Investigator Interview: Jordan Wildon

The co-founder of Prose Intelligence, developer of Telepathy, and former journalist with Logically and Deutsche Welle talks about building tools and conducting investigations

Jordan Wildon’s experience as an investigator helped him identify the need for a tool to gather and analyze Telegram data. So he built and released Telepathy, a free open source application. Wildon recently left his job at the Institute for Strategic Dialogue to start Prose Intelligence, a company offering OSINT tools, training and services. Its first product is Telepathy Pro, currently in private beta.

I’ve followed Jordan’s investigations since he worked at Logically. We recently spoke about the investigative mindset, his approach to developing OSINT tools, and how much of the investigative process to include in an article. This is the first in a series I’m calling Investigator Interview. I hope you enjoy it! Feel free to suggest other people to interview in the comments.

This conversation was edited for clarity and length, and Jordan reviewed the edited transcript to ensure accuracy.

---

How did you get into journalism and how did you find your way into the type of work you're doing now?

I started out doing radio. I fell in love with documentaries, basically, and bounced around various places. Then I moved to Germany where I was the host and producer of Deutsche Welle’s environment radio show. 

I eventually moved from the radio department into the social media department. With that I was not only managing the flagship social accounts but at the same time I could witness breaking news events. And very quickly I moved into the “bad things on the internet” beat.

Do you think of a particular story or a moment where you were sort of like, “This is the thing I want to be doing?”

I think the crucial moment might have been when I did a freelance piece for Vice. It was just because I was being a nerd and reading through Google transparency reports and found there was a German princess who many years ago got drunk at St. Andrews in Scotland and attacked a bunch of people and talked about killing Muslims. What came out of the Google transparency report was she had hired lawyers to get all the references from that removed, including articles. Part of it was the online investigation, and part of it was speaking to the experts and doing all the hardcore journalism stuff. 

Just generally my mind likes picking apart and piecing together things. Also, being at DW was great because they were doing some really good OSINT work at the time. I got pulled into the OSINT community super quickly through that.

Do you ever think about the characteristics of you as a person and how they relate to this work? There are certain types of people you find a lot in this community, but it's not homogenous. I'm curious for you, when you think about your characteristics, and the things you like to do, what do you think has matched you for doing this kind of work?

It would be an oversight to not say growing up terminally online. 

I think that being online and seeing platforms come and go and experimenting with new places and spending, god knows how long on Tumblr. Those kinds of things, I think, give you an idea of communities online and how to sort out the technical aspect of how it works. I'm not necessarily talking about on a programmatic level, but just like, “Here's how online spaces work.”

In terms of other characteristics, I guess it's just a case of that mentality of, “Okay, can I pick apart this thing and understand it?” Whether that's, "How does webapp work?” Or how to do something with programming, or to sit tinkering with physical electronics.

One of the things that a lot of people say — and I do too — is that tools are great, but they come and go. If you're focused on tools and all you know how to do is use tools, then you're screwed when they disappear. As somebody who's building tools I'm curious of your view of that, and how you balance the mindset stuff and the tool stuff.

It's probably a US military saying or something I saw online, but [I think about] the idea of redundancy. Like, if you have a tool two is one, one is none. If you have a tool and it’s only one, at some point it’s going to go and then you have nothing.

The techniques and the methods are really what the staple is. There are many ways to get information, but it’s about being able to go, “How can I verify it? How can I parse it? How can I then translate this into a way a normal average everyday person is going to understand?” 

I personally feel like I still am trying to figure out how to describe the process in my articles in a way that is not boring and is not needlessly nerdy, or in some cases showboating like, “And then we did this, and then we did that." Cutting down all of that just to give what's necessary to show your proof, but not to bog it down. Are there things you think about when you’re getting to that point of needing to communicate it to as broad an audience as possible?

