Geo Political Insiders
Los Angeles
Fineas speaks with Dr. Dylan Craig about proxy warfare, gray-zone conflict, and the changing nature of war. They discuss what the National War College does and who it trains, whether diplomacy between states can ever be separated from the threat of force, the benefits and hidden dangers of fighting through proxies, and why sponsors so often lose control of the groups they back. Using the Congo wars as a case study, they explore Dr. Craig's concept of "sovereign interstices," how drones are reshaping the battlefield, where diplomacy ends and war begins, how well we can predict where conflict will break out and whether AI will help, the security threats Americans overrate and underrate, what high school students should be doing now to prepare for careers in international affairs, and other topics.
Dr. Dylan Craig is an Associate Professor and Director of Research and Writing at the National War College in Washington, D.C., where he teaches senior military officers and government officials. He is a scholar of proxy wars, gray-zone conflict, piracy, and state formation, with a regional focus on Africa, and the author of Sovereignty, War, and the Global State (Palgrave Macmillan, 2019). Before joining the National Defense University in 2022, he taught security studies, colonial and Cold War history, and research methods at American University's School of International Service and at Rhodes University in South Africa. He holds an MA in Historical Studies from the University of Cape Town and a PhD in International Relations from American University.
Transcript lightly edited for clarity. Click any timestamp to play the video from that moment.
Fineas Hi, I'm Fineas, founder of GeoPolitical Insiders, a site devoted to conversations with diplomats, policymakers, scholars, and national security experts about the forces shaping international affairs. Today I'm joined by Dr. Dylan Craig, a professor at the National War College and an expert on proxy warfare, gray-zone conflict, and African security. He is the author of Sovereignty, War, and the Global State and has written extensively on proxy warfare and the use of nonstate armed groups in conflict. Thank you, Dr. Craig, for joining me.
Dylan Craig My great pleasure, Fineas. Thanks for having me.
Fineas I first met you this summer when you spoke at the NSLC National Security program in Washington. What originally made you interested in international relations and national security, and how did that interest turn into the career you have today?
Dylan Craig Well, I think I mentioned to all of you at the time that I was as surprised as anyone to find out there was actually a discipline that contained the skills I was looking for to answer the questions I had about the world. I was trained as a historian. In the course of my history work I became interested in conflicts in Africa, and I proposed what I thought was a history PhD on why proxy war recurs as a tool of violent statecraft for African states. When I showed it to my history advisor, he very politely said, "Well, this is very interesting, but you know there's this field called international relations, and it might be a little closer to what you're interested in." And he was 100% right.
It turns out it's not so much that the topic I was interested in was only studied within international relations, because you can come at that topic from a million different perspectives. It's that the specific kinds of questions I had, like why do states use this technology or this style of warfare, are the questions that discipline specializes in. The discipline that asks "Why do states do X?" is international relations. And once I was among other international relations practitioners, what excited me wasn't necessarily what I was studying, but how I was being taught to study it. That's what I've tried to focus on in all my subsequent work.
Fineas You're a professor at the National War College. For people who aren't familiar with it, what does the National War College actually do, and what kinds of people are you teaching there?
Dylan Craig It's a very interesting institution. It's part of a category called PME institutions, professional military education institutions. Under that heading we can put all the military academies: the Air Force Academy, Annapolis, West Point. What sets the National War College apart is that all military officers need to go through something called JPME, which is professional military education aimed toward what they call the joint operating concept. The idea is that in conflict and in statecraft, the Marines don't just get to play with the other Marines. Everyone's going to be integrating with the Air Force, the Air Force is going to be integrating with the FBI, the FBI is going to have to integrate with Treasury, and all of these people are going to work together, sometimes under the auspices of the State Department.
It's about getting at that ability to not just be good at what you do, but to understand what other people do and how they tend to think, so that when you work together, that work is efficient. That rationale underpins JPME II, which is the National War College. I tend to teach students who are about my age, late 40s or early 50s. They've usually been in the military for 20 or 25 years. They're at the O-5 rank, so lieutenant colonels, and the other branches have other names for that, plus civilians from places like the FBI at an equivalent civilian rank. These are all people who know a lot about what they've been doing for the last 20 years, but they may not know how the issues they deal with, say, in the Middle East, look to someone who has spent 20 years doing something very different.
The JPME II level is meant to give folks a year of experience thinking alongside people who are slightly different from them. At the end of that year, when we grant them their master's degree, they are as well positioned as we can make them for senior decision making. These are folks who go on to become flag officers and general officers, or go up into the senior executive service. As I like to joke, this is the last time they are junior enough to be told what to do.
The other interesting thing about the National War College is that about 10% of our cohort are foreign officers from nations that are partners with the U.S. They're learning not just how people from other branches think, but what it's like to work with Americans, and Americans are learning what it's like to work with Pakistanis and people from Kazakhstan and people from Israel. It's a level of jointness that concedes the fact that very rarely does the U.S. get to operate alone. It usually brings friends with it, and the question of how best to work with those friends is left to the National War College curriculum to deal with. Overall, I'd characterize it as an international relations degree. We're asking questions like: What is security? What should states do to achieve it? What is strategy? And by and large, it's also an excuse to rub shoulders with different kinds of federal employees and really work on becoming a better team.
Fineas How has teaching senior military officers and government officials, with all those different perspectives, changed the way you personally think about foreign policy or warfare?
Dylan Craig It's tricky. I want to answer that question in the affirmative and say it's absolutely blown my mind, and it is very different from teaching civilian students who are 19 years old. But really, what teaching military folks who are advanced in their careers has taught me has come largely not from the ways they're different from the students I taught before, but from the ways they're surprisingly similar. I have students who are massively decorated, with incredible records, but they are as scared of writing an essay as a 19-year-old freshman.
I find that fascinating, and it really helps me zoom in on what I'm supposed to be doing as an educator. If the task remains undone all the way through someone's undergraduate education, and what's holding these folks back when they're about to become senior decision makers is that they've spent their whole lives being scared of having ideas in public, then that's the thing to work on. Usually what holds people back when they struggle to write is that they're worried about sticking their neck out and possibly being wrong. That gives me a laser focus on what they need to do. These folks have so much to contribute. They know so much. It's really just a matter of laying a trail of breadcrumbs and saying: it is okay to have an opinion. It's okay to have a thought about what should be done.
