Project Maven

[00:00:00] Speaker A: Theo, do you want to talk about the logo a little bit? [00:00:02] Speaker B: Yeah, totally. [00:00:03] Speaker A: You know, you launch a podcast and what do you have to do? You need to create the, you know, the account on wherever you're hosting the podcast. You need a name for it and you need a logo. You have to submit a logo for it to be accepted in the Apple podcast repository and others. And so that was one of the first things that we had to do was come up with a logo. We went back and forth on it quite a bit before settling on the one that we did. Which is based on something that you found. [00:00:28] Speaker B: Yeah, I, rather than do my job was one day spending time on the Internet looking at these old posters I'd come across for what was called the Victory Book campaign, which was a book drive of sorts, launched I believe by the American Library association in coordination with the United Service Organization, AKA the uso, to get books to soldiers over overseas. And I thought they were awesome. They had this great mid century quality and also the art was great. And it occurred to me that they might make for good, basically branding material if you ever wanted to do something like what we're doing right now. [00:01:08] Speaker A: Yeah, the PDF that you sent me was, it looked like a work of art. It looks like something that you would find in a coffee table book about World War II government posters. The font was this mid century stylized, strong font, the eagle in the center carrying this bundle of books. And so you sent me a, a scanned PDF of it and you know, it was sort of faded colors and everything and we're like, well, we can do something with this. And so we borrowed the eagle from it and cleaned it up to fit our, our logo. And that's the centerpiece of the logo from this World War II book campaign. If you, if you have an eagle eye, you can also see that I took some of the fonts, some of the text from elsewhere on the poster and I dropped the transparency down to like 1 or 2%. And just to give it some texture in the background is actually the, the text from that poster as well. I think it's really cool. [00:01:59] Speaker B: I encourage anybody who's listening, if you're bored and even a little bit into graphic design or art and the military to look up these posters. There's one great one from World War I of a doughboy carrying a stack of books with a rifle slung [00:02:15] Speaker A: and any podcast launches to initially to 0 subscribers. We've been fortunate to see the, the listener base grow with each episode. We're approaching the point where it probably justifies the many hours we spent working on the logo before the launch of the podcast. So keep the listeners coming and, and it will justify the. The expenditure of time that we put into the. [00:02:39] Speaker B: That that was the idea. We're just producing the podcast to retroactively justify the work that we did in Canva. So thank you for helping us feel good about the way we spend our minutes. [00:02:51] Speaker A: All right, should we get to this episode? [00:02:52] Speaker B: Yes, we should. [00:03:02] Speaker A: Welcome to the MWI Book Club, a podcast from the Modern War Institute at West Point. I'm John Amble. [00:03:07] Speaker B: And I'm Theo Lipski. [00:03:08] Speaker A: And the MWI Book Club is where we talk about books and the profession of arms. [00:03:12] Speaker B: Each episode features one book, fiction or nonfiction, new or old, bestseller or cult classic, and we give it a close read to find lessons about war, about leadership, and about character. [00:03:25] Speaker A: On this episode, Project Maven, A Marine Colonel, His Team and the dawn of AI Warfare by Katrina Manson. [00:03:34] Speaker B: Welcome back to the MWI Book Club, a podcast from the Modern War Institute at West Point. This podcast is predicated on the idea that an army that reads is an army better prepared to fight wars. My name is Theo Lipsky and my co host is John Amble. Each month we read one book that bears on soldiering and we invite you to join us as we talk about it. What you're about to hear is our fifth monthly episode. So far we've covered a good bit of ground across fiction and nonfiction. We read a memoir of the Allied campaign in Burma during the Second World War, that being Defeat the Victory by William Slim, a classic. We read a near forgotten Tolstoy nolla of his time fighting in the Caucasus called Haji Marat. That was our last episode. We read a influential post Vietnam vision of future conflict in Joe Haldeman's the Forever War. And of course, to kick off the entire podcast celebrating its 75th year, we read a classic story of loyalty and dissent in the United States Navy on the high seas in World War II called the cane Mutiny by Herman Woak. We've begun to get feedback from listeners as we now progress into our fifth episode on social media. We've gotten feedback via email. Most of it's positive. We're really proud of that and we're very grateful for it. And we've got some great constructive feedback as well coming in and we wanted to, by way of reply, share a couple notes with you up front. [00:05:01] Speaker A: Yeah. First, please, please keep the feedback coming. It really helps us as we try to kind of enact this vision that we set out to to produce, which is to create a product that might become kind of a hub for a community of readers and aspiring readers across the military and the national security community. If you listeners have ideas for how we can better do that, what books to read, what topics to explore, and how to better engage with you, we want to hear them. One note, I would say we are not yet at the point where we are taking requests from authors to feature their books. We have had a few people reach out. We are looking at potentially adding sort of another vertical to this podcast which would include discussions with authors about their books. We're not there yet, but if you are a reader and you've got a favorite book that you like telling your peers about, get in touch with us. It may be one that we want to consider for a future episode. [00:05:55] Speaker B: And another note, if you are as a listener with a unit of any sort in the military or even an office in the greater national security community that could be on the line or in the institutional force, and you are looking to integrate books into your leader or officer development program, we know that there are a lot of OPDs and LPDs out there. We're more than happy to partner with your unit to read a book together or to consider the book that you're reading for an episode. If that's you, somebody out there in the force reading with your unit or alone and you want to partner with us, please reach out to our email addresses, which you can find in the show notes and we'll get talking. [00:06:36] Speaker A: And third, if you have found the podcast sort of worth your time. If you like it, please give it a rating. Leave a review on Apple Podcasts wherever you get your podcast. There's a lot of noise out there and it can be hard to find sort of the signal in the middle of that noise. Those reviews, those ratings are a really helpful tool for listeners to find us. [00:06:57] Speaker B: With that said, those are our household notes up front. Let's consider the business of this episode. So far, the books that we've read have been mostly fictional and mostly backward looking, with Joe Haldeman's the Forever War as a science fiction text being a partial exception. I say a partial exception because it's very much the Forever War, as you might remember, an artifact of the post Vietnam moment in which it was published. Either way, none of the books we've read up to this point or were published after 1974, which we are going to change this episode. And in order to set up that book, we'll get into what we've taken to calling the front Matter. For those just joining us, each MWI Book Club episode has a structure, a structure which mimics that of a book. That is to say, we've got Front Matter, which includes an executive summary of sorts, an author introduction. Then we usually consider the book that we're reading in three parts, and then we close out with Back matter, which includes some margin notes and some book recommendations. So, as the first order of business for the front matter, John, do you want to tell us a little bit about the book that we've selected for this November episode? [00:08:10] Speaker A: Yeah. As you mentioned, we are fast forwarding. This is the first very, very contemporary book that we're talking about. It was just published this year in 2026. It's called Project Maven by Katrina Manson. [00:08:22] Speaker B: Yes. Subtitle A Marine Colonel, His Team and the dawn of AI Warfare. Yeah. So what do we know about Katrina Manson? Who is she? Why are we reading her? [00:08:32] Speaker A: Katrina Manson is an