Fellows Spotlight: Johanna Wild, Investigative Journalist
An interview on risks, trends, and tools in OSINT digital research
When Johanna Wild entered the Berkman Klein Center at Harvard as a joint Nieman Foundation innovation fellow, I was intrigued. Wild works for the award-winning international open source (OS) investigative journalism collective Bellingcat. She is an expert on the creative deployment of technical approaches to support a more diverse cohort of public interest reporters and investigators, blending automated approaches with human-centered research methodology.
As someone who has supported expert networks in both disinformation and conflict documentation, I wanted Wild’s first-hand perspective on the benefits and risks of using novel open source intelligence (OSINT) tools to enable a broader, more transparent global knowledge base. We conducted this interview over email between Amsterdam and New York City.
Sam Hinds: Do you encounter specific types of people or professional backgrounds in the work of investigations and OSINT tool development?
Johanna Wild: The great thing about the field of open source research is that it consists of people from various backgrounds. Open source researchers spend a lot of time online. They find pieces of information on social media platforms, in online forums, and databases, and they compare features that they identify in user-generated online videos and photos with locations that can be seen on satellite imagery. This process, called geolocation, is used to verify online images. The nature of open source research allows everyone with an internet connection to do this type of work.
The open source researcher community is therefore a mix of people who do open source research as part of their job and volunteers who are passionate about contributing to important research in their free time. My surveys and user interviews with our Bellingcat community showed that our community consists of people working for human rights organizations, stay-at-home-parents who use their limited time to do something mentally challenging and useful, cybersecurity specialists, job seekers who want to learn new skills, lawyers, data scientists, people who are retired and many more. When I ask volunteers about their motivation, they often say that they want to contribute to research that reveals issues in the regions where they live, that they want to feel that in these times that are characterized by various conflicts around the world, and global challenges like climate change; they do not just passively sit around but actively contribute to something that creates new knowledge about those issues. Another motivation is to become part of a community with similar interests and to improve their open source research skills.
Of course there are also many journalists who are part of this community. Nowadays, more and more newsrooms are setting up teams focusing on open source research. However, journalists were more of the late adopters in this field. Most of them only discovered in the last few years how useful this type of research can be, especially if it is combined with traditional journalistic skills and methods. Newsrooms even started hiring skilled open source researchers who are completely self-taught and who have no journalism degree, which is something that is still rather unusual in the news industry.
Volunteers with a technical background contribute by building tools. These are often simple command line tools that are able to do one very specific task, for instance to scrape posts from a specific social media platform or to check whether an online account has been created on a platform using a specific phone number. Those tools do not usually turn into big commercial products; they are built by people from within the open source software community who focus on writing code that is publicly accessible to anyone. Several years ago, I clearly saw that the open source researcher and the open source software community are a very good match for each other, we just needed to bring them together. This is one of the things that we now do at Bellingcat. We organize hackathons, actively invite software developers into our volunteer community, and support them to build their own tools or to contribute to tools built by the Bellingcat team. This group of volunteers consists for example of people who have a full time job in a software company but want to do something meaningful in their free time, of job seekers who want to create their own portfolio of tools, or of academics who are already deep into a technical topic but would like to test its practical application.
Although the open source researcher and tech communities are very diverse in terms of their professional and personal backgrounds, they are currently still dominated by volunteers and professionals from Western countries, mainly from the US and Europe. The technical tool builder community is also, to date, still male dominated. This lack of representation raises serious questions in terms of who defines the future of our field and who has the power to research topics in regions all around the world. With people in many other regions still excluded from participating in this type of research, they mainly become the subject of Western researchers.
“While AI tools can be powerful, we should not expect to automate the whole open source research process. Doing open source research is a combination of specific research methods, the use of tools, a good dose of logical thinking and also creativity!”
SH: Have you seen novel trends emerge in the type of information researchers want today?
JW: I definitely observe that researchers, and especially journalists, have become more aware of how useful it is to be able to work with large datasets, to know how to scrape information from websites or to have the skills to build small tools that can speed up some of their research tasks.
Currently, everyone is of course interested in AI. Less experienced researchers are hoping for a tool that lets them input any picture or video and then spits out the exact location of where it was taken. While AI tools can be powerful, we should not expect to automate the whole open source research process. Doing open source research is a combination of specific research methods, the use of tools, a good dose of logical thinking and also creativity! Creativity is needed to spot topics that are worth getting investigated. When deciding where to look next in the vast amount of online information that is out there, creativity helps to connect multiple, often tiny, pieces of verified information which allow researchers to draw conclusions on a certain topic.