And I know I said this to you folks at the NSLC: no one in the policy world in D.C. will ever ask you a question where they know the right answer and just want to see if you know it too. It's never a trap. When people ask you, "What should we do in Syria?" or "What are the prospects of the war in Iran?" they genuinely want you to have an opinion. They want it to be a good opinion, well substantiated and well argued. But frankly, they don't know the answer. If they did, they wouldn't have bothered asking you. That's one way my students haven't changed since I was teaching 19-year-old civilians.
The other side of it is that it's easier to convince senior federal servants of what I'm looking for. I'll give an example from a Teddy Roosevelt class that I teach. The class is on strategic leadership. If I were teaching this in my last job, I'd have to deal with the fact that very few people in the room had much experience of leadership. The folks I'm teaching now have all been leaders, and they've all been led by leaders. So if we're saying Teddy Roosevelt did this and that, and this is what made him a good leader, and this is what people have criticized him for, you can quickly turn on your heel and say to the room: Does this ring true to your experience? Have you ever had a boss who did this, or have you ever been the boss who did this?
That's very different, because it's really easy for them to identify with, and it unlocks bits of knowledge they have locked away in their heads that they need to get used to putting out there in written form or in a presentation. So it's the same in some interesting ways and different in other interesting ways.
Fineas GeoPolitical Insiders, and my own interest in international affairs, started largely with an interest in diplomacy. But as Clausewitz famously argued, war is a continuation of politics by other means. We usually hear almost the opposite idea today, that war begins when diplomacy fails. Between sovereign states, is there really such a thing as diplomacy without at least some implicit possibility of force behind it?
Dylan Craig I had to laugh when you mentioned Clausewitz, because the flip side of that is the famous way one of Clausewitz's great students reframed it: that politics is a continuation of war by other means. So it really depends on which direction you're looking at it from. But to be quite blunt about it, the vast majority of international relations happens in the absence of force and threat. The vast majority of things that happen outside our windows and affect the ways we live are the result of states seeking bilateral solutions.
A bilateral solution, by definition, is an agreement where you have to put various sweeteners and various penalties on the table to convince someone to agree with you: a trade treaty, a mutual defense pact. All of these are forms of politics where the logic by which we think someone will agree is a logic of incentive. Do this and you can expect the following good thing to happen; refuse and the following bad thing will happen. Sanctions or tariffs would be examples of the bad things.
War is very different, because war isn't just where diplomacy ends. It's really where bilateralism ends. What distinguishes war from other forms of politics isn't necessarily that it's violent. Politics can be violent, and war can be very diplomatic. It's that war involves giving up on the idea that the other person needs to agree to what you're trying to achieve. War is a unilateral act. They don't need your permission to invade you. They're just going to invade. And sure, they can simultaneously be pursuing diplomacy to convince you to quit earlier, or saying, "If we stop fighting now, we'll give you back some of your territory." That's diplomacy running neck and neck with war. But what's different about war is that it is unilateral in character.
To the extent that the world around us is largely a system of states that agree to do things, and sometimes violently negotiate how the world is going to look, I would say the vast majority of it is not war. Our eyes are just really drawn to the warlike bits, because that's the part that shows up on TV, with chaos and explosions and burning tires. If we don't discipline ourselves, we may think that's the entirety of international relations, and it isn't.
Fineas A major focus of your research has been proxy warfare. What are the strategic benefits for a state fighting through a proxy instead of directly? And more importantly, what are the risks? History seems full of unintended consequences from proxy wars. How often does the proxy end up pursuing its own interests in ways the sponsoring state doesn't anticipate or can't control?
Dylan Craig You asked a couple of questions in sequence there, so I'll start with the first one. I think you're very right to frame it as: why would you choose proxy war over conventional war? You can ask what proxy war is, and the answer is pretty obvious. It's when states partner with non-states to achieve violent goals. But that misses the interesting part. The interesting part for me is why you would choose to partner with a non-state in addition to using your own forces, which in many cases you've spent a lot of money on.
The answer is twofold, as far as I'm concerned. Number one, proxy warfare is a way of signaling your flexibility to your enemy, in terms of the rules you are not willing to break in pursuit of your political goals. For example, the Russian "little green men": folks in unmarked uniforms, speaking Russian, show up in Georgia or Moldova or Ukraine and conduct what look like very Russian military operations. No one is confused about the fact that those are actually Russian soldiers. That's not why you send those folks. You send them as a subtle way of signaling to the world community: I care enough about the rules of international politics that I'm going to pretend I'm not actually doing this invasion.
It sounds kind of silly. States that use proxy war are willing to put a fig leaf on their violent action, but the fact that the fig leaf is there really matters. The next step after someone invades you is figuring out what that invasion means. How far are they willing to go? What are they willing to do? Using proxies is a way of saying: I want this thing I've invaded, but not to the point where I'm going to stand up in the United Nations and say, yes, we have invaded you, officially and in public. People sometimes talk about proxy war as achieving deniability. I don't think deniability is a goal in itself. The degree to which states can deny involvement is a way of signaling to everyone else that this is a relatively limited military incursion, and that, for example, nuclear weapons won't be part of it.
The other reason you would use proxies is this. Gearing a military operation up or down is usually a little like a light switch. It goes on and it goes off. It's a lot of effort to send out a Marine Expeditionary Unit or, historically, to send the legions. The great thing about proxies from a war-making state's point of view is that they're more like a dimmer switch. You can crank them up a few degrees, crank them back down. If the heat gets too much and the international community starts threatening sanctions, you just stop the flights carrying weapons from your armories to your proxy. They won't go away. You don't have to fly them back to your country. They'll stay in the country where they're operating and continue their civil war, probably less successfully than when you were backing them. Then the minute the international community's attention goes elsewhere, you can start the flights right up again.
So it's a really attractive, flexible option for states that aren't so much looking for the ability to inflict violence on their neighbors, which they could do with their regular forces, but that want to dial it up and down depending on what they need at a particular time. Look at the situation in and around Iran right now, and the degree to which Iran's proxies have become an issue of national security significance. The existence of these proxies is not an on-off issue for the Iranians. They can dial up their proxy network when it's necessary for them to do so, and if they feel like making concessions, they can dial that support back down. The proxies will remain there. Hezbollah and the Houthis will stay in the fight. And Iran gains a flexibility in interacting with them that it would never have with its own conventional forces. That's the answer as to why states use it.