exceptional journalist first and foremost. She writes for Bloomberg News now covering national security and emerging technology. She formerly worked for the Financial times as the U.S. foreign Policy and defense correspondent. Um, highly recommend this book. Also highly recommend the work she does as a journalist because, you know, if you follow her on social media, if you read the work that she does, she really does exceptional reporting at the nexus of military affairs, national security and emerging technology like artificial intelligence. [00:09:06] Speaker B: I think all that experience also comes out in this book, as I'm sure we'll talk about when we get into the meat of it in its three parts. How did we settle on this book, though, for the November episode? [00:09:17] Speaker A: Well, short answer was we were, we were looking back at what we had covered, the books we had covered, and we realized that the biggest gap yet was sort of something contemporary. We had done non fiction and fiction. We had done, you know, World War II, focused and older. We had done sci fi. We. We just hadn't done a sort of non fiction look at something very, very current. And frankly, if you look at the books that have come out over the past year, this was arguably the best one and perhaps one of the most interesting, certainly. [00:09:48] Speaker B: Also, I knew that I didn't know everything I needed to know as a soldier and as a national security professional. And I thought this book would be a really efficient way to furnish myself with, frankly, the chronology and the proper nouns associated with the development of AI and war and man. Did it deliver. So, John, do you want to give us the MWI Book Club executive summary of Project Maven? [00:10:14] Speaker A: Yeah, certainly. Again, the book is Project Maven. You very helpfully added the subtitle, A Marine Colonel, his Team, and the dawn of AI Warfare. It's really about that. Subtitle Project Maven tells a story of the US Military's effort to integrate artificial intelligence into what it does. That effort was called Project Maven. More specifically, it tells the story of the team that set out to do that because it really got going based on the sheer determination, the strength of personnel, and an understanding of how to engage the bureaucratic levers of the Defense Department of one person, a Marine Corps colonel named Drew Cukor. We're introduced to Colonel Cukor and given a window into why he is so committed to building AI into the way the US Joint Force does its business. We watch as he builds the team to make Project Maven happen. We see that partnerships with industry that are crucial to making it happen. We see the initial products fielded, tested, and refined. We watch as he grows institutional, institutional backing for the program. We also see some of the territorial infighting as the Department of Defense's interest in AI grows. And ultimately, we see the project's flagship product, the Maven Smart System, rolled out more and more widely across the Joint Force. Now, listeners to previous episodes understand or know already that we typically break the discussion into three parts. We look at the book, you know, the first hundred pages, second hundred pages, third hundred pages, we. We look at, like, there are usually some pretty natural dividing lines to break book into three parts. For the purposes of this analysis, this discussion, we talked about doing that in this case, but we're trying something a little bit more experimental with this one. We're really breaking it down thematically. We're going to talk about in the part one, the technology. Right, the. The. The actual technology underwriting this initiative. In part two, we're going to talk about the intersection of that technology and the character of war. And in part three, we're going to talk about the leadership element of this project. [00:12:15] Speaker B: Okay, ready to move into part one, the technology. [00:12:19] Speaker A: Yep, let's do it. Part one, the technology. This book, if nothing else, is a primer on artificial intelligence for anybody who wants to understand the technology a little bit better and specifically the way the US Military harnesses or seeks to harness and leverage that technology. It's exceptionally well written for non technologists, non specialists in artificial intelligence. And I very specifically would include you and I in that group. And so one note before we kind of get going is that, like, we're not going to be able to explain the ins and outs, the ones and zeros of how all of these things work. We're going to talk about it at the sort of accessible level that Manson's book really engages with the material. If we make any mistakes, they're ours alone and not those of the author of the book. And we hope you'll, you'll forgive us. We're going to do our best to, to, to discuss it in a way that makes it understandable. But please understand also that we are, we are generalists. We're not experts in this subject. [00:13:19] Speaker B: And one additional note, John. We flatter ourselves with the idea that it might actually be important and useful for listeners, for us to sort through the technology as near laypeople, because this technology, as the book makes abundantly clear, is here to stay in the American way of war. And, and it is fast becoming impossible for you to be a soldier, certainly a senior NCO or an officer or a national security professional and not understand the basics of artificial intelligence, the distinction between deep neural networks and simple computer vision, and maybe even what a transformer is. So we're going to stumble through this stuff as we piece it together with the hope that it might be useful as you try to do the same so that we can all be more literate in these terms as we proceed in our lives and our, along our professional paths. [00:14:12] Speaker A: Yeah, so part one of this discussion, when we're talking about the technology, I really struggled with how to do it because as we've, you know, stacked up some of these episodes, there's kind of a particular format, a particular feel to each section where we're, we're talking about what a book says and we're extracting its lessons, drawing sort of parallels between our own experiences between the, the army, the US Joint force of today, and you know, the subject matter of books, there's really no way of doing that with this one. So I really approach this first part as kind of a primer on the technology and it. The goal is, my goal is to kind of impress upon listeners that look, if you want to be successful in the profession of arms going forward, you have to have a baseline understanding of artificial intelligence and how it's going to interact with our day to day operations and everything that we do. There's not going to be lots of leadership lessons, there's not going to be lots of importance of training. And that's sort the sort of thing that often we pull out of, you know, these books. We'll get to that later in the episode. But in this part one, it's all about the technology. So to understand how the US Military is Approaching artificial intelligence. I think it's really important to understand how the US Military treated technology in the past. And I'm not talking about war fighting technology necessarily, you know, the actual things that shoot. I'm talking about the underlying computer technology of the modern era that has been part and parcel to everything that we've done as a US Military. The traditional military relationship with technology separated humans from the technology, each with its own discrete tasks. Technology in that sense was just a tool. And I think that Project Maven gives us the clearest sense of that when it describes Colonel Cukor's first deployment to Afghanistan. He deployed as a young marine officer in late 2001 just after the 911 attacks to Afghanistan. And he brought with him a bunch of computer equipment. Colonel Cukor at the time, Captain, I believe Cukor was an intelligence officer. What he brought was effectively a suite of Microsoft Office. They used Microsoft Word for writing reports, they use Microsoft Outlook for sending those reports. They use PowerPoint for developing intelligence products. So that's, that's, that's effectively how the U. S. Military operated. It used commercial off the shelf products that were not very good. I listened to a podcast episode several days ago produced by the center for Strategic and International Studies. It was hosted by a guy by the name of Greg Allen who was formerly at the Jake the joint artificial Intelligence Center. Now replaced now is the cdao. You know, because there's nothing that the US Military loves more than creating, than changing, replacing initialisms with another one. And he actually described in this discussion with Katrina Manson, the author of this book, he described this transition