Tracking atrocities in Sudan: 'The world has become significantly less anonymous for war criminals'
Since April, Sudan has been rocked by fighting between two factions of its army. At first, the violence was contained in the capital city, Khartoum, but in recent days fighting has flared up in western Darfur, ground zero for a genocide that started back in 2003 and left hundreds of thousands dead.
Arab militiamen, known as janjaweed, or “devils on horseback,” were able to kill so many in Darfur in such a short time because the area is so remote — there was no one to witness the atrocities or hold the perpetrators to account, so they continued apace.
That’s what makes this latest conflict so different: Technology is allowing third-party observers to document human rights abuses in near real time thanks to, among other things, low-orbit satellites.
Researchers like Nathaniel Raymond, the executive director of Yale’s Humanitarian Research Lab, have been using satellites not just to document the violence, but with the right on-the-ground intelligence, to predict attacks before they happen.
The team recently documented evidence of war crimes in Ukraine with a report that provided both photographic and other proof that Russia was behind the systematic relocation of thousands of children from Ukraine into Russia and Russian-controlled regions of Ukraine.
Now Raymond and the team are working with the U.S. State Department to document human rights abuses in Sudan. It is a bit of a homecoming for them — they pioneered the use of satellite analysis and open-source intelligence in Darfur more than a decade ago and now they are back with better tools and a focus on ending a crisis that is decades in the making.
This conversation has been edited for length and clarity.
Click Here: Let's start at the beginning. Can you explain how you got into this work?
Nathaniel Raymond: It’s a crazy story. In 2005, I was deployed by Oxfam America to Biloxi, Mississippi, in the aftermath of [Hurricane] Katrina. We were on the peninsula, in a primarily African American community around the Division Street Baptist Church. And I observed that there was almost no federal or state assistance making it onto the peninsula 48 to 72 hours out from landfall. But meanwhile, over on the less affected, primarily white side of town, there was a significant amount of aid. I began to think about this concept of digital invisibility, that the white population was more visible to responders. They had cell phones. They had cars where they could charge those cell phones. I had a friend who worked at Google, and he asked me to write a memo about how Google technology could have helped mitigate that digital visibility factor in Katrina. I sent the memo to them, and I called it Bat Signal — basically saying that, with a combination of online data and Google Earth, we could have a common operational picture to understand where aid was and was not going.
CH: What was the response?
NR: That memo sat in a drawer for five years. And in October 2010, George Clooney's people gave me a call and said, George Clooney and a large part of the cast of Oceans 11 — Matt Damon, Don Cheadle, Brad Pitt — want you to build what was in that memo. They wanted to work in Sudan, and they asked if they funded it, could I build it? And I said, I can sure as hell try. So I went to the Harvard Humanitarian Initiative and put a team together to begin to build what was in that memo.
CH: And, and could you explain what that was?
NR: It was about being able to break data down across streams to create a mosaic effect, what is now called open source investigations. The theory was that we could not only document attacks against civilians in near real time, but we could begin to move into a predictive posture where we could see when and where the Sudan Armed Forces or other actors might attack civilian communities ahead of time and position satellites using open source information and pattern of life information to catch an attack that hadn't happened yet.
CH: Was AI involved at this point to try and predict this or was it more mathematical models?
NR: It was done entirely manually. To the point where we even used string and markers on a laminated map in rulers to plot flight directions of bombers.
CH: Like something out of A Beautiful Mind?
NR: It looked like something out of a Wes Anderson movie, like a kid had opened up their pencil case and basically had started sketching on a map and then used string to do radial plotting of operational ranges of Antonov and Sequoia GroundStrike fighters. And really the critical part of the process was the integration of the non-imagery data taken from news reports and reports from humanitarian agencies. We would aim satellites into the places that humanitarians and peacekeepers couldn't go because if they were being excluded from areas, we began to realize that's likely where forces were massing.
And so the challenge here was: How do you get data from inside a conflict zone — by satellites, human sources, media, social media — to begin to talk to each other? Everyone focuses on the satellites, but at the end of the day, what we do is really at the intersection of technology and anthropology.
CH: Tell me a bit more about that intersection between technology and anthropology. What does that look like in practice?
NR: [When starting out] we went to the Widener Library at Harvard and we got anthropological texts from the 1950s and the 1930s, and we studied the maps in those texts of the movements of [various ethnic groups]. We used that to begin to plot potential attack routes seasonally based on where they traditionally were with their cows. And there was this one moment early on in Sudan [in 2011]. We thought we saw this massive amount of dismounted infantry moving through a mountain pass towards a village, and we got all bent out of shape about it. And a colleague of ours, from a satellite imagery company we bring in to have a second opinion, he's like, Gentlemen, you just caught yourself or herd of cows. They were longhorn cattle. Just because you see something, it doesn't mean it is what you think it is. Or if it is what you think it is, it doesn't mean that its intent is what you think it is. Over time, you begin to build the ability to see contextual patterns of life.
CH: In the conflict more than a decade ago, were you able to successfully predict attacks?
