Irregular Warfare Podcast - Drones, Automation, and how ARSOF is Adapting

Episode Date: May 31, 2024

Episode 106 of the Irregular Warfare Podcast examines how drones, robotics, and automation are changing the battlefield in Ukraine and how ARSOF has adapted.  Our guests begin by outlining the signif...icance of robotics and autonomous systems on the Ukrainian battlefield. From there, they delve into how US Special Operations formations are learning from Ukraine and changing their own formations. Finally, they end with a deep dive on how SOF forces have used drones in the past and how that compares to the modern fight and look at what changes SOF, industry and the US government may have to make to keep pace with the changing character of war.

Transcript
Discussion (0)
Starting point is 00:00:00 Hey there, this is Shauna Sinnott, co-founder of the Irregular Warfare Initiative and the current chair of the Board of Advisors. And I'm Kyle Atwell, also co-founder of IWI and the current IWI chair of the board. Before getting into today's conversation, we wanted to take a minute to announce that this episode marks the four-year anniversary of IWI. Since we released the first episode in May 2020, IWI has grown significantly. We're extremely proud of what IWI has contributed to the national security community during this time, and we're even more excited for what is to come. What began as a podcast with the mission of bridging the gap between scholars and practitioners to support the community of irregular warfare professionals has now become the Irregular
Starting point is 00:00:40 Warfare Initiative, an incorporated nonprofit organization which is working daily to accomplish three strategic objectives, building the IW community, driving professional public dialogue on irregular warfare, and investing in irregular warfare professionals. The IWI team has accomplished a lot over the past four years. This includes releasing 106 podcast episodes with more than 1.3 million downloads, publishing almost 170 articles, hosting 20 events, and building three cohorts of IWI fellows. We've established eight special projects focused on topics such as maritime irregular warfare, cyberspace, air power, proxy warfare, and the role of SOF in strategic competition. IWI has become a community that includes tens of thousands of
Starting point is 00:01:25 listeners, social media followers, and newsletter subscribers, including you. And IWI is just getting started. More than 60 committed volunteers are working hard to create robust networking opportunities for the irregular warfare community, to establish additional platforms to drive public dialogue, and to generate opportunities to invest in talented or regular warfare professionals. There are many ways you can contribute to the IWI mission. First, you can join the team. Reach out if you are interested in becoming a volunteer or leader within IWI. Our team is made up of scholars and practitioners dedicated to making an impact. Second, you can create content by submitting an article or engaging as a podcast or special event guest. Last and very importantly, you can donate or purchase swag from the IWI store.
Starting point is 00:02:10 Generous financial support enables the IWI vision to become a reality. Thank you for your interest in Irregular Warfare and for being part of this amazing community. We're honored to provide this resource to the community of Irregular Warfare professionals. Now, without further ado, we're pleased to present today's very interesting episode of the Irregular Warfare Podcast. And so when we look at that skill set, arguably the utility of being able to hit targets with a direct fire weapon at those sorts of ranges is becoming one where the cost-benefit analysis of committing the training time to that is just not there.
Starting point is 00:02:45 That is something that the institution finds very uncomfortable. With the loss of the Russian fleet in the Black Sea, and Ukraine doesn't have a navy. That's multi-domain at work. That's what the army says we want to do. Now I'm a former resident of the First Island chain. I am very interested in how land forces can destroy naval assets. That is a growth industry, in my opinion. Welcome to the Irregular Warfare Podcast. I'm your host, Matt Mullering, and today my co-host is Alyssa Laffer. In today's episode, we do a deep dive into drones, robotics, autonomous systems, and how they are influencing the USSOC force structure. Our guests begin by outlining the significance of robotics and autonomous systems on the Ukrainian
Starting point is 00:03:23 battlefield. From there, they delve into how U.S. special operations formations are learning from Ukraine and changing their own formations. Finally, they end with a deep dive on how U.S. soft forces have used drones in the past and how that compares to the modern fight and look at what changes soft, industry, and the U.S. government may have to make to keep pace with the changing character of war. Dr. Jack Watling is a senior research fellow for land warfare at the Royal United Services Institute. He works closely with the British military on the development of concepts of
Starting point is 00:03:56 operation, assessments of the future operating environment, and conducts operational analysis of contemporary conflicts. Dr. Watling has worked extensively on Ukraine, Iraq, Yemen, Mali, Rwanda, and further afield, and has written several books including his most recent publication, The Arms of the Future, Technology in Close Combat in the 21st Century. Lieutenant Colonel Eric Davis is an active duty army officer with over 16 years of experience in special operations. He's also a General Wayne A. Downing Scholar with master's degrees from both King's College London and the London School of Economics.
Starting point is 00:04:30 His assignments have taken him from village stability operations in rural Afghanistan to preparing for high-end conflict in the First Island chain. You are listening to the Irregular Warfare Podcast, a joint production of the Princeton Empirical Studies of Conflict Project and the Modern War Institute at West Point, dedicated to bridging the gap between scholars and practitioners
Starting point is 00:04:48 to support the community of irregular warfare professionals. Here's our conversation with Dr. Jack Watling and Lieutenant Colonel Eric Davis. Jack, Eric, thank you for joining us on the Irregular Warfare Podcast. Good to be back with you, Matthew. And thank you again, Jack, Eric, thank you for joining us on the Irregular Warfare podcast. Good to be back with you, Matthew. And thank you again, Jack, for joining us. Really looking forward to this conversation. Jack, your firsthand experience in Ukraine helps us understand just how difficult it is to deploy robotic and autonomous systems in military operations.