Now, the other thing you asked is how often proxies get out of the bag, as it were, and produce some unpleasant side effect. In my PhD research, I looked at four potential outcomes. The proxy and the sponsor both get what they want out of the partnership. Neither of them gets what it wants. The sponsor gets what it wants and the proxy is hung out to dry. Or the proxy gets what it wants but the sponsor doesn't, usually because the proxy has done something the sponsor didn't want, enlarged the war, or created humanitarian crises that draw international condemnation.
Practically speaking, these partners betray each other all the time. It's relatively rare for everyone to get what they want. But if you stack the two asymmetric outcomes against each other, proxy betrays sponsor versus sponsor betrays proxy, it's more often the case that the sponsor gets what it wants and hangs the proxy out to dry. Occasionally proxies get in over their heads, but states don't go into proxy relationships thinking that's a likely outcome, because they're aware that they hold more of the cards. They can withdraw support. In some cases they can betray their proxy to the forces of order.
There's a fascinating chunk of South African history where military intelligence and the police are both running proxies into Zimbabwe and Mozambique. At some point the military intelligence folks literally pick up all of the police's proxies and deliver them into a situation where Zimbabwean enforcers are waiting for them. That's basically a turf war between two parts of the South African state, but it plays out as a situation in which South Africa, treated as one unitary actor, betrays its proxy. And that's what happens most often.
Fineas In your book, Sovereignty, War, and the Global State, you describe what you call "sovereign interstices," places where formal state sovereignty and the state's actual power or control don't necessarily match. I recently visited Rwanda, which led me to develop an interest in the Congo wars, what's sometimes called Africa's World War. At different points you have Rwanda, Uganda, Angola, Zimbabwe, and other states involved, along with Congolese and foreign armed groups whose alliances and objectives keep changing. Can you use the Congo wars as an example to explain your ideas about proxy warfare, gray-zone conflict, and sovereign interstices?
Dylan Craig Three answers to that. Number one, let this be a lesson to all of you prospective scholars out there: when you pick the key concept on which your work rests, pick something that's easier to pronounce than "interstices." It seemed to make a lot of sense to me, but then no one can say that word. So that's one point.
The second point, and this is the last thing before I actually answer the question: the last thing you do before you get a PhD is a big oral exam called the dissertation defense. It can last well over an hour, and you're grilled by a panel of examiners who ask questions meant to test whether you're ready to emerge from that room with your newly minted doctorate. One of the last questions I was asked was: "Dylan, what is the conflict in Africa that most challenges your theory about sovereign interstices?" They ask that because they want to see whether you're resilient as a thinker, or whether you fall to pieces the minute you're asked to consider that any part of your beautiful little theory is less than perfect. Because if you do, you probably shouldn't have a PhD.
And my answer was the war in the DRC. The war in the DRC is such a huge black hole in African conflict studies, because of the millions of people who died, because of the enormous dogpile of foreign states that jumped into that war, which you just listed, because of all the human rights abuses, and because of how endemic the conflict became. It has just settled in and is still going on now. One of the reasons I wanted to do my work is that people have assumed all proxy wars in Africa look like the one in the DRC.
If we think about the characteristics of the war in the DRC, we see, first, the importance of plunder. Especially in the east of the country, it's inarguable that one of the motivations for the Rwandan and Ugandan interventions was access to gold and coltan. The United Nations did a report after the Ugandan invasion that looked at Uganda's mineral exports. They go from essentially zero to hundreds of tons, coincidentally at exactly the time Uganda conquers a chunk of the DRC. It's not hard to understand where those minerals are coming from. Sometimes it's direct: states send their militaries in to pull stuff out, load it onto military planes, and fly it back to the capital. And sometimes it's done informally through the gray and black markets.
In Zimbabwe, for example, it's not so much that the state benefited from the plunder available in the DRC, but that giving important generals in the Zimbabwean Defence Forces an opportunity, as representatives of the government, to line their personal pockets with the profits of illegal lumber was a way of keeping them loyal to the regime. So plunder is one big issue.
The other big one is ethnic identity. There's a commonly held belief that one of the drivers of conflict in Africa is arbitrary colonial borders that cut ethnic groups in half, so half are on one side of a border and half on the other. The degree to which conflicts like the DRC's involve countries invading neighbors to protect vulnerable minority populations has shaped people's sense of how all conflict in Africa works. But the vast majority of the proxy wars I studied for my dissertation didn't take place across a contiguous border. They were two, three, four countries away from where the fighting was happening. There's no ethnic loyalty binding those countries together. The DRC clearly does feature that.
The last thing I looked at was the idea that Africa's conflicts come from the weakness of African states. That's definitely true in the DRC. You've got a lot of countries saying: we could take or leave this conflict, but the DRC is such a vortex of crime, smuggling, and arms trafficking just across our border that we have to go in and secure these border provinces using proxies, simply to defend ourselves.
So among the lessons of the war in the DRC: it was a proxy war partly because it allowed folks to loot a lot from the DRC. It was a proxy war partly because states wanted to support their co-ethnics across the border and couldn't do it formally, so they did it informally. And it was a proxy war partly because the DRC was so weak and chaotic that there were all these security vortices into which a neighboring state would reach its hand, pick out a contender to support, and work with them to achieve border security. So clearly it's a proxy war. It just doesn't look like other African proxy wars.
Most African proxy wars don't involve plunder, and they don't involve ethnic loyalty. And this was the really surprising thing I found in my research: they're generally conducted by Africa's strongest states, against other strong states, not against weak states. The DRC is kind of the exception that proves the rule. Mostly, when states use proxy war, it's one strong state trying to keep a neighboring strong state weak, to achieve some kind of security hierarchy.
Take Iran's use of proxies, to go back to an issue that's in the news. They're not using those proxies to attack weak countries. They're using them to attack strong countries. And not because they lacked formal means with which to attack, but because proxies are a useful tool in those circumstances. Specifically, they let you get into an asymmetric fight with a strong state that doesn't give that strong state anything obvious to hit back at. If you attack with your own troops, they know your return address. They know where to hit you in retaliation. But when Hezbollah is firing missiles into Israel, where exactly do you hit? Do you attack Lebanon? Do you try to attack Hezbollah? Do you attack Iran? It's anyone's guess. That's why you would use proxies against a strong state.