to where then the US Joint Force started seeking out sort of bespoke custom technological tools. And it's, it's really fascinating. It's like 30 seconds where he just eviscerates these tools. He says, you know, these bespoke products were constantly over promised, under delivered. And then we moved into this era where they were they under delivered but they could sort of be fixed, but they were fixed in this really long slow process. It just wasn't fast enough for, to keep up with the demands of modern war. So in all of this sort of pre AI age, the furthest we got in terms of integrating technology to enhance the way that the U.S. joint Force operates was what we called intelligence driven operations. That really kind of took shape in the middle of the wars in Iraq and Afghanistan where we had the F3EAD, find, fix, finish, exploit, analyze, disseminate, targeting cycle. But it still left a gap between the humans and the technology. Technology was still just a tool to enable what humans did. There was a gap between staff sections, intelligence and operations. There was a gap, gap between workflows. The best we could do was to align these very different things, right? The intelligence and the operations, the human and the technology. It was all about alignment. It wasn't about eliminating that gap. We couldn't fuse those things entirely. The promise of AI as we're, as we're introduced to it in this book is that it could begin to close and ultimately eliminate that gap. So the book opens with a prologue. It describes Drew Cukor, this Marine Corps Colonel In 2017, watching a documentary about something called Alpha Go. Now, what is alphago? Go, for listeners who aren't familiar, is a very ancient Chinese originated game that is often compared to chess, but is almost infinitely more complex than the number of moves that can be made on the board, on a Go board compared to a chessboard makes it much, much more complex. A Google project called DeepMind set out to create an AI enabled program that could defeat what was at the time the current world champion, a South Korean named Lee Sedol. It did beat him. They played a series of five matches. This computer, this AI program versus the, the, the best player in the world. They played a series of five matches and the AI won four of them. [00:19:58] Speaker B: And there was a moment in the matches that is now a legendary moment in AI discourse and in the AI community, referred to in shorthand fashion as Move 37, which the AI system AlphaGo made at first to the ridicule of the observers, but as subsequent moves played out, they all then saw the human observers, that the AI had done something that appeared to them absurd, but in fact was brilliant and would win the match. And they hadn't understood it. So the move 37 moment, or move 37 moments are these moments wherein we realize that artificial intelligence has outpaced human intelligence, not just in rote activities, but in certain creative strategic capacities as well. [00:20:53] Speaker A: And if you read this prologue, you're given the impression, you're left with the impression that this was Cukor's sort of light bulb moment. As you get into the book, you realize that's not the case. He had that light bulb moment long ago. This was just sort of maybe like increased the intensity of the light bulb. But AlphaGo was a light bulb moment for us institutionally, I think, as a US military within the AI community, you said this is this sort of legendary inflection point. I would argue that within the military it was also this sort of legendary inflection point because it's the first time that we started having discussions about AI AI. The first time we started, you know, exploring, you know, we. We all sort of institutionally, implicitly, accepted that AI was going to change the way that we operated as military. And that's when we begin to have the discussions of how it's going to do that. Now, I was just this past summer, I was with a group of people, and there was sort of a friend of a friend who works for a very interesting AI startup in London at the moment. And she was kind of describing me, what she does. And she said. She started saying, have you ever heard of this thing called Alpha Go? And I said, of course I have. And she said she was kind of taken aback by the fact because she knew that I had a military background. She knows what I do now. And she was like, well, why do you know about it? And I said, you wouldn't believe how important this moment was within the US Military as this sort of inflection point. So I thought it was really interesting that that Manson started her book with that prologue, because I think she probably understood also that, you know, that that was this major inflection point for us. Cukor told the author, Katrina Manson, and I'm going to quote this directly from the book. He said, that's really what comes from this AlphaGo thing. You can't play these games anymore without this technology. You can't go and step on a modern battlefield without this. Otherwise you are behind. You're not serving your country well. And if the other country has it, your lease at all, you can't win. You will lose. So, I mean, I just think it's a very powerful illustration of, you know, this. This recognition across the US Military that this was inevitable. And Cukor was a pioneer because he recognized it before this AlphaGo moment. [00:23:03] Speaker B: Although the thing to track as we all educate ourselves on artificial intelligence or superintelligence and its variants and its applications is they're not all like, that is to say, the software. The technology at the heart of AlphaGo is not the same type of technology that Project Maven dealt with for the first couple years of its existence. And the superintelligence, or artificial intelligence with which our listeners now interact is also different than AlphaGo. The distinction between computer vision and reinforcement learning and large language models we'll try to parse as we proceed. The other note that I think is interesting, and I was monitoring the entire time I read this book, it relates very much to the quote you just provided us from Drew Cukor, is that in Certain ways, when you step onto the battlefield, it is analogous to a game of God, but in certain ways, the battlefield is not analogous to the game of Go. And so to the extent anybody is or is not Lee Sedol or AlphaGo is a open question and a debate that we'll try to integrate into this episode is when that analogy holds and when it doesn't. None of us know the answer to this question exactly, but in later parts, we'll delve into it anyhow. [00:24:35] Speaker A: So how does the US Military react to this growing recognition that AI would be, would inevitably, inevitably be part of how it operates? Well, the first sort of application that it saw, or individuals within the US Military saw for it was essentially for kind of data management and surveillance. During the wars in Iraq and Afghanistan, we exponentially increased the aerial surveillance using unmanned systems of the battlefield. Stan McChrystal called it the unblinking eye. We were generating hours and hours, thousands of hours of video and imagery data, and it was outpacing the capacity of even a growing body of analysts to be able to do anything with it. And so this became kind of a way to overcome not only the, the literal, like, manpower limitations that we had, but also the, the just the sheer challenge of analyzing that sort of data. There's a really interesting vignette that's shared in the book where they're talking about a guy, an Air Force analyst, who had, you know, part of his job was looking at all of this data, and it talked about these long shifts of, of kind of monitoring these, these UAV feeds. [00:26:06] Speaker B: Yeah, this is the screeners, right? That's what they call [00:26:09] Speaker A: them, the screeners. So I want to, I want to read from it to this just to give listeners maybe a sense of how challenging this can be. It said Gizardo. This is the name of the. The member of the Air Force. Gizardo developed his own cocktail for getting through. 