Starting point is 00:05:19 Additionally, you recently took part in the U.S. Army's Project Convergence experiment where you saw the U.S. approach to integrated robotics. Can you give us a synopsis on the current state of how you see robotics and autonomous systems changing the character of work? What we observe is that autonomous and robotic systems are very often a significant force multiplier in carrying out specific tasks. But they also come with a significant tail. They're not particularly flexible tools and so if you have them organic to the unit they tend to increase its tail and reduce its tempo and so where we see them have the greatest effect is when they are brought forward by specialist users as an attachment to a unit to overcome a problem for which they are optimally designed.
Starting point is 00:06:06 And then once that problem is overcome, the autonomous and robotic systems break away and the unit can exploit through, which means that we need to be training in a way that allows those specialists to understand how to attach to the planning process that that unit has and detach from it. And we also need the unit that is being supported by those systems to understand how to attach to the planning process that that unit has and detach from it. And we also need the unit that is being supported by those systems to understand how to receive that attachment and then to break free. When you try and make them something that a company commander, say, has to think about all of the time, you end up having too few specialists organic with the unit to be able to maintain those systems.
Starting point is 00:06:46 And so that becomes a vulnerability of the company. Great. Thanks for that intro, Jack. And I would pitch it back over to you, Eric, to ask, as someone with a lot of operational experience, how do you perceive the integration of robotic and autonomous systems in soft? Do you agree with Eric's take? Have you seen this long tail behind employing autonomous systems or have you seen something different? So in a lot of ways, we've seen a very similar tail experience
Starting point is 00:07:13 even with our existing drones that we have inside the United States Army. In a lot of ways, this isn't new though. We've been talking about particularly quadcopter, 3D printed, innovated drones since at least 2014. I've been part of those conversations. But the technology, it's moving really fast. It's not fully mature. And I don't know that it's going to become fully mature in the next year. I think this is going to continue to be a space where
Starting point is 00:07:32 innovation happens faster than most of the traditional processes are capable of keeping up with. Innovation happens faster than most of the traditional processes are capable of keeping up with. Within USISOC, United States Army Special Operations Command, what we're trying to do right now is take, we've got a bunch of ground up efforts. We've got people all over the regiment and the range regiment and the SF regiment, civil affairs, PSYOPs, who are all trying to do stuff with this. And so what we're looking to do is give that a clearinghouse, a place for people to send the good ideas, the lessons learned up, and then also provide the full support and the
Starting point is 00:08:04 budget of the headquarters behind it. So that's, I think, our biggest change. We've established the Remote Unmanned Systems Integration course, or RUSIC, at our schoolhouse. Starting this year, a proportion of all the Green Berets graduating the Q course are going to attend that training. We're also standing up our own AI division, and we're also trying to integrate both of those elements because we see that this is a software solution to a piece of hardware. In a lot of when you put those two together that's when you see real effects we have naysayers we have people who say we can farm the ai and we can farm the drones out to contractors or to production stateside but in our software we understand that that doesn't work for us like we are going to be forward we're going to be in the deep in a fight where we need to be able to
Starting point is 00:08:40 do these capabilities at the edge rather than waiting to reach back waiting to be able to get something shipped over it has to be able to be innovated and adapted on the ground. And I think that was something that Jack's article really hammered for us was the need for this to be innovated at the local level and have the ground up support to be able to give feedback and make adjustments. And the last thing, just so this is not a unique to the Army, the Navy's figured it out too. They've got a new career pathway for robotics with a rate. And so we are looking to do the same thing. We're actually trying to develop an MOS, which gets at Jack's point. This can't just be an extra skillset. It's going to have to be
Starting point is 00:09:10 ultimately somebody who knows this and understands the intricacies of this. And so we're trying to get a new military operational specialty or MOS established because when it comes down to it, I'm not going to give all that talent to the Navy. There's a job for Army, for all those high schoolers that are playing with 3D printers and drones. And I want them to come work for the United States Army. All really good points. Bringing back to your article, Jack, you mentioned this great concept of irreducible minimums and the recent article you wrote on War on the Rocks about drones. Can you elaborate on that concept for our listeners and talk about how it applies to the war in Ukraine? Yeah. So if we look at how UAVs are being used in Ukraine at
Starting point is 00:09:45 the moment, let's say you have an ISR UAV that is being launched to get into the enemy deep. That is something that theoretically could be completely automated, right? You have an autonomous vehicle that is a launcher. The UAV flies up. It goes over the area that it needs to look at. It offboards that data autonomously. It comes back, it lands, etc. But the reality is that there are a whole range of tasks around that that require individuals to make judgments. You need someone to liaise with the commander to actually get an understanding from the intelligence picture, but also from the commander's intent. Where should I prioritize looking? And that is something that requires a human to be in communication. Sometimes those parameters
Starting point is 00:10:29 will shift during the mission. And so even if the flying of the UAV is quite automated, you still need a pilot essentially who is able to make those injects with the system. You also invariably need somebody who is going to manage what's coming off that UAV. If anything, it's just for validation. The number of times where I've seen UAVs detect a target but misreport the coordinates of where that detection is because they are being interfered with through electronic warfare, that is ultimately discovered by humans. You can create technical systems to see whether that's happening. But as soon as the adversary understands what those technical systems actually monitor, they can refine their EW to evade that detection. And so the risk if you fully automate the system
Starting point is 00:11:16 is that you end up being deceived at scale and it becomes very difficult to problem solve to identify where that is occurring. And so you need a sensor operator, you need someone monitoring even if they are not sitting in the decision loop. And then the unit needs some force protection. Ultimately, it's a battlefield, vehicles get stuck, but you need people there who are able to do basic tasks, lifting, setting things up, making sure that the runway is clear, etc., if it's a fixed swing UAS and needs some space. And so you can't buy out that. And actually trying to buy it out leads to a far less efficient and far less adaptable system. And therefore,
Starting point is 00:11:56 it becomes one that has a very limited shelf life, because as soon as the adversary understands it, they can start gaming the limitations of that system. Whereas if you have operators, then you're able to adapt very quickly to those problem sets. And that leads you to irreducible minimums, right? You just need a certain number of people. If you have to have a pilot and a sensor operator, you can do the math on how many UAVs you have, how long their endurance is, how many orbits you need to maintain coverage, and therefore how many pilots you need to do the monitoring before they're exhausted. You need some people in that system. So identifying the irreducible minimums is important because if you start cutting below that, you either lose flexibility and resilience or you lose endurance.