The DRC is a fascinating and a horrifying war. I think it certainly showed that Africa, in the wake of the Cold War, was perfectly capable of getting along with its own horrifying wars without any superpower involvement. But it's an unusual bird, in that it has these three features which in people's minds are true of all African wars, but in fact are true of only a tiny minority.
Fineas To stay on the Congo wars: one thing I noticed in studying them is that at some point it becomes difficult to say who's really controlling whom. A state may sponsor an armed group for its own purposes, but that group has its own leaders, interests, and objectives. Does thinking of these groups simply as proxies sometimes give the sponsoring state too much agency?
Dylan Craig It really changes from case to case. Two examples come to mind of cases where it's easy to look at a proxy relationship and see that anything that happens, happens because the sponsor is pulling the strings. One is the rebel group M23 in the eastern DRC. One of the ways we know it's Rwanda pulling their strings, as opposed to them pulling Rwanda's strings, is that every now and then Rwanda essentially hangs M23 out to dry. It gives up their leaders to international judicial bodies and disavows them. It's done this a couple of times, and each time it becomes clear that if someone has the ability to conclude this conflict at any time it's politically expedient to do so, it's Rwanda, not M23.
The other example is South Africa's sponsorship of a group called RENAMO in Mozambique. The historical record shows that the South Africans specifically and on purpose gave RENAMO only the lightest weaponry possible. They had a giant stash of Soviet weaponry and were sending it to their two proxies in the region. UNITA in Angola was getting all the heavy weapons, and RENAMO was getting almost nothing. The most reasonable explanation is that, from South Africa's point of view, the goal for RENAMO was not to win its war. It was supposed to be a continual irritant to the government in Maputo, to kill a couple of people, to make the country look bad. If RENAMO had actually marched into the capital and taken control, South Africa would have had to figure out how to deal with this very strange, very eclectic group of people it had armed. And it didn't want that.
Then flip it over to the north of Africa, and you've got the Chad situation, one of the case studies I looked at in my dissertation and later in the book: Libya's support for proxies in Chad during what are called the Toyota Wars. Libya's thought process goes like this. They want a chunk of the north of the country. They know they can't just conquer it. But if they can get the Chadian government to give it to them, surely that's fine. One state can transfer territory to another if it wants to. So they arm a proxy and send it south to knock out the previous government, take control, and hand them the north. Once that proxy finds itself in power, it becomes far less interested in giving the Libyans what they want. In the end, the Libyans find themselves fighting the very people they put in charge, and that goes on for years, with figures like Hissène Habré bouncing back and forth between the sides. Clearly this is a situation the Libyans, in retrospect, probably wish they had not started, because the genie is very much out of the bottle.
Another obvious example is the mujahideen in Afghanistan. The fact that the U.S. ends up experiencing blowback from that conflict indicates that perhaps the U.S. did not hold all the cards in its interaction with the mujahideen, or rather, it left the mujahideen with enough cards that eleven or twelve years later the chickens come home to roost, to mix my card and chicken metaphors. You get what I'm getting at.
Fineas In Sovereignty, War, and the Global State, you identify the armed drone as one of the potential discontinuities in the history of interstitial warfare. Since you wrote the book, we've seen drones become extraordinarily important in Ukraine, Russia, Iran, and other conflicts. They also seem to occupy a space that doesn't fit neatly into the traditional division of warfare into land, sea, and air. Have drones created something like a new battlespace, a kind of "drone space," and how do you think they're changing the calculus of war?
Dylan Craig This is a fascinating point. For the benefit of folks who haven't read my book: most of it is about continuity in states' use of these kind of rowdy, hillbilly political actors who live out in the borderlands, the pirates, the privateers, the proxies, who can be used to scratch state itches. But even if the book is all about continuities, I think it's important occasionally to confess to the things that don't match your neat little model, the problematic cases. And the reason I flagged drones is what I see as characteristic of the way they're used. You're right to point out that we've had air-based drones for a while now, and we're starting to see them in other battle spaces.
Go back to the question you asked right at the beginning about the difference between war and politics: war is unilateral violence. In classical models of warfare, when you inject a navy into the Strait of Hormuz, or an army into Afghanistan, or an air force into the skies above Serbia, what you're generally doing is extending sovereign control. You're saying: I control the seas around here, and here are the things I'm not going to allow you to do. When you inject an army, you get to do all of those things, and on top of that, because sovereignty derives from the ability to credibly govern people, and people live on land, taking control of the people shortcuts you directly to the point where you've taken sovereignty away from whoever purports to govern them. Something similar is true of the air. What's interesting about air power is the degree to which it gives you the power to effect political outcomes on the ground.
So how do drones change this? The challenge, I think, is that it is very hard to effect politics through a drone. Look at Ukraine and Russia. Drones there are not helping either side win. They are stopping both sides from losing. The drones the Ukrainians are fielding make it basically impossible for Russia to achieve a decisive battlefield victory, because if Russia tries to roll tanks into Kyiv, they'll all be blown up by cheap little FPV drones. We mustn't overstate what the Ukrainians have done with drone technology: they've essentially produced a battlefield full of remote-piloted hand grenades. But the degree to which they are directly targeting Russian personnel and thin-skinned Russian vehicles has made it impossible for Russia to do politics on Ukraine, violent politics. Similarly, Russia is throwing hundreds and hundreds of one-way attack drones at Kyiv to blow things up. Is that going to make the Ukrainians quit? It hasn't yet.
The challenge right now, when we try to figure out whether drones are a new form of warfare or a new battle space, is to ask: with what drones are capable of right now, are there political outcomes that used to be impossible and are now possible? Or are drones just going to keep doing what states have always been able to do? I had an interesting conversation with the chair of the Air Force department at the National War College when drones really started to take off in the news. I dropped by her office and said, "You're Air Force. What do I need to know about drones?" She made it very clear that, from her point of view, the interesting thing about drones is not what they can do, because that's not very different from what we've been able to do for ages with guided missiles, bombs, and soldiers with rifles.
The thing is, she said, it used to be that you had to be a superpower, an advanced industrial democracy, to have both precise targeting, the ability to reach out and touch a specific point on the Earth or a specific vehicle, and the ability to deliver mass, huge explosions, precisely. The kinds of politics you can do by delivering huge amounts of explosives onto very precise locations used to be something only superpowers got to play with. Nowadays that's not the case. Drones haven't allowed us to do much that's different on the battlefield, but they have radically increased the number of people who can do things on the battlefield that you used to have to be a superpower to do.
