Internet "censorship" is often associated with repressive governments filtering the traffic of their citizens, but it goes well beyond that. Internet service providers sometimes filter—or alter—the traffic that they carry, companies restrict employees based on keywords and URLs, courts naïvely order certain URLs to be blocked, and so on. But it is difficult for any particular internet user to know just what it is they can't get at. That problem is what the Tor Open Observatory of Network Interference (OONI) project is hoping to help solve.

The overall goal for the OONI project is "to collect data which shows an accurate representation of network interference on the Filternet we call the internet", according to the web site. One obvious, though time consuming, way to do that is to gather information from multiple different "locations" on the internet, and that is what OONI has set out to do. Of course, the OONI project itself can only reach out so far, so the intent is to enlist other participants—essentially "crowdsourcing" the data collection.

There are other internet censorship tracking projects—Google's Transparency Report and Herdict for example—but the OONI project's README notes that other efforts either use a closed methodology or closed software. As befits a Tor project, though, OONI is fully open source. No top-level LICENSE file for OONI is present at the moment, but one would guess it will be similar to Tor's permissive license.

The core piece (ooni-probe) is written as a framework in Python, with an eye toward contributions of additional tests (called "plugoos") and reports. "Tests" are meant to detect censorship events by comparing the results obtained locally with some kind of experimental control. That control could be obtained via the Tor network, for example, or via some other means. The tests can use various kinds of "assets", which might include lists of URLs, IP addresses and ports, or keywords, as their input. Current tests include checking that Tor bridges are functioning, determining whether HTTP "Host" field filtering is occurring, checking for DNS tampering, doing address and port scans, detecting Squid proxies, and so on.

While there are plenty of tests that could be added, seemingly the area needing the most attention right now is the "reports". Currently, test failures are essentially just written to an unstructured text log file, which can be stored locally or uploaded to a server. Tools to interpret the data and to provide higher-level visualizations of the types and locations of internet censorship are planned.

While the OONI code is under heavy development, the project can already claim some successes. ooni-probe was used to detect eight blocked web sites for internet users in Bethlehem, West Bank. The probe scanned more than one million sites and found that users are blocked from eight news sites "whose reporting is critical of [Palestinian Authority] President Mahmoud Abbas".

In addition, ooni-probe found that T-Mobile USA's Web Guard "feature" blocks access to much more than the advertised categories. In particular, sites for Tor, the Internet Archive WaybackMachine, Chinese sports news, French economics and financial news, a Japanese URL shortener, and many others, were blocked though they didn't fall into any of the listed categories: "Alcohol, Mature Content, Violence, Drugs, Pornography, Weapons, Gambling, Suicide, Guns, Hate, Tobacco, Ammunition".

OONI is just getting started, but it is clearly a welcome addition to the internet landscape. In order for John Gilmore's famous quote ("The Net interprets censorship as damage and routes around it"—which seems to be an informal slogan for OONI) to be true, the internet, or really its users and operators, must be aware of where that censorship is occurring and how it is being applied. With tools like OONI (and the others, though it's unclear why they aren't more transparent), routing around that censorship will be easier. The free flow of information on the internet depends on being able to do so.