Another trend is the use of facial recognition tools. Open source researchers often find pictures that show individuals who have a connection to a certain research case but whose identity they don’t know. In the last few years, several easy to use facial recognition tools have emerged. Researchers can upload a picture of a person and the tool compares this picture with collections of photos from social media platforms. Sometimes, this can reveal the identity of a person, for instance by providing the person’s LinkedIn profile. It is obvious how useful this can be to identify individuals who were involved in serious crimes that require journalistic reporting.
However, facial recognition tools are a double-edge sword. We all know that they can provide wrong results. Two people might just look very similar and an uninvolved person might be misidentified as someone who is involved in illegal activities. It is therefore important that open source researchers do not use those tools as the only way of identifying someone. On top of that, the use of such tools raises various ethical questions ranging from the risk of stalking random people online, to questions about the data sources on which facial recognition tools rely. At Bellingcat, we reflected on how we can ensure a responsible use of facial recognition technologies and concluded that we will refrain from using these tools extensively, and never as a core element of an investigation. We also never used products from companies like Clearview AI. A good example of how we sometimes use a facial recognition tool as a starting point for further research can be found in our article on how “Cartel King Kinahan’s Google Reviews Expose Travel Partners”.
SH: Are there any overlooked tools that you like to highlight in your trainings?
JW: The best type of tool really depends on the research topic. Often a combination of several small tools can lead to the best results. For instance, our Name Variant Search Tool is basically an enhanced search engine for finding information about people. Open source researchers often start with a name and try to find out as much as possible about the person’s online presence. However, the name might be written differently on different sites. “Jane Doe” might also show up as “J. Doe” or “Doe, Jane”. The tool suggests different possible variations of a name and provides search results for all those variations. It is also possible to instruct the tool to search for a name specifically on Linkedin or Facebook.
Example: Name Variant Search results for different variants of the name “Jane Doe”
Our OpenStreetMap search tool, on the other hand, supports the geolocation process. A core task of many open source researchers is to find out where a photo or video that they found online has been taken. To do that, they try to identify specific features and compare those with what is visible on satellite imagery or maps. If researchers already have a rough idea in which region a photo might have been taken, they can input a list of features that are visible in the photo (for instance, a residential street, a school and a supermarket) into our tool, which will try to list all locations in a pre-defined region in which those features show up together. This can really help narrow down possible locations.
SH: What’s an example of an unusual story or insight one can find from OS tools?
JW: If open source researchers have no idea where a picture might have been taken but they know at which time it was captured and the photo shows objects that cast clearly visible shadows, they can try our ShadowFinder tool which is able to calculate at which locations around the world shadow lengths correspond with what can be seen in the photo at a specific point in time. This helps open source researchers concentrate their geolocation efforts to the areas suggested by the tool instead of searching across the whole world.
Another tool that has gained popularity within the open source researcher community is PeakVisor, a tool that was originally targeted at helping mountaineers orient themselves but which can also be used for geolocation tasks. For instance, we used it to research the location of the killing of Colombian journalist Abelardo Liz. This example in particular shows that a combination of research skills and the use of tools can go a long way.
SH: What frustrations or barriers do you see as a trainer, and how could the field democratize knowledge of command line tools?
JW: First of all: Teaching open source research is great. People who are interested in learning these methods come from so many different backgrounds which allows everyone to learn new things from each other, including the trainers! The topic is also quite accessible, meaning that everyone can start doing open source research with very simple methods, like using search engines in creative ways. Sometimes, this can lead to surprising results: For instance, just by googling, my colleague Foeke Postma revealed how US soldiers exposed nuclear weapons secrets via flashcard apps.
Of course not all methods are as simple, and one of the things people are struggling with the most are research tools. During my Nieman-Berkman Klein fellowship my research assistant Cooper-Morgan Bryant and I interviewed forty open source researchers about their use of tools. Their answers confirmed my previous findings on this topic: Open source researchers, who are either beginners or who are looking at a topic that is new to them, find it really difficult to figure out what tool they should use at what stage of the research process and how those tools work. With such a wide variety of online tools, some more useful and some easier to find than others, and many researchers feel overwhelmed by the task of finding their way through the landscape of available tools spread across various platforms.