14 hours a day. Glued to a screen, he'd go through a bag of sunflower seeds each shift working his fin mouth, constantly topped up with Red Bull. Others on the team chewed gum or snapped a rubber band against their wrist. Anything to keep busy and keep going. But mostly he taught himself to keep asking questions. He went into his head seeking out new details. Every question he asked himself, every detail he unturned. Every minute step he took deeper into the pictures in the scenario was a way to stay awake. He kept his brain buzzing, sleeping for four or five hours and then was back for more. [00:26:51] Speaker B: Right. [00:26:52] Speaker A: Imagine the challenge that a human brain faces trying to overcome that critically, what doesn't get a computer, an algorithm, [00:27:01] Speaker B: and this image of the exhausted screener coming up with cocktails and cognitive tricks in order to stay awake, I think is important to bear in mind. I think Drew Cukor and the minds behind Project Maven want you to bear it in mind. Because when you're evaluating the prospect of an artificial intelligence engaged in imagery analysis or even targeting, you consider all the downsides. And implicitly, sometimes we fall for the trap that the alternative is perfect human performance, when the alternative is unfortunately imperfect human performance, sometimes at great cost. And what that looks like is, yeah, some caffeine addled individual doing their absolute damnedest. [00:27:45] Speaker A: So Project Maven launches with a memo by then Deputy Secretary of Defense Bob Work. Its main objective at the time is to integrate artificial intelligence into intelligence. And this is a theme that we see throughout the book is the distinction between intelligence AI use in intelligence for intelligence purposes and in operations. And I think this is worth sort of pausing to, to discuss a little bit because as I said, the, the initial vision was this is going to help us analyze the thousands of hours of video and imagery that we are collecting mainly from UAVs and other sensors. Eventually you see that they say, well, if we can use this to analyze previously captured video, perhaps it can be integrated into the video capturing mechanism and be used in real time to identify objects and track those objects around the battlefield. There's a quote at one point in the book where the, you know, one of the characters says, we want to identify everything. They want to have perfect visibility on this battlefield. They want, they have this vision of what we now call the transparent battlefield. There's a discussion to be had about how transparent it actually is or how complete that actually is. But what's really important is that it starts as this is. This is an intelligence function. The question of whether or not it would be integrated into operations, meaning using artificial intelligence to determine what to strike and perhaps even to conduct those strikes, is sort of a third rail early on for a number of reasons. Mainly it's because the funding that Kukor's team has is from military intelligence program dollars. It means that they, I mean, he actively tries to tell his team, don't say operations, don't say targeting, don't use these sorts of words. So I think it's worth pausing for a second again to, to discuss a little bit about what targeting is because there's a really interesting, interesting sort of semantic thing going on here, right? Targeting. Like if we take it out of the military context, targeting is A gerund. Right. It's a verb, takes the ing form that then acts as a noun when we say, you know, jihadist camp was targeted with an airstrike or, you know, Houthis targeted the, the main airport in Riyadh with drone strike. That targeting sort of refers to the kinetic action of launching some projectile at another place at a target and, and destroying it. Targeting in a military context is something very different. It is the act of identifying high value targets. They could be key individuals, they could be installations, they could be all sorts of different things. And building up the targeting package in the, in sort of a post 911 context, it was typically, okay, who are the high value targets? The key enemy combatant leaders developing pattern of life, where they go, you know, what sort of person like their individual network. We use programs like Analyst Notebook, which was, you know, I have nightmares still about using Analyst Notebook to develop sort of the networks and figure out, you know, who might lead us to this person. All of that, that could be weeks, months, years long project. That is targeting. It's not just about the, the sort of kinetic action. That is what they were really sort of initially kind of setting out to do was to streamline and speed up that, that targeting process from weeks, months or years to the, the sort of, the shortest possible time from we identify a target, we develop all this information necessary to strike that target and then conduct the. The strike. But it remains intelligence driven. Now in the book, there's a reference to Bob Work sort of maybe regretting again, the Deputy Secretary of Defense at the time regretting not adding some language into the initial memo, standing up Project Maven that that would sort of include, hey, this will have an element for operations as well and an operational element, not just an intelligence element, because that becomes again, this sort of like third rail that for much of Project Maven's existence they had to stay away from that until as we see, they link up with the 18th Airborne Corps in the Army. And that kind of offers the Maven smart system a way into, into operations. [00:32:17] Speaker B: I think that this is a fascinating through line in the book, and frankly, I admire the way Katrina Manson handles it. There's a not a small bit of controversy around Project Maven in certain corners of the tech community for exactly this reason. It was presented as initially to all sorts of folks a way of integrating data into intelligence processes. But it was either suggested or set out right by various folks over Maven's early history that it was not a project that intended to integrate artificial intelligence into operations. And the bright line between intelligence analysis and operations is at times difficult to pin down. [00:33:01] Speaker A: Yeah. And I think what's important is that again, this was, this was always about speed. Right. There's a, there's a quote in the book where I think it's from Kukor, who says, I believe Google updates their core software several times a minute. We're lucky to update meaningful processes once a year. Right. There's this, it's, there's this underlying recognition that the systems and processes that drive the way the US Military operates are far too slow for the modern battlefield and frankly, far slower than what the private sector does on its products. Now, the, one of the, speaking of speed, one of the, I think most significant accomplishments of Project Maven is, you know, this is sort of a ragtag bunch of, you know, people on a small team. The memo that, that Bob Work signed standing up Project Maven was signed at the end of April 2017. By the end of that year, they had what would become the Maven smart system in the field being tested, being used not extensively. It wasn't fully integrated into any, you know, into the way any particular unit did what they did. But it was being used crucially on real world data. It wasn't on, on. It wasn't using just training data that was sort of sanitized and, and maybe had little connection to the battlefield. It was being used on real world data. And it didn't actually work very well in most of those places until it did. Because that's the key, right? With, with an AI tool, it's about learning. The more it's used, the better it, the better it gets, the more you can refine the algorithms. And that was the key to what, you know, what I think he called Cukor called field to learn. Get it out there, practice, figure out what doesn't work, fix it and get it out there again. [00:34:51] Speaker B: There is a lesson, by the way here. This is the technology section. So I'm just going to put a pin in it about the way you achieve organizational change in a bureaucracy. That is to say, you start a necessary movement before you have a complete plan. Something anybody who knows their true pleading procedures will recognize, but sometimes gets lost at higher levels. But we'll circle back to that in part three where we talk leadership. [00:35:18] Speaker A: No, that's a great point. And I think one of the centerpieces of that part three discussion that sort of brings us to a close of the technology, right? We saw from the initial spark and the determination of one person the development of a technology and its Use cases slowly expanding over the course of now nine years since it was stood up until today, when we have the product Maven Smart system in use across the joint force. [00:35:47] Speaker B: And in sum, because it gets confusing, there's Project Maven, there's Maven Smart Systems, there's all these other companies that are brought in along the way in order to write algorithms. It's worth restating what it is. So Maven Smart Systems, which is eventually. This platform that's created by Project Maven, is a platform for the integration of a bunch of data analytic tools that presents the synthesized common operating picture of those tools to the user and allows the user to both cohere that common operating picture and then act on it if need be. The thing technologically that is interesting is for those who are trying to understand AI a