Measuring the Internet to fight censorship / Arturo Filastro, OONI

[applause]

Thank you. And yeah, we'll have also some numbers, why not? Um so yeah, I'm I'm here I guess today to tell you a bit about the work um that we do at OONI, which is um this uh non-profit organization based out of Italy uh that measures the internet. Um the goal that we have set uh since uh we've started working on this, which now is um 14 years. It feels quite uh strange to even think about that.

Uh that uh we've been at this for yeah, since 2012. Is that our idea is basically we would like to empower people to be able to carry out these decentralized efforts at understanding how network interference happens around the internet. Because we might be thinking that, you know, the internet is the same wherever you are around the world, but actually unfortunately, it's changes and it's different and there are different ways in which uh access to information is being restricted. Um and the way in that we see as a um a necessary precursor to be able to actually fight against this and go in a direction of a more free and open internet is that of first uh understanding the problem of collecting evidence, of collecting data on what is actually happening so that we can then um support and empower groups, wherever they might be, to be able to take the action that they consider to be most appropriate and most necessary to change things uh in the place where they are um sort of located. And how we do this?

So, we have a set of applications that people will install on their phone or on their computer or on their server or on their Raspberry Pi. Um It's called OoniProbe. You can try it out. It is available in all major markets. And it carries out a series of network experiments or net tests, which measure different properties about your internet.

So, they check for example whether or not certain websites are blocked. They try to simulate the activity that a particular mobile application will carry out like a messaging app. They measure whether or not certain censorship circumvention tools like VPNs function or do not function. Um We measure also the performance of your network and then we have sort of let's say some more experimental nerdy tests that look into the availability of new protocols that are being actively researched and studied that um people that are working on deploying them might have certain questions on how feasible it is to use them to make the internet better. Um we then through these tests collect all of the data that is that is running and make it available as a entirely open data set.

And we are the the world's largest open data set on internet censorship that accounts for what is it now? Over 3.4 billion measurements from 245 countries and territories covering more than 30,000 networks. And it's all available for anybody to query through our web interface called Ooni Explorer where you can go, you can put in some um some questions, let's say, of of what you are interested in knowing. Let's say, where in the world Twitter is blocked, and then you get some nice charts that tell you where it is blocked, some nice numbers that uh you can um either look there or import it into your own tooling, uh use it as part of research, uh which it is used as part of a lot of academic research, as well as uh journalistic investigations, uh and more about what I will say uh in a minute.

Um people use it um themselves because it's an open data set, so anybody can access it and use it as part of their own work. Uh but I would say we are also um the number one user of our own data as part of uh research that we carry out ourselves uh in collaboration with um I think at this point uh almost 60 or 50 or 60 different uh organizations all over the world that we we work with um through these partnerships on studying certain phenomenon that is happening in their country. And we sort of provide the um the technical interpretation of the data to support the work that they are interested in doing uh inside of their country or in their region or in their community. Um so, we sort of um we're not an advocacy organization, we're not a policy organization, we um support other groups that do these kinds of interventions with the necessary tools and technical interpretation to carry out the action or intervention that makes most sense uh to their context. And we've published, yeah, more than 110 reports uh on internet censorship covering a variety of different topics from uh blocking all the sites around the world with Out Right International, uh, covered the the blocks in Iran.

Um, during the the coup in in Myanmar, we were looking into the investigations there. And and and I guess a trend is that you see, uh, in certain instances of, um, social or political tension, an intensification of blocks, uh, whether it is, uh, elections, protests. Um, it is becoming more and more the case that restrictions affecting the internet, uh, become a tool to be able to, uh, control dissent and, uh, steer public opinion in a certain direction. Uh, and also we definitely see it being used as a tool to further repress and discriminate, uh, marginalized communities and groups all over the world. Um, and so we document these kinds of things.

Um, I wanted to put here another, um, point which I think maybe frames this also in the context of something that is maybe also closer to where we are now, uh, in that I think we tend to think that censorship is sort of just this, um, you know, the sort of over there problem, you know, of of places where, uh, maybe there is not such a strong rule of law. But as a matter of fact, internet censorship is something that that happens and is a problem also, um, right here. Um, so, you know, maybe you have heard, but as part of the package of sanctions that was, um, implemented following the, uh, Russian invasion of Ukraine, um, the European Union passed, um, a decree requiring, uh, countries to restrict access to, uh, Russia Today and Sputnik News, the two state-run, uh, media organizations, uh, which led to a variety of different countries implemented implementing more or less successfully and consistently mostly unsuccessfully and inconsistently various restrictions on these two sites. We have other cases as well like recently it's quite popular in Europe to you know because football is a big thing. People love to watch football.

There is a lot of money in football. And a lot of money means a lot of interest and as a result for example Spain Italy and France have started to roll out these changes to a regulation affecting how filtering gets deployed giving sort of unprecedented authority to in the one case in in in Italy the regulator that doesn't have to go through a judicial process in order to request that internet service providers implement restrictions and in theory the idea there is that the regulator should provide some level of oversight. However as was the case a few years ago when Google Drive got blocked the oversight does not seem to be sufficient. In the case of Spain this was basically entirely handed over to a private company. Which implements restrictions in a way that leads to a lot of over blocking because it is being performed in a in in a certain way by by IPs that are shared by multiple different sites.