In addition, the majority of open source researchers are not able to use command line tools since this requires a certain degree of technical skills. However, those are exactly the type of small tools that the open source software community is building most frequently. There is a clear divide between those who are building tools for open source researchers and the researcher community itself, for whom those tools often turn out not to be accessible.
“Open source researchers want complex tools that are easy to use and that are stable and well-developed but such tools need funders and teams who build them, and these conditions are not always easily met in the open source research and journalism space.”
On the other side, open source researchers are often not aware of the resources that are required to build mature tools that have an easy-to-use interface. It is getting easier now, but tool builders need to invest a lot more time to build such tools and this is difficult for people who do this task in their free time and without any funding. Open source researchers want complex tools that are easy to use, stable, and well-developed, but such tools need funders and teams who build them. These conditions are not always easily met in the open source research and journalism space. I hope that researchers will become a little bit more open to learn some basic technical skills, and even more importantly that they understand that not every tool that is useful for their research has to function like a fully built commercial tool.
At Bellingcat, we focus on bridging this gap between tool builders and open source researchers. We work with tech communities —often through programs like hackathons or fellowships — and make them aware of how important good user guides are, even for seemingly easy-to-use tools. On the other hand, we teach open source researchers how to use command line tools. We also launched a video series with the goal to help researchers make their first steps towards the more technical side of research tools.
SH: Tools take a lot of resources to build. Do any OSINT tools have a complicated provenance in terms of private sector origin or geopolitics?
JW: It is definitely problematic that researchers and journalists can be so dependent on tools provided by big tech companies. Meta’s social monitoring platform Crowdtangle will be shut down in August and this has caused a lot of discontent amongst journalists, in particular amongst those who are covering elections. For instance, many of the platforms and tools open source researchers use are provided by Google, like Google Search, Google Maps and Google Earth Pro. We are often at the mercy of the decisions that big tech companies take regarding use of their tools.
However, their tools are usually provided for free, which is not the case for other commercial tools. Open source researchers definitely need to look into the companies from which they are buying tools. One risk is that tool providers might be able to see what type of keywords people are typing in or on what topic someone is working on. Researchers and journalists need to be sure that their sensitive research topics are safe from being monitored by tool providers.
At Bellingcat we focus on mostly small open source tools, but those tools come with their own set of challenges. For instance, it is often not clear who is behind a tool that is offered on code-sharing platforms like Github, which can raise security-related questions.
“I would love to see universities getting more involved in building and maintaining tools for open source researchers and journalists…since both sides have the common goal of advancing research in the public interest”
This is why I really hope we can build a different tool ecosystem for open source researchers in the future. I would love to see universities getting more involved in building and maintaining tools for open source researchers and journalists. I think that such collaborations could work well since both sides have the common goal of advancing research in the public interest, and many of the tools that are used by open source researchers are equally useful for academic researchers. I also see opportunities to research security-related aspects of widely used tools together, as journalists and open source researchers could definitely use some help in assessing the risks that some of the tools they are using might be posing. If anyone who reads this would like to discuss these topics with me: Feel free to get in touch!
SH: Misinformation, disinformation, conspiratorial thinking: What are some of the uses and abuses of “research” you see in these contexts?
JW: What is most common — especially during conflicts and wars — is that people share either photos or videos from a different conflict or old imagery and make people believe that they are related to current events. In the context of the Israel-Gaza conflict since October 2023, this phenomenon has reached a new scale with countless examples circulating online. For instance, Bellingcat found videos that were shared with the claims that one showed rockets that were fired at Israel by Hamas and another that claimed to show recent Israeli strikes on Hamas; both turned out to be recycled videos that had been uploaded to YouTube several years prior.
“People who post such pictures might sometimes think they are doing ‘research’ and that they are sharing relevant information about an ongoing conflict, without realizing that they are actually sharing incorrect information.”
What is dangerous is that some of those posts go viral and are able to reach significant numbers of people who will never know that they fell for misinformation. People who post such pictures might sometimes think they are doing “research” and that they are sharing relevant information about an ongoing conflict, not realizing the information is incorrect. Others, however, will do it on purpose to evoke emotions either in favor or against one of the conflict parties. Users of online platforms cannot really do much to prevent being confronted with such posts. This is another reason it is essential that we all learn to question what we see online and to invest some time in learning basic verification skills.