little better. For much of its prehistory and its early history, Project Maven is really a matter of what we call computer vision, which is to say writing algorithms that are good identifying objects on a drone, feed on a wide area moving image. This is a type of video Produced by our MQ9 drones primarily, at least was during the war on Terror. But right around the time that Bob Work is signing the memorandum which creates Project Maven formally, because he has been so taken with the demonstrated efficacy of the computer vision which Kukor oversees the development of at Google, a paper is published which now, looking back in 2026, we can see clearly changed human history. And that was attention is all you need, which introduces this idea of a transformer. A transformer is a piece of technology and architecture that allows basically a model to look at entire sentences at one time, rather than word to word. That completely changes the trajectory of large language models. It allows them to scale much more rapidly. And much of the things that we now called artificial intelligence are models that integrate transformers. They're large language models. That's the T in ChatGPT. And so Maven later, because it's a platform, integrates large language models into it so that they exist alongside and are integrated with computer vision. The parsing of those technologies I find is useful because otherwise things get quite amorphous pretty quickly when, when we talk about AI. But if you want to understand what exactly is going on when you open up Maven Smart systems and why things happen when they did, it's good to have these different concepts at hand. [00:38:34] Speaker A: So that's sort of the. That covers the discussion of the technology, a little bit of hardware, a lot of software, but the missing component that we haven't talked about yet is data. And that's going to be a kind of a centerpiece of part two. So, Theo, are you ready to transition to part? [00:38:51] Speaker B: Sure. This is part 2, AI and the character of war. So, irrespective of what sort of artificial intelligence you're talking about, be it computer vision or a large language model, and both, again, play a big role in this book, you are going to need a lot of data in order to train the intelligence to be as effective as possible. And one of the great storylines in this book is Project Maven's search for data, that is to say, everything from labels to the underlying video footage, from drones to information about targets. The stories that Katrina Manson wrangles in order to illustrate just how hard the fight was for data, I think are worth talking about. Both because they show what's possible, if you think creatively, even in a relatively calcified bureaucracy like that of the United States military. And because the way that the Project Maven team and the way that the sort of legacy military, which in cases Project Maven treat data, both imply something about the character of war and what artificial intelligence can do in war. So, John, I don't know if you felt any flashes of recognition when reading this stuff, but I certainly did feel some flashes of recognition as I thought about my own work in relation to the stories told. There's one great moment in the early days of Project Maven when, having decided that they have the basic software teams in place to write the computer vision algorithms in order to get eventually where they need to be, the Project Maven team goes looking for data. On my page 47, they even refer to it. And I thought this was a kind of helpful, earthy metaphor for what they were doing. They refer to data as the feedstock of AI and found that it was, quote, bundled away on old slide decks and hard drives and computers and cupboards that sat as good as discarded. No one was bringing original raw data to share, sort or understand it. And what's more, the services, when contacted by Project Maven, which sat up under the Deputy Secretary of Defense at the time, were disinclined to share what data they did have because they were, in the telling of Katrina Manson and her interviewees, scared that by providing data to Project Maven, they would imperil their own services programs. And I know in Part three, we're going to talk leadership and bureaucracy, so we might circle back to the perverse incentives inhibiting data sharing there. But eventually the Project Maven team goes out to the force in order to find those Dusty hard drives, wherever somebody's willing to share them. And so two project maven team members end up, and now I'm Looking at page 73 in Maryland at the Patuxent river drone office. And they go literally diving into closets where they find hard drives, quote, stacked up and collecting dust since the days the Marine Corps were deployed in Afghanistan in 2010. The use of this data is that it provides that wide area motion imagery that is extremely difficult for a human to process for the sweep of it, but is therefore a very good use case for computer vision, that is to say automated screening. They iterate, as you said, and in these environment, operational environments, at first the AI doesn't perform very well, but it gets better. What's interesting to me is John, about the story from, you know, the first dumpster diving through the successful unification of a bunch of data streams on more mature platforms, you know, later in 2018-2021, is that all the while the data for which they're looking and the data that they use to train the models is data collected for targeting, as you described it. That is to say, usually you're trying to find people and usually you are using drone footage from these medium altitude drones with these exquisite cameras on them, which generates a lot of data, but ultimately speaks to a pretty narrow function in the sweep of military operations, targeting of individuals with imagery provided by a slow flying unmanned aircraft. What, what interests me and what I think the book doesn't ever fully settle is the extent to which the processes by which these artificial intelligence systems are trained can expand to account for much, much messier data environments. They spend a good bit of time in the book. That is to say, the author spends a good bit of time in the book talking about the effort, the human effort that has to go into labeling hundreds of thousands of hours of wide area moving imagery in order to get the training off the ground. But that is still relatively clean imagery taken when drones can fly in looking for pretty specific items. I wonder if this process can be replicated for actions that are more than tactical actions, more than just striking an individual van and environments where the data that you would need in order to train models are far scarcer. I don't know if you, if you had this reaction when, when you were reading it, but the thought dogged me all the way through the book. [00:45:01] Speaker A: I don't think I did because I could see the boundaries that the Manson was putting on this discussion. She was, she was engaging with a project as it was and wasn't necessarily engaging with these Sort of secondary and tertiary questions that will become relevant going forward as we try to apply artificial intelligence to things other than targeting. As you said, some of the thornier things that maybe AI is less optimized toward, toward accomplishing or helping us accomplish. [00:45:34] Speaker B: Yeah, there's another use case that one might bring up in response to my question, I realize, which is the submarine use case. There's a great section. We don't need to dive into the details here, but in order to demonstrate the applicability beyond the war on terror in a more great power competition context. Katrina Manson tells the story of the attempt to use artificial intelligence to detect submarines. And that of course depends on data collected from the sea floor using sonar technology. But that too, for all of its many challenges, which the project maven team seems to have very admirably overcome, is a relatively discreet, clean environment in the sense that you're looking for a very discrete object in a pretty uniform domain. And in that way it's much like the go board, you know, which for all of its complexity and possible combinations or permutations of moves, is a discrete space with a fixed grid and is perfectly visible. So even though the submarine example is a non gwat context and is a cross domain context, it doesn't to me answer the lingering question that I mentioned earlier about how well this sort of artificial intelligence will perform in environments where the data is corrupted or completely inaccessible, where the factors that matter are not readily observable, or where the iterations are so low in number that, you know, you don't have hundreds of thousands of hours of video footage to consider them, you know, individual battles or individual wars. [00:47:22] Speaker