Starting point is 00:12:44 And that becomes something that the enemy can quite quickly overcome. That's great. So you both have started to touch a little bit on changes to the force and force structure that might be needed to make the most out of UAS in the battlefield. I'm curious for both of you, whether you've seen already any changes that Ukraine has made to its organization and its TTPS as well as the Russians. Have they adapted to these between units that have dedicated specialists on these systems and units that don't. Where you have units that are just being given equipment and trying to work out how to use it, you get efficiency rates of maybe 20%. You get high attrition rate on systems. attrition rate on systems, where you have dedicated UAV units, for example, and of course, they can force generate packets to support other elements. But nonetheless, it is a unit with its own command and control structure. Not only do you get collaboration between different kinds of UAVs,
Starting point is 00:13:55 but you also get a much higher, there are several drivers of that. Partly it's about pilot skill. I'll give you an example. I was talking to a Ukrainian UAV operator who described what he did when the video link that he was using to fly the aircraft was severed by EW. And he said, well, it's quite simple. I turned the aircraft 180 degrees and I reversed the timing and the inputs that I made on the controls exactly. So I fly it back to myself blind along the same route. That requires a lot of expertise and experience to be able to do it. But that means that pilot can get their UAV back even under some pretty intense CW effect. You also need the back end. You need the access to mission
Starting point is 00:14:36 planning data in terms of electromagnetic spectral surveys. You need access to weather, meteorological data, enemy air defense, what the intelligence tells you, and that allows you to route plan and flight plan. And so again, you need a dedicated formation in order to be able to integrate all of that data at the speed of relevance. Whereas if you just have units that are trying to put these things up, then they will end up pushing them up in environments where it is highly adverse for those to fly, and therefore you will lose them. I would just like, I could take the moment to just thank Jack for the fact that I know the lengths that he's gone and the distances he's traveled in Ukraine to get this research. I really want to say thank you again for that. Not just back in the UK at RUCI, sitting in a think tank, you're actually going all the way out there and getting some great lessons.
Starting point is 00:15:22 And we appreciate you bringing those back. in a think tank, you're actually going all the way out there and getting some great lessons. And we appreciate you bringing those back. To go broad a little bit. So to bastardize an F. Scott Fitzgerald quote, the test of a first rate Green Beret is our ability to hold two different styles of warfare in our mind at the same time and still function. So we advise and assist conventional forces through security force assistance. That's your combined arms maneuver warfare fight. And we have to be able to do that all over the globe. At the same time, we also advise and assist a regular warfare forces, typically in unconventional warfare. And the central variable in both of those is the partner force. So whatever advice and assistance we're giving has got to be informed by the culture and
Starting point is 00:15:52 the capabilities of our partners. So what does that mean? Well, that means when it comes to drones, we need to be able to understand both the approaches of how these are being used in a conventional, larger, high-scale fight, but also how they're being employed in an irregular warfare capacity. And then we need to understand how the partner force can acquire and adjust the systems. And so that's where some of that RUSIC, that schoolhouse bringing things back together is some of the stuff that we're working on capturing lessons learned and TTPs isn't going to come through the traditional army doctrine because some of the stuff that our partners are going to use are things that we cannot purchase because of various policies in the United States. We can't buy DGI Mavics, even though this is the
Starting point is 00:16:28 number one platform that everybody's innovating off of. And so I can't go to a partner force and say, hey, I don't know anything about that. I'm sorry, I can't. No, we have to understand how to use those and how to adapt those. So that's probably the biggest change is understanding how is this impacting our partner force and whether those are a regular warfare or regular warfare. And to briefly hit it, those are not mutually exclusive. We have a tendency to think that one stops, the other starts. You see it on the line in Ukraine today. The regular warfare forces and the conventional forces are integrating to fight against the Russian aggression. Jack and Justin did a great job in their article talking about the close and the deep and the integration of
Starting point is 00:16:59 those two components. And they're both using these new tools in new and innovative ways. And that means we've got to adjust our doctrine. That includes the new specialty, but that's why we also talked about this needs to be a separate MOS at the end of the day, more than just a skill set like having a sniper on a team. This is something we're looking at. We haven't finalized yet, but we're looking at changing the standard operational attachment alpha.
Starting point is 00:17:21 What does that package look like when it needs a skill set? Is that skill set a skill set that I need all the time? Or is it something similar to what we did with the special operations, the SAUD-A's, where it was a SIGINT capability that we would attach? They had a higher level of capability. They're trained in the soft formations, but they're not part of the regular detachment. So we're working through that space right now, but it's unfortunately, we're figuring it out right now. And admittedly, again, things like Jack coming back and helping us understand how Ukraine is innovating and how they're determining what is the right package that goes forward and where has been really impactful for us. So, Jack, in your article, you gave the example of MQ-9 having increased the cognitive load of data analytics and intelligence processing teams, which has ultimately led to an expansion of the manpower required for a single MQ-9 mission.
Starting point is 00:18:07 Despite you breaking this down very clearly, there are still a lot of senior leaders that seem to believe that AI and automation will actually decrease the processing requirements and decrease the personnel requirements as well. What would you say is causing this misunderstanding and this disconnect? I think it's a willful blindness. I think it is because our defense establishment is facing recruitment challenges and also facing budgetary pressures. And as we try and increase the skill sets of the force, and often there is a significant institutional inertia that means we're not prepared to remove previous layers of training to bring in the new things.
