosting or a shared IP range is a plausible explanation before you conclude that each site was individually targeted.

How to read a result responsibly

Treat OONI data as evidence about networks, not as a country-level score.

1. Check how many measurements support the claim. One blocked result on one network is an anecdote. A consistent pattern across many networks, ISPs, and time periods is a finding.
2. Check where the measurements came from. OONI Probe runs on volunteers' devices, so coverage is uneven. A country with few volunteers will have thin data, and that thinness is not itself evidence of anything.
3. Check the local network conditions. A failure can come from a misconfigured router, a captive portal, a local outage, or a filtering middlebox — not necessarily a national censorship system.
4. Look for the mechanism, not just the outcome. Distinguishing a DNS block from an IP block from throttling changes what you can conclude about who is interfering and how.
5. Use OONI's own explanatory material. The project publishes its testing methodologies, a FAQ, and a glossary specifically so that people interpreting the data don't have to guess at definitions.

What the data is good for

• Tracking change over time. Because measurements accumulate continuously, you can compare a service's reachability before, during, and after a politically significant event.
• Documenting specific cases. OONI's research blog covers concrete findings — for example, work on the blocking of reproductive health websites such as Women on Web and Women Help Women, on censorship of abortion-related resources globally, and on social media throttling in Türkiye during protests.
• Supporting journalism and advocacy with citable measurements. The dataset is open, so a claim can be traced back to the underlying tests.
• Comparing networks within a country. Differences between ISPs in the same country are often more informative than a national average.

What the data cannot tell you

• It cannot tell you intent. A measurement shows interference occurred; it does not identify who caused it or why.
• It cannot tell you the full extent of censorship. Tests only cover the specific sites and apps in the test list. Censorship of something not tested will not appear.
• It cannot give you uniform geographic coverage. Volunteer distribution is uneven, so absence of measurements is not absence of censorship.
• It cannot isolate a single cause from a single result. Local network problems and censorship can produce similar-looking failures.

Practical rule of thumb

If you want to cite OONI data, state three things alongside the finding: how many measurements support it, which networks and locations they came from, and over what time period. If you can't state those, you have a lead worth investigating rather than a conclusion worth publishing.