I’ll try and use an example from one of our Logically stories, the GhostEzra investigation. That had a huge trail of [information]. Most of that made it into the story. Some things have to be cut because of protecting people around this individual that don't deserve to be pulled in. You've got not only the legal but the ethical things of, “How can I explain this so that you, the reader, trust that this is true?” But also, how can I do it so you’re not going to sit there and replicate the entire thing and then go after someone's family members?

So you have the legal, ethical [concerns]. And then explaining it as clearly as possible and also doing it in a way which isn't bogging down with really technical details. I agree that showboating is a big one to avoid.

What goes in and what stays out? I don't know. I think it works on a case by case basis.

That GhostEzra story was definitely one I wanted to chat about because at the end of the day he was the one who provided a lot of the clues to his identity. It's sort of like knowing what to look for and having the willingness to scroll through potentially thousands of posts here or there, pluck things out, and know what to do with them. For me it feels like a very familiar thing: Are you willing to get really boring and scroll through a whole bunch of stuff to see if there is anything there for you…

Yeah the quite interesting part of that was looking through the Telegram posts, for example. That was when I had the open source, personal version of Telepathy. So I had a CSV with every post [from his Telegram]. Being able to search through them and then to download all the media files and scroll through them.

Nick [Backovic], who also worked on the investigation, noticed that some screenshots from his phone still had the Android bar at the top and there were specific apps open. So we're like, “Can we work out what all of these apps are? And then locations that could possibly be related to them?” Some of them were finance apps and we knew the guy had an eBay account and I think an Amazon reseller account, or something like that. It's noticing those details and thinking what can I cross reference elsewhere to work out if this is something relevant? 

That's the thing some people sometimes miss — they think tools provide the answers. In this case when you're talking about Telepathy, the tool did what most of them do, which is help speed up or systematize a process. But you still have to sit there and know what to search for and know what to look for. It doesn't replace the thinking.

Yeah, it potentially shortcuts it but you still have to go through and verify it. Absolutely. With our company and also with the progress of Telepathy, that's all coming from being an analyst and doing it and thinking, “Okay, how can we make these things a little bit smoother?” Thinking what methodologies, what techniques do we use and how can this be shortened in a way which is useful for what analysts or investigators or journalists do.

You worked at Deutsche Welle and InVID is one of the few purpose-built journalism OSINT tools out there. It's got so much in there but the core of it is a workflow tool. For the most part, it's making it faster to do reverse image search or pulling out some stuff from videos. One of the trends that’s happening with tools, and I think a lot of it is mostly in the synthetic media area, is trying to give you a score of whether this is an AI generated photo, is this a deep fake video or not. I'm curious about your thoughts on deterministic tools where it's like, “I’m gonna give you a score, but you're just gonna get the score.”

It's interesting you bring up InVID. I had some freelance work testing that through various iterations. And the same thing came up at times of needing to explain to the person using this how we got to this point. I've seen a lot of tools that do that kind of black box thing of like, “We’ve determined this is false.” 

Whether it’s an investigation or even a breaking news piece or doing analyst work or anything else, you need to be able to stand up exactly what's happened. Tools that just tell you, well, you can't then prove how you got there. You can't show your working. The attitude I always have towards it is, how can I use tools that give the raw information and then process it myself and then explain it?

Let's talk a little bit about Telegram and the tool you built for it, Telepathy. In the last few years Telegram seems to have become far more important. Part of it is the full scale invasion of Ukraine, obviously. Telegram’s hugely important for that. And as the big social platforms tightened their moderation, Telegram became a place for certain types of extremist groups and other communities to go to. For you, how do you see the evolution of Telegram and what got you started wanting to build a tool for it?

One of the big things I was looking at initially was when all of the deplatforming, especially from Twitter, with QAnon happened. Then also just in general with Telegram becoming a thing in the Western world. 

There's now a significant chunk of the user base that goes, “Well, I can just go and say anything here and no one will stop me” and then they also get pulled into other stuff. And it's just like an incredibly interesting platform in terms of how those networks work and how things move around and how things are shared. 