Starting point is 00:18:46 And so getting someone to a basic deployable standard simply becomes more and more costly and time consuming. Therefore, policymakers want to be able to say that their investment in technology will reduce their force requirements. It's an attractive argument. It is an argument which wins them budget for the kind of modernization and transformation that they know is needed, but it's also wrong. And I mean, we have to qualify that, right? There are areas where automation will increase efficiency, reduce the number of people that are required for a specific task. But the two things to bear in mind is that unlike in civilian industry, you have an adversary that is actively trying to destroy your systems using the same efficient tools that you were
Starting point is 00:19:30 employing. And therefore, that creates a dynamic where you need resilience, which means that mass becomes relevant. But there's also simply the fact that using the predator point and the reaper point, when you start putting these high-end sensors on the battlefield, you start generating opportunities that didn't previously exist. And if you are to be able to exploit those opportunities, whether it be increasing the fidelity of intelligence, whether it be increasing the lethality of units, then you need people. you need people. And so very often what you end up doing when you integrate modern technologies into formations is displacing your mass rather than eliminating it. If I could, just one thing that Jack said that I really wanted to hammer on is like he mentioned it, we have a recruiting challenge. However, we know we have a bunch of Americans right now, high school kids all over America that are playing with 3D printers, playing with drones, playing with coding, playing with AI. So we have a recruiting challenge. We also have a recruiting opportunity,
Starting point is 00:20:28 which is to shift our idea of who's our target recruit and how can we employ them. And that's one of the places where SOF has an advantage in the sense that I can bring in an 18 x-ray straight off the street, train him up and turn him into a Green Beret in the space of two years. We do the same thing with our psychological operations. And I think in particular, when we talk about some of the AI stuff, if you've got a high school kid playing around with AI and he's interested in psychological operations, there's a pathway to get him direct in. So as opportunity, we just have to maybe change who we're looking at for our recruiting. And just to grab the institutional inertia point, because I think this speaks to the
Starting point is 00:20:59 changing composition of an alpha team. If we look at sniping as a skill set, high prestige capability requires a huge amount of training in the modern force. Now, there are a lot of elements that make up a sniper that are going to retain their relevance in terms of the intelligence, surveillance, reconnaissance function, communications function. However, when it comes to the shooting element, think how long it takes to get somebody to be able to make a hit at 1200 to 1800 meters, and the assurance, the reliability with which they'll be able to make those hits, versus somebody who can very reliably hit a target, even a protected target at 2000 meters from a concealed position if they're using an FPV,
Starting point is 00:21:41 and how long it actually takes to teach them how to do that. It's not that long. And so when we look at that skill set, arguably the utility of being able to hit targets with a direct fire weapon at those sorts of ranges is becoming one where the cost benefit analysis of committing the training time to that is just not there. That is something that the institution finds very uncomfortable because people's identity and their prestige is tied up in having this skill. And so we have to work out how we transition the force without either trying to do both things simultaneously, which becomes uneconomical, or trying to, you know, you end up in a situation where sort of institutional habits either deny the ability to innovate or you seize upon innovation in a very crude manner where you throw out the old thing entirely and you lose the elements of, in this case, sniping that are still pretty critical. The ability to get into a position, accurately observe targets, and accurately communicate what you're seeing.
Starting point is 00:22:43 So building on that example that we just discussed of the MQ-9 needing an increase in data analytics, are we actually seeing an increase of software developers or data analysts at the command cell level to increase the cognitive load of processing all this data? Do you think this is something that we need to incorporate in addition to the personnel that we're using to manage drones? And are you seeing an increase in analytic capability in the command cells of Ukraine? In terms of data analytics capability, I think it still sits within an intelligence function, and it's probably going to be one that is primarily done through reachback. The server capacity, the infrastructure, the permissions to properly integrate data
Starting point is 00:23:24 is something that is much more assured at higher echelon and much easier to protect. The challenge then becomes how do you distribute and disseminate that information without breaching security or classification boundaries? So I think we're probably going to see more data analytics supporting operations, but it will actually be held at quite a high echelon. And then the novelty is that that will be made available and accessible to sub-tactical units to be able to exploit that picture. If you try and do, you know, you want to minimize what you process, what you end up having to transmit back. And so
Starting point is 00:24:01 shifting to data rather than raw material and having edge processing to cut down what you have to communicate is very important. But once it gets back, the ability to fuse that picture is one where you want the greatest amount of data possible so that you can make those judgments with the right context. The number of times in Ukraine I've observed people reach conclusions at the front based on limited data sets as to what is happening to them, that has proven to be incorrect when you fuse the picture at higher echelon, highlights how dangerous it is to draw high confidence conclusions from limited volumes of data or very incomplete data sets. over to you for a moment, Eric, reflecting on this conversation that we've just had. How do you see the balance between autonomy and human oversight evolving in the future of regular warfare operations? Very briefly, it's going to be centaurs for a little bit. Jack hit on it earlier. We can't turn the keys over directly just for the variety of fallibilities
Starting point is 00:25:02 with the system. Nevermind, people will talk for days about the legal and ethical, moral. It's just capabilities not there. So I've been working with AI for a while. Many people are familiar with Project Maven, but I was using Picasa's facial recognition tools in a mud hut in Afghanistan back in 2010. So this is a tool that we've always used, but it augments human performance vice
Starting point is 00:25:21 replacing it in a lot of places. Sometimes it frees up your time to go work on more important things, but it's still going to be an integration. We're still in a very human intensive time for drones, which I think is what Jack really hit on with the War on the Rocks article. You look at anything, it's not just in the military. So centaurs are winning. Centaurs are winning in chess. They're winning in business. They're winning with a lot of other places. It's going to continue to be this augmented tool because the humans do edge cases really well and there's other things we do. I think what's been so transformative when you talk about things
Starting point is 00:25:46 like chat GTPT and the large language models is they're in everybody's hands now. That was not the case. Nobody got to play with Deep Blue, but like now everybody gets to use these things. So it's not in a high-end military platform. It's not at DARPA. It's here on my desk on my computer when I'm playing with stable diffusion to learn more about it. The Army of 2030 is heavily reliant on deep strike capabilities and that's going to require deep sensing. And as Jack mentioned, there's people who were like, there's going to be a robot, it's going to do everything. And I am a skeptic and I don't think that's going to happen. I think what you see, Ukraine's a great example. Deep sensing is happening, but it's not coming from a stack of robots or a stack of drones.