That's what I think we really see in Ukraine. It's not that Ukraine is doing anything new. We have a situation a little like World War I, where there's a new technology, in that case the machine gun, which means that smaller countries that used to be politically irrelevant can no longer be considered irrelevant. It's not yet going to change what we use warfare for. And until it does, I don't think we need to spend as much time thinking about it as we do about cyber war, because with cyber you really can do things you couldn't previously do with a Marine and his rifle. You can achieve a variety of political effects that weren't possible before: deepfakes, cyberattacks. That's a technology that really deserves some thinking. I don't put drones in that category yet. They're going to have huge effects, but they'll be effects related to who's around the table, not what is happening at the table.
Fineas Your concept of a sovereign... how should I be saying that word?
Dylan Craig Interstitial. You know, this is why I say I should have just called it gray zones. If I could go back and do it all again, I would have just said gray zones. But because I was out ahead of the pack, and I thought, well, an interstice is a gap between things, I just assumed that would be the right term for it. Gray zones is much better.
Fineas All right. Your concept of gray zones is partly about places where the traditional rules and boundaries of sovereignty don't quite match the reality of violence. Do drones create a new kind of gray zone, not necessarily a geographic territory but a technological space in which states and nonstate actors can use force across borders while sometimes avoiding the traditional consequences of military intervention?
Dylan Craig This is a really good example of what I was just talking about. Take a hyperpower like the U.S. The U.S. has always had the ability to take an area of the world and make you very scared to put navies into it, because it had long-range cruise missiles and satellites and all these fancy toys it could use to say to its enemies: we can get you anywhere. What we have in the new world is that countries like Ukraine, which didn't used to be able to make that kind of threatening gesture, can do it with drones. It's not that anyone is going to fight in a different way than before. People have been threatening to blow up each other's ships since there have been ships to blow up. But the technology it used to take to do that has now evolved to the point where terrorist organizations, non-states, the Ukrainians, the list of people capable of playing in that space is hugely different.
That means people are going to try to get things done with war that they weren't previously capable of turning to war to do. Organizations that used to have to do outreach campaigns or letter writing or crime will now be able to look at the battlefield and say: we have a ticket to get onto that place. So it's going to make things a lot more violent. It's probably not going to change what the world looks like so much as make the world a lot more loud and chaotic, because more people will be trying to do the same sorts of things that a relatively small group of states used to be able to do.
There's a famous motto that Louis XIV had engraved on his cannons, and I found a couple of U.S. cannons down in Savannah last time I was there with the same motto, whether because we got them from the French or because American armorers liked the saying so much. It's the Latin phrase ultima ratio regum: the final argument of kings. That's how Louis XIV describes war. When you've made all of your pleas to logic, to interest, to honor, to law, and there's only one argument left, the argument of force, then you make it with your cannon. What I see as different in the world of drones is that it's no longer the final argument of kings alone. Kings are now just one player that can use it very effectively, that can bring mass and precision to the battlefield. I don't think that's going to fundamentally change the things wars are fought over, but there's going to be a far wider diversity of people who feel that war is a legitimate and practical tool for them, and that's going to create a more conflictual future.
Fineas Does gray-zone conflict make the distinction between diplomacy and warfare even harder to maintain? If a state is funding an armed group, conducting cyberattacks, imposing coercive economic measures, or supplying intelligence while deliberately remaining below the threshold of open war, at what point should we say diplomacy has ended and warfare has begun?
Dylan Craig It's a really interesting question. Remember that the position I come at this from is that proxy war is nothing other than an attempt to have a conversation. It's a way of saying to other states: I want this thing, but I'm also willing to be subtle about it, and potentially that opens the door to a negotiation. It's not inconceivable that you could offer me something I want, and I'm bothering to keep my hands clean here so that, if necessary, I can walk away from it. That's what Rwanda has done every time it's agreed to give up its proxy. It's extracted some sort of concession.
If you see things the way I do, the fact that more people are running to the battlefield with shiny new drones and shiny new cyberattacks doesn't necessarily mean the end of politics. It just means more people are going to be sprinkling a little violence into the cost-benefit calculation they want other countries to make when deciding how seriously to take them. It's very much like the way the IRA used terrorism in parallel with the political process. The violence is there to add urgency, to remind you that instead of talking, we could just be blowing you up. Is that the end of diplomacy? I wouldn't say so, because in the end they're still trying to set up a world in which someone decides it's less trouble to give them what they want than to fight them.
Contrast that with a situation in which diplomacy is considered to have failed: we've tried talking, and now we're going to war. If you look at Clausewitz, war is the moment in which you fight the enemy because you're trying to remove the thing that allows them to say no to you, which is their military. You use your military to destroy their military, because once their military is destroyed, they have to do what you want. That's not a negotiation. That's not saying, "We're ready to negotiate at any point." It is, in fact, a repudiation of the idea that there's any point in negotiating at all. In World War II, for example, the Nazis try to initiate a round of negotiations near the end of the war, and they are quite firmly told no: it's unconditional surrender or nothing. To me, that's the distinction between war in the Clausewitzian sense, violence intended to compel our opponent to fulfill our will, and war as a side dish, a way of putting emphasis on the negotiations you're actually conducting.
I think there will just be more and more people who feel that violence is something they can add into negotiations, but the final result will still be a negotiation. And if states really are willing to take whatever pain an adversary dishes out, Iran being a really good example right now, then there's no amount of violence you can inflict on them that will bring them to the table. What's happening now in the Persian Gulf is a very interesting exhibit of what it looks like when one side thinks war is a side order, something you use to bring people to the table, and the other side sees war as an end in itself. Those two sides are never going to come to a conclusion, because neither is willing to quit: one because there's no negotiation coming, and the other because they're still capable of resisting. It's not a good situation to be in.
Fineas When you look at proxy warfare, drones, and gray-zone conflict together, do you think the traditional idea of war as something that happens between two sovereign states is becoming obsolete? Or are these actually very old forms of conflict appearing with new technologies and new terminology?
Dylan Craig I think it's really important to remind ourselves that, through human history, and this is what my book is about, war has very rarely been between two sovereign states. We have this idea mostly because we're fresh out of the 20th century, and the rules of sovereign war have come to mean uniforms, formal militaries, rules of engagement, the Geneva protocols. Historically, that's a weird, idiosyncratic way war played out for maybe 50 years in one part of the world. At exactly the same time, countries were fighting brutal counterinsurgent, postcolonial wars in Africa, in Asia, in South America, in which they weren't doing any of that. And just before and after that, they were fighting wars against non-states, in the form of communist revolutionaries or Reaganite revolutionaries in Nicaragua.