A: I think the, the answer is that it won't function right. Because this is. Artificial intelligence can function if you don't have enough data. Right. There's, I mean, there's a sort of four part construct that, that some listeners will be familiar with. I'm familiar with it from Paul Shari's book Four Battlegrounds that says essentially for AI to be incorporated into, for AI to work, broadly speaking, you need to have four things. You need to have institutions, talent, computing power, and data. And the data problem is always going to be one. If you had access to all the data in the world, you're going to have a pretty good AI. If you have access to less than that, every increment, less than all, then perfect data, perfect fidelity is going to reduce the effectiveness of the AI. And look, the data could be mislabeled, it could be corrupted. We see an example of that happening in the Book, it could just not exist. The, you know, where we see in the book instances where data is collected and then it's deleted at the end of, you know, whatever cycle. We actually, there was a, around the time actually that Project Maven stood up in 2017, I believe it was John Spencer, who at the time was still serving the military on staff at mwi, wrote an op ed in the Washington Post drawing attention to the fact that it's harder to learn lessons from our wars because we had become so reliant on technological tools at the time, still basically Microsoft Office. We had hard drives and hard drives full of all sorts of different things, reports that were sent up. Anybody who served, you know, deployed to Iraq or Afghanistan knows just how much documentation was produced. And we, we deleted all of that going back. All of those reports are, are pieces of data that could have been used for training, a large language model, for instance. And we saw the same thing with, with moving imagery, with UAV video that was deleted from time to time. So I think it is a, it's a, that is the challenge that Maven faced in standing this up was having enough access to high fidelity qual or labelable data. And it's the challenge that we're going to continue to see as we seek to, as you said, apply or utilize AI and AI enabled tools in place where we don't have access to that data. [00:49:47] Speaker B: And I think it's worth parsing the several different challenges that confront the effort to get quality, contextualized, relevant and easily ingestible data to artificial intelligence. One is the retention problem that you mentioned, right? Many of the historical use cases that might bear on a given model have seen a lot of the data that depicts them deleted the stories of reports from Afghanistan and Iraq. Another is an example from the book the P8 Poseidon hard drives that were going to be instrumental to training artificial intelligence to detect drones Project Maven Teammates Discover has been deleted, never saved. Then there's the question of whether or not the data is accessible even if it does exist. There are technical obstacles to access, but there are also organizational obstacles to access. I alluded earlier to the way that services in the book appeared jealous of their particular data. Ironically, Drew Cukor, who at first is the great red tape cutting liberator of data, becomes himself a jealous proprietor for data when his entity Project Maven comes in a contest with the joint Artificial Intelligence Command JAIC that we mentioned earlier becomes cdao. And there's this kind of interesting self reflective moment halfway through, through the book wherein the author Asked Drew Cukor if he hasn't become the villain that he derided for much of his career. And he says, I could. I think the quote is, I know that's how it looked. The irony is not lost on me. So there's the access question, but then again there's the applicability question. You know, just even if we were to recover all of the data about individual patrols or, or targeting from Afghanistan or Iraq, or even if we were to recover every single hard drive worth of P8 poseidon sub hunting data, would it be useful for artificial intelligence to scale up to a point where it's super intelligence in the context of war? This problem I think is less acute in civilian uses than it is in military uses because there is nearly infinite data about the world in non war contexts. But owing to the nature of war, how disruptive it is, how every system is trying to destroy the other systems, and owing to how difficult it is to observe and how in certain ways unique each war can be, there isn't infinite data. And what that means is that the wartime uses of artificial intelligence might not scale at the same rate and to the same level that the peacetime, so to speak, artificial intelligence can scale. And I was struck by just how tactical all the data is discussed in the book. To that end. You hear a lot about scaling and scaling laws. And when people talk about the promise of AI or the trajectory of AI, it's not totally unrelated to the levels of war which we talk about when we discuss military operations. And I'm not sure that the tactical data can scale up to strategic executive functions. And people might point out Go is a strategic game, but we end up with the same problem that we alluded to earlier, that Go is in certain ways analogous, but in certain ways not because the data is perfect and the board is so discreet and bounded. [00:53:37] Speaker A: Yeah, no, that's a good point. You know, this is very much a, a tactical application story about, about artificial intelligence. It does become more difficult to think about or I guess the, the operational level campaign plan, campaign design applications of AI are a little bit less obvious. And certainly when it comes to strategy, military strategy and beyond grand strategy, you know, how AI is going to, going to apply at those, those levels is, you know, that is like I said, much less intuitive than it is being able to train an AI to recognize an AK47 or you know, a particular weapon system, particular vehicle, be able to track it. That's just much more obvious. [00:54:25] Speaker B: Yeah. And this point is addressed towards the end of the book to very specifically in the context of Ukraine. So here's a. A war that is decidedly not like the global war on terror. Here's a context that's decidedly different and closer to great power competition than Afghanistan or Iraq. And Katrina Manson has some great discussions, it seems, with now General Retired Chris Donahue, in which Chris Donahue lays out this exact problem. He says, and I'm now looking at page 284. To the extent Maven was successful, it wasn't the algorithms themselves, but high quality underlying data, the system in which the algorithm was situated, and how seamless the workflow the teams could develop. It's that first leg of that he mentions, high quality underlying data that I think is going to continue to devil us as we attempt to integrate artificial intelligence effectively not just into the targeting process, but into more operational and strategic matters. It's that third thing that General Retired Chris Donahue mentions is the key to the success of Maven that we now want to talk about, but at a more meta level probably than he meant it, that is General Retired Donahue said the success of Maven also depended on just how seamless Project Maven could make the workflows and staffs and the military can make the workflows using Maven Smart system. This whole book is in part a story of the attempt to build a workflow. And there are a lot of different parties involved across the bureaucracy and a lot of different personalities across the bureaucracy. So even if you and I don't think you should scrape out all of this stuff about artificial intelligence, all the stuff about the war on terror, all the stuff even about Ukraine, there would still be a useful story here for any member of the national security community or the military, which is how you change bureaucracy and how you get teams to work together despite personalities and contravening incentives. That's the topic of our third part, the Leadership Factor at the heart of this book and at the heart of Drew Cukor and Project Maven's story. So, John, do you want to move on to part three so we can talk? My favorite topic, bureaucracy, and everyone else's favorite topic, leadership. [00:56:47] Speaker A: Yep, let's do it. Part three, the Leadership Factor. This is a book about AI, but as you alluded to in that sort of transition, it's a book about leadership. It is a fantastic window into the leadership, particularly of one person. You know, that was clearly Katrina Manson's intent by giving it the subtitle that she did. This is a book about Drew Cukor's leadership, but you also see elements of leadership above him and below him that are just, just invite so much exploration of what it means to lead in the sort of