Starting point is 00:26:18 It's coming from an integration of people using these incredible tools, but it's still coming from a regular warfare partners and from all the other things that go into getting that intelligence. And I think Jack, what he just hit on is very important. Integration of that data, getting a whole or picture is going to continue to be the value of the headquarters is that ability to see it. Cause you're going to see contested. You're going to see information put into the, your system that is wrong either because it's been jammed deliberately or it's missing disinformation. And so the higher headquarters where you can get more of that picture is going to help. The challenge for us in SOF in particular is that we are going to always see a demand for edge computing.
Starting point is 00:26:53 And so we are looking at a lot of that, like, hey, what is the minimum amount of weight that I can put in a backpack to put a GPU forward to be able to provide some processing? And when does that just put such power demand on the drone that as soon as I turn it on, it crashes because I killed the battery. And so we're working through that space right now. And then the other thing for us in particular is these are not replacements for the big platforms, and I think everybody's really hit on that well, that these are augmented. They're tools. They are a new piece of kit that the sniper gets to employ to allow the sniper to be the value what he's always been, which is more as an intelligence asset than as an actual lethal asset. Lethal was just a capability that the sniper brought to the table.
Starting point is 00:27:27 The number one thing a sniper brought was his intelligence, his ability to see the battlefield and tell you what it looks like. And we just gave him a new tool. So we didn't replace the sniper, we just made him better. Now, there's an upper limit to how much a sniper can do. And so maybe we need to get another guy that helps to do that. And so it's this complimentary thing. And in particular, mass produced modular drones, I just see all kinds of opportunity for them. But also to go back to it, I got regular warfare partners. Sometimes they don't have HIMARS. And so it's not a matter of, is it a FPV drone or a HIMARS? And as you're seeing right now, it was the 155 round was the round of choice in Ukraine. Why? Because it was the one they had. And then they didn't have it and became the FPV drone,
Starting point is 00:28:01 because it's the one they had. And so there's something to be said for knowing that these are not about either or as much as it is what is available to you and how can you fight with what you've got. And again, just a reminder, we talk a lot of high-end conflict. A lot of this has been focused on like the stuff that's going on in Ukraine. But all these tools are available in the crisis and competition phase too. And they mean new things. And so that's the place that I think SOF in particular has to really reinvent what we can do and communicate it to the joint force so they understand how to employ us. Just to build on that very briefly with something quite tangible, a lot of the time in the occupied territories in Ukraine, the role of human in being able to find targets that are subterranean or in buildings and that kind of thing is extremely important. subterranean or in buildings and that kind of thing is extremely important. And it's important for the Russians as well. But human reporting is not very good at generating a category one grid
Starting point is 00:28:51 that can enable a strike, right? The turning that into something actionable often requires automation, surveillance, technical means, et cetera. And so you see actually the blending and complementary blending of very old capabilities and these new ones. And that allows you to get after those kinds of deep targets, which I think speaks to the question in SOF and what's the relevance for SOF. There are areas where SOF needs to change how it works. The success rates of cross-line infiltration by, should we say, green profile special operations units in Ukraine is not great, not least because that concept that soft units have often had, which is if you have a contact, you need to win the initial firefight so you can break contact and then
Starting point is 00:29:37 maneuver away in an environment where the enemy can bring down quite a lot of precision fires that will chase you once you're detected. And when you have that saturation of sensors over the battlefield, the ability to infiltrate and the ability to break contact and win that initial firefight for a small team is just not there. So it's not a survivable approach. But the question then becomes, how do you use a different profile to gain access? But that profile might not allow you to bring the tools that you need to do the job. So then how do you queue on tools? And that's where things like automation becomes really critical in being able to queue on effects. So I think, again, you see a complementarity
Starting point is 00:30:16 between very old skill sets and very new ones. It's one of the things that I think we have made our name in the last 20 years. There's a lot of people think about soft as something that was invented in 2001 and don't realize that in fact, most of our units were invented in World War II. And a lot of the skill sets from World War II are becoming very relevant again. One time pads are back. So one thing I just wanted to mention was the experience from Project Convergence. There seems to be a disconnect between how drones are being used in Ukraine and how the U.S. Army is currently viewing robotics. And I think a good example of that is what you're currently seeing with FPV drones. I think the U.S. Army sees this as
Starting point is 00:30:50 something very comparable to using like a switchblade or other high-end weapon system. But when you look at the price, you see a distinct difference between what the U.S. Army is currently looking to purchase and what is being produced by the Ukrainians with the FPV drones. Do you think this is something that is tied to more like cultural aspects in the U.S. Army? Or is it because the technologies are getting a lot better faster than how our acquisition systems can keep up with the requirements process and how we decide what it is we need to purchase? I think there is a disconnect because the U.S. Army's procurement systems treat procurement of