If you look at the history of war, the slice represented by states fighting states under agreed rules of engagement is small. So when I see the non-state triumphantly reinsert itself onto the battlefield using drones, I don't see a scary new future. I see a reversion to a scary past. And I do mean scary. I don't think it's necessarily going to be a good thing that all these people we managed to banish for the longest century you can imagine are back. It's not going to be good for the people who have to live amidst their warfare. We got rid of those folks for a reason. But it's not going to be anything we haven't seen before.
Louis XIV, of the final argument of kings: just look at the kinds of non-states he was willing to partner with, from religious factions to pirates to North American Indigenous peoples. For him, the fact that war belonged to the state didn't mean the only people on the battlefield were going to be guys in French uniforms. He didn't have a problem squaring those. And I imagine in the future we won't have any difficulty squaring the existence of contractors, tribal militias, and state-sponsored proxies alongside our own troops. We'll find there's actually a long precedent for that in human history. What's going to become harder and harder to imagine is the century, that brief little window, in which we managed to ban those people from the battlefield.
Fineas You've also worked on using mapping as a tool for conflict prediction. How good are we at actually predicting where wars or political violence are going to happen?
Dylan Craig It depends on how fine-grained you want to be. We have some really good scholarship, for example Tanisha Fazal's book on all the states that aren't with us anymore, all the states that have been exterminated throughout human history. A statistically very significant number of them had the misfortune to be stuck between two other states. So if you're buying real estate as a state, the worst piece of real estate to buy is in between two people who have beef. At that very broad scale, we can be very clear about where it's dangerous to be. The challenge comes with getting more and more fine-grained.
It's easy, for example, to look at the Strait of Hormuz or the Strait of Malacca, various geographic choke points, and say: these are places where a lot of conflict is going to break out. That is and isn't true. The Strait of Hormuz is certainly really important for our current global economy. But it's important because it's expensive to circumvent. Going around Africa isn't impossible, it's just more expensive. So when we treat it as impossible, what we're really saying is that it would not be financially viable for the people who currently ship oil to take 2% or 4% less profit because they have to spend more on fuel and wages going around Africa. In other words, the degree to which some part of the world is closely linked to future violence is not eternal. It's a reflection of what our global economy and technology look like right now, and it could change tomorrow.
There are all sorts of parts of the world we've spent so much blood fighting over that economics has left behind. France and Germany don't have to fight over Alsace-Lorraine anymore, even though they fought over it plenty of times. It's been rendered irrelevant by the march of technology. So at the big level: don't get stuck between two fighting states. We're very clear on that. At a kind of meso level somewhere in the middle: avoid being stuck on real estate that's economically critical. But if you do end up in an area like that, have some hope. In the future, technology may change the degree to which that place is a shooting gallery.
When we get down to, say, a hundred square kilometers of the border between the DRC and Rwanda, we really start losing the ability to figure out where conflict is going to take place. The reason is that once you get down to the human scale, we have to grapple with the fact that the places where fighting happens are not necessarily the places that produce the violence. Imagine a civil war where the rebels hide in the mountains and come down to the lowlands to fight the government. The shooting is happening in the lowlands. Does that mean the lowlands are producing the violence? Or the mountains?
There's a scholar called Stathis Kalyvas who looked at the Greek Civil War and plotted the altitude of villages against the number of people being denounced as spies, either for the government or for the communists. He produced a really interesting graph, not because altitude literally makes people more or less interested in supporting the government, but because altitude becomes a proxy for how far away the government is. Up in the really high mountain villages, the government is far away. That's rebel territory. So if there are a lot of people being denounced as communist spies in the lowlands, it's not because that's where all the communist spies are. They're up in the mountains. What's happening in the lowlands is that there's enough government support that people feel confident that if they betray someone as a communist, the communists won't get them tomorrow.
This is the challenge we run into at the human scale, and it's why I think it's really important to keep working on our explanations of why conflict happens. Violence doesn't just happen because bad people do bad things. It happens for reasons that can go quite far back in history, and that are usually more subtle than you think. If we're not keyed into those reasons, we'll look at a map of altitude and people being lynched in the night as spies and not understand what it means. The danger is we'll say altitude causes violence. No, it doesn't. Altitude is correlated with something else that causes violence, and now we need to find that thing. That's the ongoing chore, if anyone wants to study security studies.
Because if I'm right that we're going to get more kinds of actors using violence than ever before, because the bar has dropped, then all of our old explanations of why violence happens will need new security scholars, folks in your generation, to answer new questions. When do tech startups go to war? None do yet, because tech startups are new. But at some point they might, and we'll need an explanation of what motivates them, the same way we currently have books and books about why states go to war, why terrorist organizations and criminal cartels use violence. We're going to have different kinds of actors on the battlefield of the future, and we need to know what makes them tick.
Fineas You touched on some of the warning signs you look for in predicting major conflicts. What are some other warning signs that a major conflict is about to start, and is there anywhere in the world today where you see several of them coming together?
Dylan Craig Warning signs colliding. Whoo boy. If I had a really good answer as to where the next conflicts will break out, I'd be downtown in D.C. selling that information right now. I don't have a crystal ball, but there are a couple of trends I think are concerning. A lot of the security we've counted on historically has come about as the result of a commitment to collective defense: the sense that in order to defend a state and its interests, you have to defend all of its people. It's right there in the U.S. founding documents. What does government exist to do? One of the things is to provide for the common defense. Not the defense of New England, not the defense of people who vote one way or another, but everyone together.
I think there are a variety of challenges right now making it harder and harder for people in countries around the world to feel like there even is an "us" worth defending, a political collective we should be willing to die to protect. We've managed to skate along. Maybe a better metaphor is Wile E. Coyote running off a cliff: it takes him a while to look down and realize there's no ground underneath him anymore. And I think we've been in midair for quite a long time.
Take the issue of collective defense becoming an increasingly hollow concept. The U.S. used to have a very strict set of legislative hoops administrations had to jump through in order to conduct violence abroad. There's the well-worn phrase: Congress has the power of the purse. The idea is to tie the government's war-fighting ability to the will of the people. Why? Because the people are smarter about war? No, because the people are going to suffer the effects of war. Their kids are going to come home in flag-draped coffins. Their 401(k) is going to tank. So you should consult them before you go to war, because they have a better sense of what war is going to cost than you do.