environment that Project Maven sort of emerged and, and operate. [00:57:27] Speaker B: Invite so much exploration is a very tactful and artful way to put it. I, and I, maybe the listeners will see why in a second [00:57:36] Speaker A: and I should, I should caveat this with a note. I really struggled with this because look, I think that there are examples of exceptional leadership and I think there are examples of poor leadership in this. Any of the examples of poor leadership that we talk about is not intended as an indictment of any particular individual. Because, you know, in prepping for this episode, I watched a number of interviews, listened to a number of interviews with, with Drew Cukor, retired Marine Corps Colonel Drew Cukor. I'm fascinated by, by him as, as an individual, as a leader. I think, you know, as the book makes clear and a lot of people say in the book, Project Maven could not have happened without him. I'm, I'm a believer in that sentiment. So this discussion, again, you know, is not intended as an indictment of any individual as much of as it is an exploration of leadership under very, very unique and, and trying circumstances. When, when Colonel Cukor set out on Project Maven, he faced a number of, of challenges and I'd like to talk about three of them. The first one is he had to move a very change resistant bureaucracy and he had to move it faster than the bureaucracy ever wanted to move. This was always his goal, as we see early in the book. He wrote his master's thesis, you know, in the early two, I believe, at the Naval Postgraduate School on modernizing the, the Marine Corps intelligence enterprise and, and freeing it from these inherently, these inherent bureaucratic limitations that kept it moving slowly and never able to keep pace with the threats that it was, that was, that it was intended to keep pace with, to deal with. So I think to understand how the Project Maven team overcame some of these bureaucratic constraints with Project Maven, you have to rewind back to earlier in the decade, around 2010, 2011, with the effort to bring on board systems designed. Well, a system designed by what is now a very well known and somewhat controversial company called palantir. Now in 2010, I remember this, I was assigned to a reserve unit I was doing. I was in grad school at King's College in London. I was assigned to a reserve unit called the Joint Reserve Intelligence Support Element at RAF Molesworth, a US military base about an hour north of north of London. I was there when General Flynn who was at the time, I believe the ISAF G2. The. The. Well, I guess would have been J2. The ISAF J2, the senior intelligence officer for coalition military operations in Afghanistan. He and I believe it was a Marine Corps captain who worked for him wrote this in the intelligence community, the military intelligence community. It was sort of earth shattering, wrote this report castigating the military intelligence enterprise and how poorly it was performing in places like in Iraq and Afghanistan with the specific challenges that they were facing that one of the outcomes of this document was that they needed better tools. Tools. One of these tools, it would quickly become known, was produced by Palantir. It was basically a data management product. There was a joint or joint urgent operational needs requirement assigned to it in order to get it into the field more quickly. Basically said that the military intelligence enterprise in Afghanistan needed Palantir. The army rejected it. I think that that was a sign of the. The bureaucracy that Cukor would run up against starting then and continuing when he got to Project Maven. [01:01:36] Speaker B: And it's worth noting the army rejected it because the army thought that it had something that already did what the ju on said the army was spending hundreds of millions of dollars on it. [01:01:46] Speaker A: And that's it. That's. That is an, an obvious and inevitable bureaucratic response. Right? That is what a bureaucracy is set up to do. We already have this. We're not going to bring in a competitor that creates chaos, that creates anarchy and it doesn't work for what we need to do. Kukor at the time, I believe he was working in the Pentagon. He got wind of something. It was sort of a plus up money that was on top of the allocated budget. The book makes clear. Look, the defense budget at the time as it is now is extraordinarily high. There are hundreds of billions of dollars that go toward defense programs. The problem is they're all allocated. They're all, they're all programmed for certain things. It can be exceedingly difficult if you want to spend a relatively small amount of money because there isn't just free money sitting around that, you know, it's not that the Pentagon gets a check for. At the time I think it was 600 and some odd billion dollars and said spend it on what you need. Cukor got wind of this plus up of money for a Pentagon. I believe it was an irregular warfare kind of R D lab. This lab spent money on a wide array of projects, but it now had this extra $36 million and nowhere to spend it. And nobody was claiming it. Nobody. Candidate Cu Cor got them to give him $10 million, a pretty trifling sum of money in the grand scheme of things at the time considering the defense budget. He then got the Army Research Laboratory to create a statement of work that became a public solicitation that really only one company could meet. That company was Palantir. He was authorized to do, do this. The, the sole source authorization was included in the, in the statement of work. I, I guess this would foreshadow the way that the Maven SMART system in. Would originate its own funding when he got there six years later. Some odd [01:03:34] Speaker B: right. And it's worth noting that how personal and how bespoke a lot of this work was. When discussing the joint urge and operational needs statement that General Flynn put out the JU on the author quotes one observer as saying that the juon had clearly been ghostwritten by a Palantir engineer, but in fact it had been ghostwritten by Kukor and arl. The Army Research Laboratory puts out the solicitation for the contract that Kukor now has the money in order to pay for. It's doing so on Cukor's behalf. So Cukor is. One can imagine somebody almost laying down the planks of a bridge across a valley or a great canyon as they cross it in order to make, make, you know, the, the, the passage a reality. It's at once depressing and also kind of breathtaking. I don't know if that's how you felt about it. [01:04:35] Speaker A: It's almost a little bit inspiring. If you're somebody who feels sort of perpetually battered by the bureaucracy, it's sort of inspiring to see that somebody kind of got one over on the bureaucracy. Manson refers to him, to Cucor at one point as a bureaucratic ninja. And I think this is the first time you kind of see that in action. [01:04:53] Speaker B: Right. But you can understand why people, people at the time thought it was borderline corrupt, that Kukor was forcing a very particular solution rather than engaging with the processes in a good faith way. And of course, history seems to have vindicated him. And one thing that I think we're going to consider this part is whether or not every single excess or extra procedural move was necessary or worth the cost. [01:05:17] Speaker A: Yeah, it's a, you know, in terms of the bureaucratic challenge. Right. This was the first leadership challenge that he faced. I think you have to chalk it up as a success for Cukor and the team. Even though this is again, this is, this predates Project Maven. But this is his first sort of win on that front. And really kind of provides a template for what he would do later going forward with Project Maven. [01:05:41] Speaker B: Yeah. [01:05:42] Speaker A: Now, he's also. One other element of this is the bureaucracy is what it is, right. It is offices, it is programs, it is line and block charts, it is budgets, it is systems, it is processes. It is this complex thing that sort of binds us all into our little place. But there are mechanisms everywhere that sort of exist in parallel to it or alongside it or that or inside it that just give a little bit more space for discussion, for ideas, and sometimes for action. One of these was one that he sort of, I guess, kind of shoehorned himself into while he was working the Pentagon. He got wind of this thing that was called the Breakfast Club. [01:06:24] Speaker B: Yeah, he invites himself. [01:06:26] Speaker A: He did. He invites himself. He did have a pretty powerful sort of three star backing him that gave him kind of not carte blanche. But, you know, he was a colonel, but he was known to be operating on behalf of a three star. And that's a pretty effective way to maneuver inside. Inside the Pentagon. So he invites himself to this thing called the Breakfast