What we have also been seeing is that supporters of conspiracy ideologies are increasingly using open source research tools and presenting the information as journalistic findings. For example, Qanon supporters in German-speaking countries started using flight-tracking sites to search for flights which they falsely believed were circling above “deep underground military bases” in which children were hidden and mistreated. This is problematic since people who are not aware of the methods and standards of open source research might not be able to differentiate between serious research and the distorted version of it.
SH: What are some of your favorite guidelines or best practices for journalists who aim to cover (and fact-check) broad conspiratorial thinking enabled by OS information?
JW: Looking at their business models can often be a very promising approach. More often than not, conspiracy-minded communities have business-savvy people amongst them who manage to benefit financially from those communities’ beliefs. When I was researching QAnon online communities in Germany, big platforms like Amazon and eBay had started implementing measures to ban QAnon products from their platforms. However, this seemed to have created new opportunities for QAnon influencers who were offering merchandise via their own small online shops. On top of that, customers in Germany were able to buy QAnon products from abroad, for instance from Chinese or British companies who offered products targeted specifically at German-speaking customers. It was interesting but also concerning to see how international today’s conspiracy merchandise markets are.
To research online shops, it is always worth researching what payment options those shops are using and to look into their potential use of cryptocurrencies. It is also important to take some time to learn the terminology a certain group is using. If you are looking into the far-right, for instance, it is crucial to learn how to interpret the symbols they use.
”Open source researchers are often portrayed as some type of ‘nerdy hero’ who spends time on his laptop to research ‘the bad guys’ and is celebrated once he succeeds. The idea of one hero figure who solves all the research challenges is really the exact opposite of how open source research works best…”
SH: How might international organizations build stronger support for women, femme-identified, and gender-nonconforming media and research professionals?
JW: In the field of open source research, there are definitely tendencies that I would like to see changed in the future. It is well established that women and gender-nonconforming people have traditionally had a much harder time to enter and succeed in the space of investigative journalism. Those issues are far from being overcome, but the journalism world has started to talk more openly about it, and the fact that academic researchers have published work on this topic has also been helpful.
My impression is that as open source researchers, we have not yet put enough effort into reflecting on what is happening in our own field. Maybe we thought that since it is relatively new, those issues would not appear as strongly. Unfortunately, however, they do, and it’s time to recognize this.
There are definitely many contributing factors, but one that has had a strong effect on me is that open source researchers are often portrayed as some type of “nerdy hero” who spends time on his laptop to research “the bad guys” and is celebrated once he succeeds. The idea of one lone wolf who solves all the research challenges on their own is really the exact opposite of how open source research works best, which is by nature collaborative and often requires the efforts of many to put together various small pieces of verified online sources for a specific research case. For those of us who don’t want, and are also not able to fit into this commonly portrayed male hero picture, this field might not necessarily feel like a good fit.
However, since more and more traditional newsrooms are setting up open source research units right now, I see more women entering the field and hopefully, this will also change how we publicly talk about open source research over time. To everyone who organizes a public event on open source research, I recommend to not only approach the few already well known voices in the field but to take the effort to find and invite speakers who can contribute new perspectives and who have done research on topics that are not always in the spotlight.
SH: What were the most meaningful conversations you had during your time at the Berkman Klein Center? Do you plan to use any of your connections or insights from the fellowship in your future work?
JW: I am very grateful that I was able to be a Berkman Klein Fellow this year. It was a great opportunity to be part of a community of people who all reflect on how we integrate new technologies in our lives but from various different angles. Each fellow and community hour provided me with insights into a different technology-related topic and I liked the “surprise” effect of being able to learn new things about topics I usually don’t have the time to think about. This has definitely had an impact on how I approached my own projects with Bellingcat. I feel that being immersed in such a knowledgeable and collaborative community has unlocked my creativity and I am looking forward to continuing to learn from everyone in the Berkman Klein sphere in the future.
Johanna Wild was a joint 2023–2024 Nieman-Berkman Fellow in Journalism Innovation, a joint fellowship administered between the Nieman Foundation for Journalism and the Berkman Klein Center for Internet & Society at Harvard University. Wild is currently Investigative Tech Team Lead at Bellingcat.