Starting point is 00:31:29 these devices as buying a product, a finished product, which can be named Desert Hawk, Puma, Switchblade. And those things are certified and operators are certified on using that platform. What we see in Ukraine is that any platform is rendered obsolete pretty quickly, unless you can get in, play with the software, adapt the tactics, sometimes swap out the sensors, sometimes swap out the antennas or the radios. That requires a different skill set in the unit, but it also requires a procurement system which allows you to draw upon different manufacturers components and then integrate them into your systems quite rapidly and it requires a regulatory and a training pipeline which says we trust our people to do that and what they produce it may come with
Starting point is 00:32:17 a safety challenge but so long as they have factored in and taken those considerations into account they have the freedoms and the permissions to be able to do that. We accept the risk. And so how the US military is able to move away from saying, well, we've got this one piece of kit and we're very happy with it, it's well assured, to being able to update these capabilities to keep them effective in conflict every two to 12 weeks. It is a major challenge for how money is allocated, budget lines are managed, and how procurement is governed in U.S. military procurement. Jack, that was a fantastic point of how the bureaucratic hurdles can sometimes stand in the way of what's needed of the force and needed of acquisitions. I know that other countries struggle with this
Starting point is 00:33:05 as well, though. It's just not suited to adapt as quickly as we need to be to the new technologies coming online. With that point, Eric, I wanted to ask you for your thoughts on the replicator initiative that's just been stood up to field new UA and get them down into the force, hopefully more quickly. Do you think replicator is going to offer more flexibility in acquisitions and allow us to uptake new technologies at a rate on par with how fast they're evolving? Is there hope there? Are there other acquisition programs that I'm not aware of that might? It really honestly is going to come down to a bit more of a paradigm shift. The speed of drone
Starting point is 00:33:44 acquisition is such an important topic that even the economist is covering. There's a whole issue they did recently about the United States military's inability to keep up. Jack's research he mentioned a second ago, like the UAV complex needs to update every two to 12 weeks. Maximum, that's once a quarter. So the Army procurement process, planning, programming, budgeting, and execution is designed to take about five to seven years. So put another way, we are 30 to 45 evolutions behind. That is a bowl full of OODA loops that we are not catching up. So we need to transition away a bit on some of these things. Platforms are still platforms. We're still going to buy tanks. We're still going to buy great big airplanes and all those things. They're still going to be platforms, but there's a system approach that
Starting point is 00:34:22 needs to be much more innovative and less, hey, I've got it, I've solved it. Because when the EW environment changes every week, you've got to have the ability right there on the edge to tailor your device. And we're seeing people that are able to do that. They're making these little small tweaks, changes in the firing mechanism, all kinds of things to constantly defeat. And if you wait for a polished finished product, you're not going to get there. General George, the Army Chief of Staff, has been pushing at this with the transforming and contact. It's trying to speed up the Army's evolutions, replicators a step in that direction.
Starting point is 00:34:50 I think we're going to need more steps, and I think we're going to need to really shift some of the paradigm. The challenge for all the Army is nobody gets to sit down. So first off, I get 3,000 people deployed in an average of about 80 countries every day. That's just breathing for our formation. We don't get to stop doing that, but that's also an opportunity for us because we get to go and I get to work with 80 countries every day. That's just breathing for our formation. We don't get to stop doing that, but that's also an opportunity for us
Starting point is 00:35:06 because we get to go, and I get to work with 80 different partner forces, and I get to try and innovate with things with them and take the lessons and learn from them, bring it back and incorporate them. One place that we didn't really talk about, and Jack didn't get into his article, it was the shift in doctrine we're seeing
Starting point is 00:35:18 with the loss of the Russian fleet in the Black Sea. And then Ukraine doesn't have a Navy. That's multi-domain at work. That's what the Army says we want to do. Now, I'm a former resident of the First Island chain. I am very interested in how land forces can destroy naval assets. That is a growth industry, in my opinion. And that is another thing that the current Army procurement process doesn't tend to look at, that we need to start looking at the opportunities there. Jack, so we've kind of been discussing around this, but you write in one of your reports you
Starting point is 00:35:43 did for RUSI, where the number of drones being lost per month is close to something like 10,000 on each side. And additionally, President Zelensky recently mentioned that one of his goals this year is to produce 1 million drones from the Ukrainian industrial base. In addition to choosing the right systems and also being able to change, what other aspects of the industrial base do you think you're going to have to see a shift in the next year in the war, whether it be from the United States or in Europe? One of the most unhelpful aspects of the term drone or UAV is that it covers everything from a $400 toy to like a 150 million UAV that's the size of a regional airliner. And that elasticity, conceptual elasticity, means that when
Starting point is 00:36:26 we have statistics like that, 10,000 UAV losses a month, it's difficult to quantify why that's happening and what it means. If anything, the number has gone significantly up since then, but that's partly because FPV drones are one-way attack munitions. So by definition, you're using them. If we were counting the expenditure of bullets, that wouldn't tell us very much. So your rate of attrition is different depending on what the UAV is. That loss rate was extremely high, was driven by the demand, which is a requirement for infantry units to have organic ISR that they can throw up and maintain situational awareness. They lack that capability. They will not be competitive with their adversary in terms of situational awareness and decision-making, and therefore they will suffer. They need it, but because they need it all the time, they can't do any complex route planning.
