April 10 2025 GM

From TCU Wiki

Network Measurements - How to Gather and Use Evidence of an Internet Shutdown Happening in Your Country?

Join us on 10 April with Elizaveta who coordinates internet censorship measurement and research at the Open Observatory of Network Interference (OONI) to learn about:

  • Open datasets that allow to investigate internet shutdowns (IODA, Cloudflare Radar, MLab, OONI etc.)
  • Tools to coordinate rapid response to social media and other websites blockings (OONI Probe, OONI Run)
  • Communities to be part of to get support to hold local ISPs and governments accountable for shutdowns

Elizaveta, Community and Research Coordinator, works with the OONI where she coordinates internet censorship measurement and research in collaboration with international partners. Previously, she worked with Roskomsvoboda, where she researched Russian journalists’ and activists’ digital security and censorship in Russia.

What is Glitter Meetup?

Glitter Meetup is the weekly town hall of the digital rights and Internet Freedom community at the IF Square on the TCU Mattermost, at 9am EDT / 2pm UTC. It is a text-based chat where digital rights defenders can share regional and project updates, expertise, ask questions, and connect with others from all over the world! Do you need an invite? Learn how to get one here.

Notes

With a quick Community Update roundtable, participants share that OONI had a major release of new OONI Explorer pages last week (https://ooni.org/post/2025-ooni-explorer-thematic-censorship-pages/), that the DRAPAC25 in Kuala Lumpur, Malaysia is taking place (https://engagemedia.org/2025/announcing-drapac25-kuala-lumpur/), and the launch of DRAPAC VPN Bot (https://engagemedia.org/2025/drapacvpnbot/)