The Telepathy tool at first was network mapping, just like: we have these channels, how are they linked? And then it grew into archiving, because the Telegram API gives you quite a lot. Telepathy was always based around working on individual channels and then being able to get a sense of everything [in them] and being able to do data analysis on posting times, top forwards or top replies, just those kind of shortcuts. 

In terms of the functionality piece, are you as part of the enterprise version building a database of Telegram channels that people can have access to and search across? Or is it more like the tool assists people as they discover channels and want to archive and monitor them to sort of set their own universe?

A big part of it is we want to be doing this as responsibly as possible. The way that it works is Telepathy Pro is the tool that takes the data and transforms it in ways that are useful, like with image hashing, or character recognition and things like that, and then provides it to clients within their own systems. We don't hold any data. 

It would be nice to be able to build something like that but we have to get past various hurdles to get to that point and be able to do things like a TGStat or similar kinds of tools. But we're not based in Russia so we can't just get anyone's data and go, “Here you go.”

Legally it’s not an option and secondly, and the most important one, is ethically, how do you build it so it’s good for us while also maintaining individual people's privacy? And put it in the hands of people that are actually going to be responsible about it as well. That's why there's some things that I'll probably never put into the open source version of Telepathy, because people can do damage with it. 

Give me the quick description of your company, Prose, and what you guys are doing. One of the things that I always found interesting about Logically is, in one part, it seemed like they were selling a product, a tool. But also, of course, they had a team doing fact checks. With Prose, are you guys purely positioning yourselves around building tools? Or are there services or public facing things you're planning to do?

The Logically situation was classic and nice because what kept the lights on was them being a tech company. And then we would be able to have the affordances to sit there and go, “Okay, I'm gonna take three weeks to do this investigation” and be able to do that, whilst also having entire editorial independence from the rest of the company. That's how we managed to do those investigations. 

With Prose, obviously, we've got the tool itself and that comes from having it already be out there and needing work to continue developing and building that. One stream is the tech stuff. There's a bunch of things that I'd want to work on down the line in terms of tech. But then also there's consulting. We're trying to properly launch that soon. 

We've done a lot of government work, a lot of investigations for different companies, and also worked helping journalists find things in the past. So we’re kind of trying to work out the best of both. 

One of the things I always like to hear from people doing this work is the element of controlling for bias and trying to keep yourself from going too far in one direction or another. As you're trying to figure out what's really going on, like who is behind this, who is behind that, you may have some theories, some hypotheses. How do you approach that element and keep a sense of awareness while you're engaging in the work? You are human, you interpret things in certain ways, and we all bring our own experience to things. How do you manage that?

When you're working with a team of people investigating, you can come out and you can absolutely know something and be like, “I know this is the case.” And someone will be like, prove it. 

Being able to have people that check your work and assumptions, whether that's within a company, within a community of people who work together, or that's just people that you know in the same field is really important.

Then there's a lot of keeping yourself in check. Making sure that you also hold yourself to that same standard of checking it two or three times in two or three different ways. 

Ultimately, it's being willing to admit that you're wrong at times. I think one of the important parts of doing this work and doing it well is being able to accept that you're wrong at times.

For sure. Being willing to drop that thread of the investigation, or the whole damn thing if it comes to that. So I’ll put you on the spot: is there a story or report that you really like? One that you always think of and recommend people read?

The first thing that comes to mind is the Deutsche Welle investigation into Uyghur camps in China. I saw it from afar with the colleagues that were working on it at the time and it was just amazing how they came together to say, “Okay, we're going through all these records, we’re working with different people in different places and then we're going to tell you a story in a way where you can see it.”

Julia Bayer’s Twitter thread, and this video in particular, is amazing. I really like how they took open source investigation, interviews, documents and put them all together to paint a full picture. Blending satellite data, information from the documents and interviews with 3D animation is such a great way to make a story accessible.

In terms of pure “Wow, how did they do this?!” The classic example from BBC Africa Eye, “Anatomy of a killing,” is still one of my go-tos for showing how clever OSINT can be.