Basically when there are La Liga football matches the internet doesn't work in Spain. And people are very upset. And so this I guess is to highlight the importance also here of monitoring and understanding what kind of network interference we have and you know, the potential impact to you know, our freedom of access to information or even just you know, work. So um another thing I wanted to point it out point out that I think is is is also maybe um overlooked sometimes is how you know, the the internet is kind of this complex system of many different entities that interact with each other in in different ways that we um we would like to understand but in reality are not fully able to appreciate uh the for example second order effects that certain actions um um will result in and so this is an example of uh in the case of um when in Europe uh these restrictions were implemented affecting uh Russia Today and Sputnik News on behalf of uh European countries which weren't really effective like we in the paper that I that I cited before that we we worked on we see that as a matter of fact most countries are not really restricting access that effectively uh on these sites. But it did generate a retaliatory action on behalf of the Russian authorities which basically ended up blocking a variety of different uh state-run media inside of Europe.

And so again, I'm not saying that it it's necessarily should not have been done but I think it's questionable how much the net positive is of this action in that now in Russia a bunch of sites that were previously accessible are now no longer accessible. Um well, I also wanted to to point out uh like a report that we just published the other day. So, um we worked in uh with the in collaboration with this um this organization called Woman on Web, uh which is a reproductive rights organization that provides uh telemed consultation to uh women that want to practice um safe abortion uh in places where it is uh either legal or uh there are certain obstacles to being able to carry that out. Uh and we documented, for example, um instances in which their site was getting restricted in uh seven countries. I guess Iran, Turkey, the Philippines, South Korea, Spain, Kuwait, and temporarily also in Argentina, it seems.

Um it's worth pointing out that we had previously worked with them uh to uh investigate the case of Spain because they had received some reports that they were getting blocked in Spain. We were able to um confirm, as a matter of fact, that that was the case. Uh and they used our data to submit a petition to the Supreme Court, uh which they won the case. Uh and it was supposed to get unblocked. Uh however, it was only um partially unblocked, and then it was deemed technically unfeasible to only partially unblock it.

So, the net result is that it is still blocked in Spain, uh as we can see here. And um yeah, they are going to be pursuing now uh additional legal measures to um attempt to get it uh unblocked. Um Another point I wanted to make is that so, I think there's this um this idea that somehow, um you know, uh internet measurement is somehow something that you can um you know, you can just go out there and measure the internet with like a ruler and a and a scale or something. But but the reality is that it's um there's there's a lot of of conditions that are not necessarily um going to lead to a metric that is fully precise. Um and the sensors know this and are actually getting better at trying to make the censorship more difficult to measure and more difficult to be perceived from the end user.

Because it's one thing if you're trying to access a site and there's a page telling you you can't access this website because of this and that reason. It's another side of It's another case if it's just degraded to a point where you get frustrated and you're not able to access it and maybe you even blame it on the website itself. Uh and so what we're trying to do with the work that we do is that we're observing um sort of something that happens on the network to prove that there was intent behind intentionally restricting access to that specific target. But we don't have access to the sort of original information that would allow us to prove intent. Like unless we somehow intercept the communication of like the person in the government that calls up the ISP and is like, "Yo, block Facebook."

Um we can just observe what we see on the network. And And what we see on the network, we we try to collect enough evidence to um disprove all other um causes that might just be network weather. Uh and in the in the case of throttling, this is an example that I think is sort of an extreme case of this is where the performance towards in accessing a specific target, in this case Twitter in Russia, is intentionally degraded to a point where um it just works very slowly. And so, here what we can see, I mean, I'm going to go too much into detail, but here we're basically looking at the speed at which um data is read when um performing a certain network operation targeting Twitter. And we can see that you have kind of a cluster in the bottom here.

And then, which is like low, so good. And then at a certain point it starts clustering up. So, this is a signal that somehow it started like Twitter started to be hella slow. And so, um this shows that it's something that could not just have been happening by chance. Why I say this?

I say this because um what we've seen in recent years is actually that internet measurement data is starting to be used as evidence. It it's uh I brought up the case of Spain before. Uh we also participated last year in a a petition to the Supreme to the High Court of Kenya uh related to uh a case uh of the blocking of Telegram during the national exams, basically in 2023 and 2024 uh during the national exams the whole Telegram stopped working in the whole country. Initially, the government put out a statement saying well, no, actually Telegram is not working anywhere in the world. Uh but then we see from our data quite clearly that it's not actually it's working fine everywhere else.

It's It's just in Kenya that it's not working properly. Um and so, in order to take this data and make it be something, you know, I think the the previous model was somehow that um a lot of this data was primarily being used as part of um let's say advocacy policy making and things that are more informative to a public debate right so it's more something that you know is doesn't necessarily have to be used as a method to enforce accountability and and something that should eventually become part of a verifiable public record and maybe this is something that resonates with with some people working in the space of you know being able to actually independently verify something and being able to you know say well actually I can look at this open data set I can look at how it was collected I can look at how they they analyzed it and I can come to the same conclusions hopefully and so I can believe it and so this I think is like where where we're moving towards and the fact that we're moving in this direction makes I think the need of having standards and rigor ever more important and that's why I think for example you know some principles that we've always followed at Ooni is you know the the necessity of having open data methodologies that are documented openly that are reproducible and that everything is available as free and open source software because ultimately evidence is the strongest when it can be scrutinized publicly and that anybody can look at it and come to the same conclusions um so I just put this pitch out there as well so we are part of this fundraising campaign too which uses quadratic funding it's a thing that web three people like so I think I think it's it's good. uh We there's many other projects. We are one of them. And I don't know.

We

[clears throat]

If you have some spare crypto, you can throw it our way and then it does this quadratic formula scaling thing and then we can do more Uni. And yeah. And that'll be great. So, I'll leave it with this. We shall measure all the things and if you want to contact me, you can you can find it there.

Thank you.

[applause]