NR: Definitely. The first major Sudan Armed Forces attack happened in Abyei [a region on the border of Sudan and South Sudan] over a Memorial Day weekend just over 12 years ago. Two to three weeks out, we had been watching [the Sudan Armed Forces] build roads for tanks. We were looking for HETs — heavy equipment transports — which they used to bring the tanks in. And they had begun to burn villages. We could see a pattern to these arson attacks, and two to three weeks out, we had seen enough to know that these initial arson attacks were preface moves to a larger strike on Abyei. So I went and warned everyone in the consortium, including George Clooney, that the [Sudan Armed Forces] was about to attack. And then I was at a friend's wedding and it was Friday night on Memorial Day weekend, and the call came in that it had begun.
They attacked just as we predicted — in a horseshoe pattern, from two directions in the places where they had built up forces, north of the burned villages. And once we got shots over Abyei, we could see tank treads everywhere. We could see that they had blown the bridge out of Abyei after civilians had run across it. And what was incredible about that moment is that we had gotten ahead of it enough to be able to issue a human security alert and we said Abyei is about to be attacked. There were people at that time who thought we were crying wolf. But we were correct.
CH: Have satellites really changed what you can do there now?
NR: Oh absolutely. I mean, there are days in Ukraine where we'll get more imagery than we would get in a week or a month in Sudan [a decade ago]. And that’s because we now have micro satellites. [Commercial satellite] companies like Planet Labs and BlackSky get us multiple shots on goal a day, and that begins to give you a fancy term called MTCD: multi temporal change detection. The trick here is trying to get as many change-detecting shots as close to each other to build a time series. And in many ways, our biggest challenge now is that the sheer amount of data can overwhelm your ability to conduct an investigation.
CH: The firehose problem.
NR: Exactly. And this is a place where [we can use] AI and RSS feeds and semantic scrapers. But that's a double-edged sword: you have to have those tools to be able to chew through this much terabytes of data, but if you set your parameters wrong, you can end up degrading the quality of the product. New ways of seeing create new ways of being blind.
CH: Can you talk a little bit about what you're doing now with the current Sudan conflict?
NR: What we are doing now, which has recently been announced by the U.S. Department of State, is our team at the Yale Humanitarian Research Lab at the School of Public Health has been supporting the Department of State as part of the Jeddah negotiations on the ceasefire between the Sudan Armed Forces and the Rapid Support Forces. Our primary objective is to support situational awareness for humanitarian actors that are seeking to deliver aid. So what that means is detecting villages that have been attacked. And we're doing that not only through imagery, but through thermal sensors through NASA satellites. They have names like VIRs [Visual and Infrared Imaging Spectrometer], which is a thermal radar. We're looking for roadblocks, we're looking for attacks on humanitarian facilities, we're looking for indicators of displaced people in humanitarian need.
CH: And are you trying to do the predictive stuff as well?
NR: Yeah, there are two pieces to the mission here. One is providing situational awareness that will help humanitarians deliver aid to those who need it most. The second part is human security warning and basically determining where people could be in danger. It's poignant for many of us who have worked on Sudan now for 13 years. On one hand, it's tragic that we are still doing this. On the other hand, it’s rewarding to be able to go back to a place where some of us have looked at these locations in such detail [for years].
CH: Is there a specific thing that has happened that has made all this come full circle for you?
NR: Well, thinking back to Abyei in the spring of 2011, we began to read the language of arson attacks on what's called “tukels,” which are traditional African huts common in East Africa. Some are pyramidal; some are conical. When I was deployed as a consulting advisor to the U.N. Mission in South Sudan in 2015, I spent all my free time visiting different tukels. We had been seeing them from space, but to be able to look at them from the ground was critical. At that point, we were developing an algorithmic tukel detector, and we took an algorithm from the Jet Propulsion Laboratory at NASA used for seeing craters on the moon and Mars and inverted it to see the roofs of tukels to speed up arson detection. So as we come back to, unfortunately, the same tragic phenomena — burned villages — we have built a lot of tools to be faster at before-and-after counting. We're looking at it with new eyes, new capabilities, and the decisive factor, which is experience.
CH: And what effect has the work had on the situation on the ground?
NR: Well, that's a good question. I think we've been able to support the State Department and other actors with situational awareness products we’ve produced, to basically bring the parties to the table and show evidence [of alleged war crimes]. Information itself is no substitute for political will. But we've been able to really show the value in this conflict of having rapidly produced, high-quality, and scientifically validated open-source products that can be shared with multiple parties to a conflict as a basis for negotiation.
CH: Is the current satellite situation in Sudan such that it's genuinely hard for fighters to cover up what they're doing, or is that too optimistic?
NR: The thing about Sudan right now is that we can see more than we could back when we first started. But to what degree is there a causal protective effect from our monitoring? I still haven't figured out a methodology to measure that. You know, both sides in this conflict are aware that they're being watched. And so there's this constant responsibility on our side, which is to figure out how we can communicate to the alleged perpetrators that we are seeing them without giving them an intelligence advantage.
Back when there were fewer high-resolution or medium-resolution satellites, [fighters] could time their killing and dodge [the satellites]. Now there's a higher chance that if you are operating both day and night, something is going to pick up a tell that you were there. The world has become significantly less anonymous for war criminals. And the fact is, data is people and data is power. And the story I'm telling here is about how we use that power for the most vulnerable people on the planet.