That's breaking down across the board. As we get more technologically sophisticated ways of fighting, it becomes less and less necessary to have society behind you before you go to war. We don't need a draft anymore. We don't need a million dudes with M1 Garands. If you gave any of our commanders 100,000 conscripts, they wouldn't know what to do with them. We don't need legions of guys doing Saving Private Ryan anymore. We need smart people who can use an Xbox controller to fly a drone. We need people with electrical engineering degrees. On the one hand, great: we no longer need to ask every single person in the U.S. to equally share the burden of defense. But that has made it easier and easier to fight wars without asking people to bear the collective defense.
Drones are going to make it worse. Cyber is going to make it worse. And the more unequal we allow society to get, the easier it's going to be to have a smaller and smaller group of people you need to get on board before you can use violence. That's not going to stop violence from happening. There's still going to be plenty of it. The problem is that it's no longer going to be violence that even has to pretend it's for everyone's good. It will be explicitly the kind of thing the American founding fathers thought democracy would get us out of. When they looked at war, they said the reason war happens is that princes get to go to war whenever they want and don't have to deal with the consequences. So let's have democracy instead. Let's have a constitutional republic in which the government has to ask the people.
We've spent the last 250 years rolling that back. I get why we did it, and it's certainly exciting to have all these new tools and tricks. But we should be very wary of where that road ends, because we've been there once before and decided we didn't like it. Just because shiny tech toys now give us the ability to not ask the people doesn't mean we should stop. And I think that's true not just of warfare but of basically anything you see outside your window. We can't just say to the parts of society we no longer see as productive, or as "real America": bye, we're going to be over here doing this, and we don't need you anymore. Whether we need them isn't the issue. The issue is how much of a fall you're setting yourself up for by allowing "us" to get smaller and smaller. Where does that road end? What kinds of problems will you find yourself in once you no longer need to sell your idea to everyone, and only need to sell it to, I don't know, Elon Musk?
There's a funny little saying in Afrikaans, which I'll treat you to. In Afrikaans, the word for "us" and the word for "ounce" sound the same: ons. So when you say in Afrikaans, "How many ounces are there in your pound?" on one level you're asking about ounces, and there's always the same number. But really what you're asking is: how many of us are there in your "us"? When you say "us," are you counting me? Or do you not consider me part of whatever it is you're part of? I think the next generation is going to have to come to terms, in a series of difficult ways, with how many ounces are in our pound. Just because we can do it without those people doesn't mean we should be excited about it. Doing without them will bring bad stuff down the road, definitely in the military case. I'm a little out over my skis when it comes to the rest of social science, but we did agree that war is the continuation of politics by other means. If something's true on the battlefield, why wouldn't it be true on Main Street?
Fineas With the incredible amount of data now available, and increasingly powerful AI, do you think we'll eventually become significantly better at predicting, and maybe even preventing, wars? Or is human behavior too unpredictable?
Dylan Craig I don't think human behavior is unpredictable, especially in the aggregate. Every time we get on the road, we're successfully predicting that 99.9% of people will drive on the correct side. We're very good at predicting behavior when we have a clear sense of where it comes from. Why don't people drive on the wrong side of the road? Because there are legal penalties, and damage they might suffer if they crash into me. But look at the degree to which people's increasing fondness for enormous F-250s and F-350s is ratcheting up road and pedestrian fatalities. The minute you change how understandable a system is and introduce a new variable into people's thinking, predictability starts to break down. If some part of the population feels they don't have to worry about crashes, because they think their onboard driving system or their huge truck will keep them safe, that changes what people are going to do.
So even though AI can crunch a lot more data, that doesn't get us off the hook for trying to understand why things happen. AI will be very good at identifying patterns of what happens. It is very unlikely to be able to tell us why those things happen. There's a guy called Tyler Vigen who has a website called Spurious Correlations, and it's basically him doing what even a moderately good AI could do: finding things that are apparently linked by correlation but have nothing to do with each other in terms of causation. The number of street signs in El Paso, Texas, against the number of children drowning in swimming pools somewhere else, and the lines match perfectly. AI is going to make it increasingly easy to grab two data series like that and see they match one for one. But unless we have the mental discipline to stop and ask why, and whether those two things could really be causally connected, we're in trouble.
We're going to have to hold AI on a tight leash, because it's going to speed up the degree to which people can be confidently wrong about things. And it's going to seem like weird sorcery to people who don't know how LLMs work. As with all technology, there's going to be a period in which a bunch of hucksters and snake-oil salesmen convince people who don't know better that they've found the secret to the universe in their branded little chunk of LLM. That's what I see among my students. Some of them are really technologically smart and really understand what AI platforms can achieve in the security sector. What terrifies them is not AI. It's what one of my students described last year. He was put in charge of a task force on AI that was responsible for generating a policy, and he and the other two people on that task force had never worked with AI before. The boss basically looked around and said: you, you, and you, you're now in charge of deciding how we're going to use AI in this branch of the military.
These guys looked left and looked right and said, "Us? You're asking us?" But the fact is, people are going to be put in positions where they don't yet have those critical filters, the ability to tell bad data from good data, and they will make rash and bad decisions based on the outputs of very obedient, very smart-seeming systems. There's the long-held phrase in computer science: garbage in, garbage out. It's true of computer programs, true of websites, and AI is just the most recent iteration. If we're not smart about what we ask AI to do, and smart in how we read what it spits out, we will fail. AI will not fail. We will make dumb decisions based on what AI served us, and that will be on us, not on AI.
I'll give you an example. A student came to me chuckling and said, "Ha, I used AI to write my final paper, and here it is." I read through it and said: don't you find it interesting that this paper, which is about South Africa, we could do a find-and-replace on, take the words "South Africa" out and replace them with any country in the world, and it would still make grammatical sense? In fact, it would make logical sense too. Notice how the AI is saying a lot of words but isn't sticking its neck out. It isn't making any definitive claims where it could be proven wrong, because that's not a smart move for AI. It doesn't hang a target out there so you can say, "That's not right." It hangs up these relatively non-confrontational, relatively shallow interpretations of the facts, in which you could replace South Africa with Ecuador and still get a paragraph that sounds nice and plausible.