Can you introduce OONI for those who may not be familiar with the organisation? What is it about, and what tools do you build? And most importantly what does the octopus stand for?
  • OONI-the-organisation (not-the-octopus) is a small team building open source tools to empower people to monitor their local networks for censorship. The project was created back in 2012 as part of the Tor Project (which I guess everyone in this chat knows very well), and since last year OONI is an independent non-profit organisation.
  • Over the past years, we developed multiple tools, I think many people in the chat have used them, and I saw a report which refers to our data was shared in the chat right before our meetup.
  • For those, who haven't used those, here are short explainers and links where you can read more:
    • OONI Probe (https://ooni.org/install/) -- an application which allows you to run tests on your local network to see if any website or application is blocked there. You can run it both on desktop and on mobile, but mobile likely will be more effective since the latest version of the app is always released first for the mobile users, and only after a few weeks or months for desktop.
    • OONI Explorer (https://explorer.ooni.org/) -- a public database with a dashboard that allows you build charts and investigate the collected data. This database stores all test results that we have collected since 2012. It's super easy to navigate, and you can also check out a page with short reports on the latest censorship events: https://explorer.ooni.org/findings.
    • We also have some smaller tools built for researchers, data analysts and for rapid response to emerging censorship, I won't get into the details here but you can find most links and user guides on this page: https://ooni.org/get-involved/
  • The legend about octopus is quite shady but I think the initial idea was that octopuses have decentralised nervous system, and OONI has a decentralised network of measurements.
Before we can into the details of how to use the tools for network measurement, can you tell us what the different forms of censorships and shutdowns out there are? Its seems like its not always a total outage. From your research, what trends have you observed regarding the most prevalent types of shutdowns?
  • Absolutely! There are many many types of censorship, and we document only a few of them. Generally, we define censorship as intentional control or suppression of what can be accessed, published, or viewed on the internet. This can include censorship on social media platforms, takedown requests, outages (blackouts), blockings, trolling, doxxing, even self-censorship.
  • Network measurement projects, such as OONI, focus on the network-level censorship: outages and blocks. We do not track social media moderation requests, as we do not track court orders and takedown requests.
  • Unfortunately OONI mostly focuses on blocking of websites and applications, and monitoring social media would require a whole different methodology and pack of tools. So far, I do not know any projects who would track social media-based censorship in an automated way, but if you know any organisations or teams doing that, please share.
  • A good example I can think of, even though it is not social media, is AppCensorship which documents platform-level censorship, namely the removal of apps from Google Play and AppStore:
  • Regarding the trends, I can answer only from our very limited domain, (a) there is more and more censorship and outages each year. KeepItOn is doing amazing job documenting these and reporting on this: https://www.accessnow.org/campaign/keepiton/. These can be measured through some NPL models but the main issue is the accessibility of the data -- not many social media are willing to share any content from their platforms.
What is causing the increase in outages?
  • Three main reasons for outages in 2024 (according to KeepItOn report https://www.accessnow.org/wp-content/uploads/2025/02/KeepItOn-2024-Internet-Shutdowns-Annual-Report.pdf) are military conflicts, protests and exams.
  • Some governments use local outages (or social media blocks) to prevent cheating during state exams. The most known is India, which has been doing it for a while now, but, for example, in Kenya, we haven't seen any intentional outages or social media blocks for a long time, and they suddenly blocked Telegram in 2023 during KSCE, and repeated this experience in 2024. It doesn't stop people from cheating, but it legitimises connectivity disruptions.
And what about the Rwanda outages, because protests don't happen there? Is it conflict related?
Since we are on the topic of data from the shut downs, how does one go about using open data sets to investigate the various types of censorship? What data sets are publicly available and where can one find them? What next when the censorship event is documented and confirmed? What happens, how do you report it? Are there advocacy efforts associated with shutdowns? If so, what are they? Especially when in some cases advocacy may feel hopeless
  • I will comment here only on the network-level censorship (blocks and outages), but I think part of these recommendations are applicable to any open datasets:
  1. Understand how this data was collected. In case of OONI, learn how the tests work, what exactly is being tested, what are the limitations of the tests.
  2. Analyse data in aggregate (build charts) to identify censorship patterns and rule out false positives (faulty data).
    • For OONI it means to check if the discovered patterns are consistent across time and networks -- some ISPs can be more slow to implement blockings, some can not implement it at all.
    • For some datasets, it would be also important to check geospacial data -- if there is an outage, is it limited to specific networks, or to a specific region or city?
  3. Compare your findings with other sources: reports from the ground, other public datasets, published research reports. For example, if you see in OONI data that Telegram blocked in country N, but people from this country tell you that they can access it -- maybe something is wrong with the collected data (but maybe not, and the data is collected from the network that is different from what people you are talking to are using at the moment)
So which resources help in verifying measurements with Geospatial data? When most are difficult to source, or have lots of politics around it?
  • I think it depends on the research question you are asking:
    • For the outages, in most cases, it would be reflected in the data sets because their measurements are collected automatically and do not put anyone at risk.
    • For the blocks, it is not very often that governments block Twitter, or Facebook messenger only in a specific city.
  • But when it is happening -- you can, for example, compare the list of networks where we see the block with the reported region of the block to see if these ISPs are present there, are these networks limited to this region etc.
  • There are some edge cases when, for example, you want to compare censorship in an occupied region and in the country-occupier. This requires more thorough methodology and "manual" approach to collecting data. We are building tools that make this process easier, but still researchers need to curate such studies very closely to be able to compare censorship in such regions. This is an example of such research: https://public.opentech.fund/documents/Measuring_Internet_Censorship_in_Disputed_Areas_Crimea_Russia_ICFP.pdf
How can folks in digital coordinate their work? Are there suggested tools and processes for this? Given that theres a lot of work in verifying the data etc? Does it influence any changes?
  • What we do now is that we coordinate with KeepItOn team, their community and our partners network. So usually whenever some outage or censorship is happening anywhere in the world, we hear reports almost immediately.
  • But then it's not always the case that we already have data from this country, and sometimes we need to find people on the ground to run tests. In these cases we usually publish a call, and reach out to our community members privately to gather some measurements, in most cases this works.
  • After we collect data, depending on the type and scale of censorship, we publish short or bigger reports
  • Both types of reports can be used for advocacy efforts, but, for example for the rapid response (when there is an ongoing developing censorship event), advocacy organisations usually refer to the findings since these are easier for us to publish and we do our best to release them as soon as we have enough data to confirm the block.
  • For longterm advocacy and litigation, usually organisations need more details, and thus we collaborate on publishing bigger research reports.
  • Depending on the scale of the blocking, advocacy can be led by local and/or international digital rights organisations. For example, when Pakistan blocked political parties' websites ahead of the elections last year, advocacy against this block was led mainly by the local digital rights organisations. While, in case of Senegal, where back in 2023 were blocked social media, and then outages happened, both local digital rights and human rights groups, and international organisations were involved in the advocacy.
  • In terms of effects, it is hard to say, but afaik KeepItOn managed to prevent some shutdowns last year with their Election Watch campaign: https://www.accessnow.org/campaign/2024-elections-and-internet-shutdowns-watch/
  • In some countries we have also seen successful litigation against shutdowns, including Indonesia, Togo and Nigeria
Does OONI Explorer work with low bandwidth?
  • OONI Probe does, you can run tests.
What are the limitations of the existing tool? How is OONi and other partners in the digital rigths space working to overcome these limitations if there are any?
  • There are a ton of limitations, and it can become a huge rabbit hole, but I will try to list the main ones:
    • Community-sourced data. It means that there will be always countries and regions where we lack outreach and data coverage.
    • Granularity of data. As we've discussed above, we can look only in per-country datasets.
    • We cannot test everything. So we test only lists of websites which we call test lists, these are also crowd-sourced, which means that sometimes they become outdated. So it may happen that we have measurements from country N but they are not relevant because the test list hasn't been updated for some years. https://ooni.org/get-involved/contribute-test-lists
  • For Community-sourced data and the fact that we cannot test everything, we are working with numerous organisations to expand our coverage and update test lists, so if you want to join our efforts, reach out or join our Slack: https://slack.ooni.org/. For the granularity of data, we are looking for a technical solution but so far we are not very sure if it becomes possible in the next few years.
What opportunities are available for people who are interested in conducting their own research but are not experts in network measurement?
  • Check out this course by Advocacy Assembly: https://advocacyassembly.org/en/courses/63
  • Then try to formulate a research question for your first research. Check out the datasets I shared earlier in the chat. Understand which ones are more relevant for your question and start reading documentation.
  • I can't emphasise more but DOCUMENTATION is there, it is public, sometimes even localised into many languages, so you can read it, learn and do your research