Starting point is 00:37:16 They can't concentrate on flying the UAV because they're actually in contact. And so the inability to decide or determine when and where they employ the capability means that if it's a bad EW environment, they're probably going contact. And so the inability to decide or determine when and where they employ the capability means that if it's a bad EW environment, they're probably going to lose the thing, but they still need it up. So they're just going to burn through them. And so that drives a very high attrition rate. If you accept that that is just a reality, the dynamics of that kind of platform, then using something like Replicator, you need to drive down the price
Starting point is 00:37:43 of those platforms and you need to treat them like ammunition, which is just going to be expended. When we talk about long range ISR UAVs with good stabilized cameras, strong data links so that the data can get back with low latency, the strong inertial navigation systems so that they can get into the right place and accurately determine the location of what they're looking at. Those are going to be expensive. There's no way of making them cheap because the sensors are the expensive bit. And given that expensive sensor suite, you want to improve the reliability. Now, if you're sensible about that, you can make those things last a month and you can probably get them for $120,000 to $200,000 a piece. So they're not horrendously expensive. If you start going way above that in terms of cost and complexity, you're probably asking
Starting point is 00:38:27 it to do too many things. You will therefore have to make the airframe bigger. It will show up more readily on air defense systems. They'll be shot down in greater numbers and there's a greater reward for shooting them. So there are some quite specific balance points for different classes of platform as to where you get a good return on investment and how that drives you to assess what trade-offs you make and how you ruthlessly prioritize what you put on the platform and what you don't. The biggest challenge about the adaptability question, which we were discussing
Starting point is 00:38:55 before, is how does that tie in with your ability to scale the capability? It's where the Ukrainians often fall down. They adapt systems very quickly. Their ability to scale those adaptations is often slower than the Russian ability to counter what they've come up with as an innovative solution. The Russian system is a bit slower to adapt, but much better at scaling when it does. So you see a difference in the two sides. Eric, how do you see this influencing resource allocation and investment priorities within USOSOC and other IW-focused organizations. Are there any shifts in doctrine or strategy that you think are warranted at this point in light of
Starting point is 00:39:33 these challenges that we've been discussing? Absolutely. I mean, again, none of our doctrine, I believe, really mentions a concept of just a person being able to throw a drone up and get ISR of their situation, their platform, their immediate area, which went from being a cool perk to being a requirement to survive. That's changed. Jack was just talking about having to go back and you look at some of the systems. We bought some of these early UAS systems. I'm having flashbacks to Lieutenant. I'm being sent to downtown Ramadi to go recover a small UAS that crashed because it crashed. And I'm like, I don't know if this is a good return on our investment, guys. There are a lot of people out here, a lot of IEDs for this one drone. We didn't have that thought process of, hey, this is a dispensable versus, hey, we do have
Starting point is 00:40:11 other drones that are incredibly expensive and we would like to recover the pieces. And so we're going to need the munitions approach. I think, Jack, you got it. We said the same thing. Like some of these things are just going to be, drones are going to be a class of munition in some cases. And FPVs, that's exactly what they are. But even some of the ones that are small ISR for the squad platoon,
Starting point is 00:40:32 maybe even the company, like just look at them as expendable assets that have an attributable capability, which I know is very hard for a lot of people because we love accountability of property in the United States military. The other thing that goes with the doctrine though, is again, like that lessons learned piece. And that's where RUSIC I think is really going to help us. The course there at SWCC, best example given is 2023, early 2023, we're working with Ukrainians and we're training them to help them get ready to doing the traditional soft role of advising them on how to do combined arms maneuver. But we don't know anything about FPV. Like that's new for us. And so they're coming out as we're doing the rotations of the training, they're showing us literally, hey, this is how I'm building it. This is how I'm adapting it. These are things. And
Starting point is 00:41:04 we're taking all that stuff back. Our ad is we show them how to do the higher level things. Here's how you take that individual FPV drone and layer it into a combined arms fight that allows this to be more than just a one-off. You killed a tank, but to actually create an effect on the battlefield, which is the hard part. It's not to pretend that the stuff that the Ukraine is doing isn't incredibly difficult. It's just that getting it to the next level is the tricky part for all militaries. But what we have now is a doctrinal place with RUSIC for all that lessons to come in and to be cycled through. And to go back to, again, what I talked before, like our partners are working with capabilities and with equipment that we don't have and we can't purchase.
Starting point is 00:41:37 And so there has to be a place for us to be able to learn those lessons and be able to then transfer those lessons when we're going to take it to another partner in the Indo-Pacific theater and give them the lessons learned so that we can be the value add that we need to be as soft. Again, partner force is what we bring to the table, but it's also the most important customer for us. How do you see the role of experimentation and innovation and also collaboration between policymakers, practitioners, and also industry stakeholders in addressing the complexities of integrating autonomous systems into regular warfare operations. Yeah. I mean, on the innovation point, I think one of the
Starting point is 00:42:12 challenges is that if we're talking about a large-scale conflict, your first echelon is going to be destroyed. It's going to be attrited, whether that's because of loss of systems, the loss of effectiveness of their systems because of the introduction of countermeasures or attrition of personnel, it's going to happen. And what we observe is that the generation of the second echelon is fundamental to your ability to continue the fight. Now, what you end up generating as the second echelon may use some of the same platforms as you had originally. But if you have much less time to train people, what they are able to do with them might be less. How they employ them might be different. And actually, given the adaptation cycles involved, some of the equipment will need to be changed.
Starting point is 00:42:58 And so I think when it comes to the relationship between innovation and war planning, one of the really interesting questions is essentially how do you achieve asymmetric rearmament, right? When you're generating that second echelon, how do you make sure that the second one rolling out the door is not a slightly less well-trained carbon copy of the first echelon with capabilities that the enemy has now suffered from, adapted to, and is ready for, but is instead being trained and equipped in a way that means that they come out the gate still relevant with day one capabilities for that second echelon fight that continue to give you technological advantage. That is a major challenge because until the war starts, you don't necessarily know what those adaptations are going to be. And if there's one thing we've seen in Ukraine, it's that industry takes quite a lot of time to spin up. The decision
Starting point is 00:43:48 points come early. And so you have to make some pretty significant strategic bets upstream about where you commit your resource. To give you just one specific example, Excalibur ammunition, so Precision 155 rounds, when they entered Ukraine in 2022, had about a 70% success rate of hitting their target. And the 30% of misses was largely incorrect generation of target coordinates. It's currently at about 6% effect. And that's because of hard EW countermeasures. Now, if you don't adapt, if you can't adapt to the capability, Excalibur rounds, then you're not going to be able to bring that effectiveness back up. If you had made an investment in 2022 to massively ramp up Excalibur production, you would now be seriously behind.
