Top Indicators of Internet Censorship

Internet Censorship Indicators

The internet has quickly become essential to the lives of many in the developed world. It is crucial for communication, access to information, commerce, and more. Autocracies have an incentive to censor the internet to prevent the free flow of information and communication between their citizens. As the technology to access the internet evolves, so does the techniques employed by autocrats for blocking the internet. Fortunately, the methods for detecting this censorship are evolving just as quickly. Here we provide a list of the top sources for measuring whether or not the internet in countries is being censored.

The list is divided into two categories, datasets which provide general indicators comparable across countries on an annual or semi annual basis, and trackers that look at specific instances of censorship, but do not have a clean and aggregated annual indicator.

Indicators:

1. Access Now

The top indicator for measuring internet shutdowns is the #KeepItOn STOP dataset. This dataset complies, validates and reports out remotely sensed internet shutdowns as well as the blocking and throttling of social media. It has been found to be less prone to false positives than other datasets on this list, at least in part due to its remote sensing of shutdowns. The data include a great deal of specific information about individual shutdowns, including when exactly they start and end, the geographic scope, the services impacted, the government entity responsible, and links to the underlying trackers that captured the shutdown.

The data themselves are a bit messy, but they have improved the quality of the reporting in the past few years. If you have questions about the dataset, you should reach out to Zach Rossen.

• Update Frequency: Annually, usually in February

• Country Coverage: Global. The only countries listed are those where shutdowns were sensed

• Data Gathering Methodology: Remote Sensing

• Concepts Covered: Internet shutdowns, social media shutdowns, internet throttling.

• Further Disaggregation: Data include information on the length of the shutdown, the geographic area, the services and users targeted, and more.

• Ranking and Aggregation: No Ranking or Aggregation in the dataset, some are available in the annual report.

2. Digital Society Project (V-Dem)

Where Access Now provides a great deal of details on a specific kind of internet censorship (shutdowns), the digital society project provides a much broader range of information with less depth on each question. This dataset uses the Varieties of Democracy platform to ask these questions, which surveys thousands of Experts around the world, then aggregates their responses into individual indicators and indices. In addition to covering questions of internet shutdowns, this dataset captures issues of censorship, disinformation, and the political polarization of social media in a country. While these data provide a good picture of the general environment, some have argued that there is evidence that they may not actually be picking up on the specifics differences between shutdowns and content censorship and more the general censorship level in a country.

Because of their normalization process there are several different versions of these data, the model outputs values ranging from -5 to 5 (no suffix), but these are also transformed into a version on the original scale for the questions: generally 0 to 4 (_osp suffix), and an ordinal version which takes the most likely whole number value 0, 1, 2, 3, or 4 and reports that (_ord suffix). For more information on this, take a look at their codebook. Micheal Coppedge is a good point of contact for these indicators.

• Update Frequency: Annually, usually in March

• Country Coverage: 173 Countries (74 LIC/LMIC) in 2025

• Data Gathering Methodology: Expert Survey

• Concepts Covered: Censorship, shutdowns, misinformation, political polarization.

• Further Disaggregation: Data claim to be able to differentiate between foreign and domestic misinformation, but it can be challenging to differentiate these without a full network analysis.

• Ranking and Aggregation: Several different estimates available, including the original model estimate, the original scale and whole numbers on an ordinal scale as described above.

3. Key Internet Controls (Freedom House)

Freedom House has historically produced a Freedom on the Net report, which captures obstacles of internet access, limits on specific content online, and violations of the rights of internet users. The report uses a similar methodology to Freedom House's flagship Freedom in the World report, which asks experts to rank counties on a series of questions, then attempts to normalize those answers across countries. The report also includes a series of binary questions in the "Key Internet Controls" component on whether certain content was blocked in the last year, whether new laws further censoring or surveilling internet users were passed, and if internet commenters have been physically assaulted or arrested.

Some challenges with this report are that they only cover 72 countries, and that the future of this report is very uncertain given that it was funded by the US State Department which recently cut off funding.

• Update Frequency: Annually, usually in October

• Country Coverage: 72 Countries

• Data Gathering Methodology: Expert Analysis

• Concepts Covered: Censorship, access, and user rights

• Further Disaggregation: Broken down into individuals questions and categories. Key Internet controls dataset captures individual binary questions on specific questions about censorship.

• Ranking and Aggregation: Countries are grouped into Free, Partially Free, and Not Free.

Trackers:

In addition to the indicators above, there are several tracking organizations which report on individual instances of internet shutdowns, censorship, and disruption, but do not aggregate their data into a single indicator.

Netblocks

Netblocks is an organization that uses automated systems to monitor internet outages around the world. Not all of the outages that they capture are censorship, and they do not capture all outages, but the do have good reporting on outages when they detect them including both the technical details and an explanation of what is going on for the layman.

Censored Planet

Censored Planet has a massive, real-time database which captures disruptions and censorship of the internet around the world automatically. Unlike other systems listed here which either do not validate the initial identification of censorship or which use in-country volunteers to validate censorship, Censored Planet uses sophisticated system to validate the identification of instances of censorship automatically allowing for them to produce massive datasets on censorship around the world in real time. A challenge of these data is that they are disggregated and quite technical, so are less of a final product and more of an intermediate source of raw data for someone looking to construct an indicator.

OONI

OONI allows anyone to download their software to run int he background and help detect internet censorship and outages. These notifications are then collated into a dataset and reports on specific instances of censorship in countries. This source uses different methods than Censored Planet, relying on in-country users to validate instances of censorship, though they publish a similarly massive dataset. They also publish reports similar to Netblocks about specific instances of censorship that they detect, though these are less frequent than the ones published by Netblocks.