I don't think AI is going to change security. If anything, it's going to empower some of the tendencies we need to struggle against at all times in security thinking. There's going to be a couple of years in which people who don't know how AI works treat it like a magic pill you take that produces great strategy. When has that ever been the case? It wasn't the case when computers showed up. It wasn't the case when air power came into existence. New technology requires smart brains to drive it. It's not going to drive our smart brains, as a lot of people will figure out. And I think AI skepticism in your generation is generally pointed in the right direction. You're not boomers or Gen Xers asking, "How did the computer do that magic?" It's just something normal to you. So I don't think it's going to cost your generation as much as it has cost older ones.
Fineas Are there any national security threats that Americans give too much attention to? And what national security threats do you think we don't give the proper attention to?
Dylan Craig This is tricky, because I could say, for example, that Americans are far too scared of the loss of American lives. Currently, all it takes to draw the U.S. into or out of a conflict is to kill a few dozen Americans. There are 340 million of us. It shouldn't be possible to completely reorient U.S. foreign policy by killing 12 of us. A multiple of that number die every day from things like slipping off a stepladder, and we don't do anything about that. I think it's equally ridiculous to think we should go to war because 12 Americans have died as to think we should leave a war we're already in because 12 Americans have died. That's the wrong question to be asking.
Obviously, every dead person is a tragedy, someone who cannot be replaced for their family and their community. That's true. But we're not trying to decide how their community or family should respond. We're trying to figure out how the resources of our collective defense should be reoriented around events whose main impact on our society is that there are 12 fewer of us than there used to be. I can say that coldbloodedly as a security scholar with relatively little skin in the game. I would probably change my tune very quickly if one of my family members were one of the 12.
On the other hand, something we're relatively unconcerned by that maybe we should be very concerned about is climate change. A stat I encountered recently, from one of our guest speakers at the school, really put me in mind of something important. I'm not an aeronautical engineer, so I'm going to mess the details up a bit. The speaker showed us a graph of airplane engine efficiency relative to temperature. Practically speaking, what you and I and your viewers need to understand is that the hotter it gets, the less efficiently airplane engines burn, and the more fuel they need. The speaker's point was that if global temperatures increase by a couple of degrees, and again I've forgotten the exact numbers, there will be something like a 10% drop in the efficiency of aircraft engines.
That doesn't sound bad until you realize it means we need 10% more helicopters to do what our helicopter force does now, and 10% more fuel to keep our Air Force in the sky. Have you ever seen footage of how much fuel U.S. forces in Afghanistan used to consume every day, going through those mountain passes from Pakistan? It's an astonishing amount. Ten percent more is a big, big deal. The idea that climate change will hurt us primarily by flooding our seaside properties needs to be weighed against the fact that climate change is going to radically resculpt which technologies are useful for maintaining national security. A lot of what we currently base our national security on is big, dumb, fancy machines that burn a lot of fuel, because fuel has historically been very cheap. We're not going to get as much bang for our buck as we do now. If more people thought about climate change in terms of effectively shrinking our helicopter force by 10%, I think they'd see it as more of a threat than they currently do.
So: probably too much attention put on lives lost, and probably not enough thought put into helicopter engine efficiency.
Fineas That's really interesting. I'd never really thought of climate change as a national security issue beyond the civilian effects.
Dylan Craig It's subtle, and it's a really good example of how often the really important things driving international phenomena aren't immediately visible on a common-sense basis. You or I would not know that thing about helicopter engine efficiency. You'd need to be an aeronautical engineer. That's what should help you understand your job as someone interested in national security. It's not just sitting back and having opinions. Anyone can do that. You've got to become smart on something the rest of us aren't smart on, so that you can be the one ringing the alarm bell about helicopter engines.
Fineas Last question. For a high school student who knows they want to work in international affairs, diplomacy, or national security someday, what should that person be doing now that would actually prepare them for that kind of career?
Dylan Craig That's a perfect segue from what we just discussed. Number one: start asking what value you'd like to add to a national security job, and start cultivating that skill set. I told you about my teaching assistant from a couple of years ago, Ernesto. He ended up getting a really nice opportunity at the Department of Homeland Security because he'd taken a video production course once as an undergraduate. What that equipped him to do was look at social media videos and figure out whether they'd been edited when DHS was vetting people. Was he the world's best videographer? No. Was he the world's best security thinker? Also no. But he was the best person they could find for that job.
Every single high school student has the ability to bring something: video production, the ability to speak French. It makes it so much easier to get where you want to go if you think of yourself as a once-in-a-lifetime combination of skills. Maybe you're not as good as the best people at any one of those skills, but you're going to be the one who brings that combination. It's a kind of weird math effect. When you shuffle a deck of cards and deal it out, you're dealing a sequence that has probably never been dealt before and will never be dealt again in the history of the universe. That's how big the number is when you multiply 52 by 51 by 50 by 49 and so on.
Thinking about skills that way, that you may not be the best person at a set of five skills, but you may be the very first person to combine those five skills that your employer has ever encountered, takes a lot of the fear out of preparing yourself. It means you don't have to be the best, but you do have to be distinctive, and you need a story about how your various skills, interests, and passions work together to equip you to go in a certain direction. Don't feel like the solution to getting into international security is doing fewer things better. What you might have to do is more interesting things you genuinely care about, and then spend some time thinking about how you could use them to add value to a national security profession.
The other thing is to work on focusing your mind. One of the reasons I was so pleased to get this interview request from you is that I recommended you folks do something a lot like this: stick your neck out, have opinions in public, meet people and discuss things with them, produce a body of work. To go right back to the question we started with, about what's new and different about teaching national security professionals: don't wait until you're 45 or 50, with sleeve tattoos and 20 years in a rucksack, to learn how to have an opinion in public without freaking out. Because the sooner you have that skill, the sooner you'll start talking your way into being in the room where it happens, to quote Hamilton.
So: diversify your skill set. Think of yourself as a one-of-a-kind combination, and ask yourself how that combination could work together. And lastly, get used to stepping up and getting your ideas out there. Yes, people are going to say mean things about your ideas, and people are going to laugh at you and call you cringe. It really doesn't matter, because eventually you're going to have to do this, and you're either going to learn how to do it at 16, or at 46 in my classroom at the National War College. Follow that, because everything you want is on the other side of having the bravery to have an opinion about the world and back it up.