Starting point is 00:44:31 So how you avoid making those bad bets is a major planning challenge. So really, that was awesome. Thank you. So Red Queen effect is not new. Not everybody wants to pay attention to it, but this is the case. I really love what you mentioned, like losing the first echelon, like just knowing that as part of the case, it's kind of a fail early approach, which is very different than again, what we've talked about with the standard procurement process, but it's more of a recce approach, right? So instead of trying to solve for the solution, having the solution, building the solution, the big five approach,
Starting point is 00:45:00 it's, Hey, we know this is not going to be the final solution. So what's the best bet we can make on, but also position ourselves to be able to rapidly take it to the second generation, take it to the third generation. And so it's much more a startup approach, I think, than the traditional military procurement process. But I think it's absolutely the one we need to transition to. I would just say the difference now is because of some of the technologies, because of the ubiquitous and the lower cost of the technologies, like the difference is, is you can do all that innovation now, like
Starting point is 00:45:29 100% at the lowest level, people can innovate in ways they never could before. The question is, will you? This has been a really great discussion. Unfortunately, we are, we are nearing the end and this brings us to our final question. And that's what final advice would you have for policymakers, practitioners, and anyone who's worried about how drones and automation are changing warfare? And what focus would you expect to see for the next couple of years? If I could give policymakers, practitioners two thoughts, if I could. First one is pay attention. I was just at the pre-command course
Starting point is 00:46:02 last month in Fort Leavenworth, and more than one leader was very skeptical that the lessons learned in Ukraine apply to us in any way. Multiple people say, we don't fight that way, or even heard one leader say, our tanks don't sit still, they drive fast. Those are the future battalion and brigade commanders of the United States Army that are saying this. So absolutely pay attention. This idea that when electronics go down, I heard that critique, then we won't have any of these systems. I'm like talking the most contested EW environment on the planet and they're using this stuff. So for the love of God, we're not going to go back to flintlock muscles, pay attention and scope up as much as you can in the lessons learned and really make the changes we need to. My second point, we didn't really hit on it in this, but I do want to highlight one of the unique things about the drones is there's this unmatched IO impact that we get from drones. So like I can watch a video of a javelin strike, but I can't share that. I can't use it for effect. There's Twitter is loaded, Telegram loaded with FPV drone strikes on tanks. There's an IO component to that messaging. It's not an accident. That's why you're seeing strikes of dead tanks, but it doesn't matter because they're trying to get
Starting point is 00:47:00 funding for all the other things. So So there's this IO component to this technology that is a whole other untapped space in the battlefield for how are you going to use that effect. I definitely think that we need to look at the scale of the IO opportunity that some of these tools are bringing to the occasion. The biggest thing I would say is that we need to get this stuff into the hands of soldiers, and we need to be refining the concept of operation for them
Starting point is 00:47:24 and then scaling what we discover from actually turning them on and getting soldiers to use them. Far too often, these capabilities are kept away from troops, developed in isolation, tested without a wider context around them, and then you actually get them into a training area and they fall over on day one because they jam each other. They don't manage. There's no way of doing spectrum allocation properly. All sorts of problems start to emerge. And then the users don't trust the
Starting point is 00:47:58 system because they haven't been integrated in the process from day one and they're not familiar with how to resolve those challenges. And so they just blame it on this thing sucks and if people are skeptical they won't adopt so i think we need to make some bets we need to get it out there so that soldiers are familiarized with the idea that this is something that is just a tool they're going to have to use and they can start finding the surprising opportunities the concepts of operation that we didn't think up when you had them being mapped out on paper, because that's really where you get advantage when you have people who are skilled at applying these tools, adaptable, and can innovate with them in their hands.
Starting point is 00:48:37 You're not going to generate the perfect solution through a deliberate requirement setting process when you're talking about an emerging technology where the people who are setting the requirements don't fully understand either what the thing is capable of or the limitations that it imposes. What a great way to end it. Jack, Eric, thanks for joining us
Starting point is 00:48:56 on the Irregular Worker podcast. Jack, thank you again. Like loving your research, everything you're writing. Really appreciate it. Thank you. It's a pleasure to be with you. Yeah, always good to have a back and forth on the practicalities of these things.
Starting point is 00:49:07 It's really valuable as an academic to be able to engage with the practitioner community in this way, because I think you can come up with ideas, but until you understand the practical frictions involved, they just remain ideas. Thank you again for joining us on the Irregular Warfare podcast. Be sure to subscribe to the Irregular Warfare podcast so you don't miss an episode. The podcast is a product of the Irregular Warfare Initiative. We're a team of all volunteer practitioners and researchers dedicated to bridging the gap between scholars and practitioners to support the community of regular warfare professionals. You can follow and engage with us on Facebook, Twitter, Instagram, YouTube, or LinkedIn. You can also subscribe to our monthly e-newsletter for access to our content
Starting point is 00:49:56 and upcoming community events. The newsletter signup can be found at irregularwarfare.org. If you enjoyed today's episode, please leave a comment and positive rating on Apple Podcasts or wherever you listen to the Irregular Warfare podcast. It really helps expose the show to new listeners. And one last note, what you hear in this episode are the views of the participants and do not represent those at Princeton, West Point, or any agency of the U.S. government. Do not represent those at Princeton, West Point, or any agency of the U.S. government. We do not represent those at Princeton, West Point, or any agency of the U.S. government.
Starting point is 00:50:34 Thanks again, and we'll see you next time. I'm going to make a delicious

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.