 In this video, we'll talk to Jorge Ruiz Reyes, one of the authors of What Can Go Right. We'll be discussing how civil society groups can use machine learning to help locate clandestine grave sites. How did the collaboration between your group and HR DAG come into being? Yes, our collaboration started in 2017 when we invited Dr. Patrick Bolt to a talk at Universidad Iberoamericana. He talked to students and to professors about the work that the Human Rights Data Analysis Group has been conducting throughout the years. And after that, we had meetings to see what kind of projects we were involved and how we could create a possible collaboration. And that's how the project started. We had meetings and we saw the different datasets that we had and we kind of brainstormed the possibilities that we could do with this data. And then Patrick told us, well, I think we can apply some machine learning to this project. And that's how it started. Then after that meeting, Patrick stayed for four or three days more in Mexico. And we were just working on the model and that's how we got our first results where it was like a very quick and healthy relationship among the organizations. And then how did you start to work with Megan as well? Well, I knew Megan thanks to Patrick because she was aware of what we were doing with Patrick in the project from Mexico. And then we started also talking with her. And that's how I met Megan. I didn't actually knew her physically until we met at the rights conference. It was mostly through emails and some quick talks, but that's how we finally were able to meet each other. How did you find the working process when it was online just through emails? Was it a smooth process or is this kind of work easier to do in person? I think this kind of work, it can be adapted fairly easy to work online because there are some tools that help us to do this. For example, we have repositories where we can work on different locations when we don't need to be in the same physical space. So that's how we have been able to do it. And luckily, the most important aspects of the project, we were able to develop it in 2017 while we met in person. But after that most, the majority of the collaboration has been online after 2017 because HRDAC is in San Francisco and University of Iber-Americana and DataCivica we are at in Mexico City. So yes, we had to adapt to online work since 2017 and it is sometimes Patrick, for example, came to visit us like once a year in 2018 and 2019 to see new steps or how we were advancing and to tune some things of the project. But yes, that's how we have been working actually online and just a few visits from Patrick each year until the pandemic. But we have been working online since the project started actually. Can you tell us more about how students contributed to the project? How do enforced disappearances and involuntary disappearances affect the younger generation? Well, the Human Rights Center of University of Iber-Americana, it's a research and advocacy center that has main researchers, but it gets a lot of support from students, whether they are undergraduate or in a postgraduate. They can work at the Human Rights Center as voluntary or also, for example, in Mexico we have social services, it's called, where they have to fill some credits so they can be able to fulfill their undergraduate. So that's how we have been receiving the work from students for this project, the disappearances project and the clandestine graves project we have had over 20 students since 2016. They have helped us to get the data that we are using, well, part of the data that we have been using for this project, for example, reading the press notes where clandestine graves are reported either at the national level or at the local level. And they help us to read the notes, to classify the notes, and then when they give us back the notes, we then put it in the database and use it for the statistical analysis set and the predictions. And I think disappearances, well, the problem in Mexico is that for disappearances or involuntary disappearances, it is a widespread phenomenon. It has been classified like that, for example, for by international organisms such as the U.N. And right now we have more than 93,000 missing persons in the country. So it is a problem that affects the majority of the states and persons from different ages. So I think disappearances have now become a major concern for a large part of Mexican society. And I think it affects us because, well, it is an ongoing problem, right? We are not documenting disappearances from the past. We are documenting disappearances from the past while disappearances are still happening, right? So I think that's how younger persons are trying to work on the issue and try to generate solutions to it. What was the outcome of using machine learning in the advocacy strategy, different from traditional or non-technological based advocacy? I think, for example, in the Human Rights Center and in the Pacifica, we were using more like classical statistical approaches, for example, descriptive statistics of how many graves have been found in Mexico. And that has worked because we didn't even have, for example, official diagnostics of the phenomena. But we know that we were not documenting the whole universe of clandestine graves that have been found in Mexico. And we have been just able to document a small fraction of the graves due to different factors that affect the social production of information of the phenomena. So that's how machine learning has helped us. Machine learning helps us to identify places where either official sources or non-official sources are not able to document graves because they're not able to document economic or violence related or political factors that they don't allow to document graves in certain places. So that's how machine learning helps us or is helping us. It helps us to identify new possible places where we can find more graves or where we should be conducting more search strategies. And yes, it definitely has helped us in advocacy because we have been able to provide, for example, policy briefs or policy reports to authorities, but also to groups of families with missing persons to assist them also in their advocacy processes. So it is a way like technologies can support the work of these groups. So what are you working on now? And what are the next steps? Well, we're still documenting clandestine graves. We're trying to develop different models that can support the search strategies of authorities and groups of families with missing persons. Right now, we are focusing on more specific regions of the country because this first machine learning model that we have developed with HRDAG and DataCivica, it is a model that predicts municipalities in the whole country, right? But now we are developing specific models for, for example, certain states in the northern Mexico, where we are identifying regions within those municipalities, right? So we are making our predictions, the area, the geographical areas, like we are narrowing the area because we are now using, for example, also satellite images and other geospatial approximations. So we are now identifying like more specific regions within the states or municipalities where we believe we can find clandestine graves. So we are, we are, we are, we now have new results for this and we are also showing these results now to authorities and to groups of families. So that's what we are doing and we're still documenting these appearances and trying also to support the documentation processes of groups of families with missing persons because, well, they have a lot of information, but we need to find ways to, to help them to structure this information to organize it and also to keep it safe, right? So that's what we are also doing with the groups of families. Do you have any advice for civil society groups that are facing similar challenges and are planning to use similar strategies? And is there anything that you wished you knew before you started the project? Well, I think my advice is to try to find collaborations with different organizations. It's, it's better to work in, in group. I think since 2016 I don't remember any project where it was only the Human Rights Center or only DataCivica. We have been, all of our projects are projects with two or three organizations because we have different strengths and weaknesses. But by working together, that's definitely how we have been able to, to produce relevant results in, in a small fraction of time. And I think also don't, don't be like shy or don't be afraid to reaching like these organizations that sometimes they have like big names, but they are actually very nice people and very kind people that they want to help, right? They, that's how we met HRDAC and DataCivica. So I think that's, that would be my advice and also to, to document as much as you can with the tools that you have, you know, it's, it's also, it doesn't matter if you have like the most advanced technologies, even if you only have some notebooks written by hand or a common spreadsheet, there will be persons that will be able to help you to make sense of that data. You just have to document and document. It doesn't matter how you document. And some things that I wish that I would knew, I don't know. Well, maybe some of the, of the statistical approaches that now we have been able to learn, but I think it's also part of the process of, of working together. So I think it's just, yeah, just have like good definitions of what you're trying to do, good methodologies and document as much as you can. And by working with a team, you're going to get the solutions. So I think that's, that would be my advice.