Club. What is the Breakfast Club? It is this sort of informal, regular meeting amongst people to talk about, you know, sort of all kinds of things. But it was deliberately set up to sort of provide a discussion space that sidestepped the. The typical kind of endless litany of planning meetings, budgeting, procurement, and all of the other layers, bureaucratic layers that wrap around any and. And, and really stifle technology within, within the Department of Defense. This is also around the time that, again, the Deputy Secretary of Defense at the time was Bob Work, who I, you know, for people who listeners should all know who he is, but he is a sort of legendary figure within this community. I had the chance to speak with him a few times for podcast episodes, engage with him at a conference or two. And I really think, you know, the story of Project Maven would not have happened without Colonel Cukor. I also think the story of Project Maven would not have happened without Bob Work as Deputy Secretary of Defense at the time, because he was fully committed to an idea that has long underpinned the U.S. defense Department's sort of broad approach to doing what it does, and that's offsetting enemy advantages with technology. It did this with the first offset strategy, it did this with the second offset strategy. And now at this time in the mid middle 2010s, there was emerging the. The third offset strategy, which. Which Bob Work was the architect of. So this. This the Breakfast Club. Cukor gets himself well he doesn't get himself invited to. He just sort of shows up and says, I invited myself to. He starts to see this as a space that will be useful for him to do what he wants to do. And then the broader context is that this is happening with a deputy Secretary of Defense who's probably going to underwrite some of the things that he wants to do within third offset strategy. So one other sort of element of this bureaucratic challenge is that Kukor really wants to do something that he calls field to learn. He wants to get kind of, you know, in the private sector they would call it a minimum viable product. They want to get it out, get it tested, figure out what doesn't work so they can iterate rapidly and improve on it to improve the product, rather than doing all of the sort of detailed, precise work on it, to try to put this perfected product out in the field far, far into the future. This is a challenging thing to do in, not only in a bureaucracy, but in a bureaucracy whose business is war fighting, where the stakes are, lives. The idea of, you know, in Silicon Valley, they talk about moving fast and breaking things. That doesn't necessarily work when your business is war and in. When your business is war in a democracy where accountability is key, you know, he is going to be challenged by the bureaucracy on that front as well, probably rightly so. This is, you know, this is the part of the discussion where I find myself most on the side of the bureaucracy because it exists to protect lives and to ensure accountability when lives are not protected. So he's going to run into that challenge as well, [01:10:04] Speaker B: right there. So many good snippets in this book of what exactly that looks like. There's a story of one of the teammates on Project Maven briefing some skeptical operators and their staff about the applications of AI to their operational fight. And he starts getting lectured. I believe it's Joe Larson, but I can't remember who about ia. And he says, no, it's AI, Artificial Intelligence. And he realizes that they're not making a mistake. They mean IA Information Assurance, which is the shorthand in the military. I'm sure many of our listeners know, for the mundane processes by which we ensure that the information touching our systems does not result in a vulnerability or a threat. One of the first things you're asked when you show up to unit is if your IA training is up to date so you can get an account. It was hilarious to know that Project Maven, this now extremely sexy hot topic, encountered the exact same issue. There's another Story from the anecdote about submarine data when Maven is trying to break into the Navy service culture wherein one of the teammates is confronted with concerns about basically, security classification guidance. To your point, it presents as annoying and obstructionist in the book. But these are practices that emerge from real problems. And it's difficult to measure just how useful they are, things like security classification guides, because that depends on detecting what hasn't happened because of them, which we can't do. So we kind of have to implement them on faith. The question I think, that arises in the book is, well, when can you take exception to them or make exceptions from them? Yeah. The other note that I had, John, while thinking about the absolute insistence with which Bob work and Drew Cukor got the Project Maven software out into the operational environment was this distinction that is drawn in this book, Bureaucracy by James Q. Wilson. Whenever I try to understand bureaucracy rather than just complain about it, this is the first reference to which I turn. I cannot recommend it enough. And in it, Wilson spells out this taxonomy of bureaucracies. It's a quad chart. The X axis is whether or not the outputs are observable, yes or no. And then the Y axis, the horizontal, is whether or not outcomes are observable. And what he says is that the US Military, particularly in peacetime, is what's called a procedural organization that occupies the top right of the quad chart where the outputs are observable but the outcomes are not observable. Observable, meaning we can see what we're doing, we can see the procedures we're undertaking, but because it's peacetime and not war, we do not know whether or not the outcome that we desire is being achieved. And that is why bureaucratic processes pile up, according to James Wilson, because for lack of the ability to observe outcomes, it's hard to demonstrate that the bureaucratic process costs more than it gains the organization. And for the inability to do that calculus, they can't be rejected, and so they get added on. This should sound familiar to anybody who's worked in the public sector or any big private sector firm. Well, one way to escape the procedural organization that James Wilson says forms when you have a situation where you can observe outputs but not outcomes is to get out of peacetime is to go to a place where the outcomes are observed, observable. And if you're in the military, the outcomes of concern are operational success and strategic success on the battlefield and in the theater. And so what I see in Drew Cukor's insistence that they get the technology out to the operational environment is not just an effort to get relevant data to train models, but also an effort to get to places where people don't have time for non productive bureaucratic hurdles, that is to say, the places closest to the fight. [01:14:07] Speaker A: Yeah, that's a great point. The, I think to sort of close out this sort of discussion of the bureaucratic challenge and, and the, the challenge that Cukor faced of, of getting a change resistant bureaucracy to change is that, you know, we, we, the narrative is sort of a series of wins. Win after win that Project Maven stacks up against the bureaucracy. The problem is that success sort of breeds a new challenge which is the bureaucracy sort of reimposes itself with, with added layers. We mentioned at the outset that kind of, you know, 2017, 2016, the AlphaGo moment sort of changed our institutional understanding of AI and its importance. It existed this recognition across the department at the time that would continue to grow, it grew much faster because Project Maven was able to accomplish what it did in its first few years, as it did. And DoD sort of as an institution responded. It said it did what bureaucracies always do. It creates sort of additional layers. It stood up other entities. There was the, again we've referred to the Jake, the Joint Artificial Intelligence center, which would go on to become the, the Chief Digital and Artificial Intelligence Office. And you see this battle between the new AI entity, the Jake, and the old AI entity which is at this time what, two years old or something, maybe three years old. But it's sort of the original player in the space and there's a battle that really ultimately takes shape about data. You talked about it in, in Part two, right. Cucorb is pretty adamant that he and his team were not going to share their data with the Jake. And he, he made the argument, was made based on, you know, his argument was that, well, we didn't know if the Jake could be trusted to protect that data and we already have the systems and processes in place to protect it. And, and so it's a, it's a really interesting sort of picture of how you can stack as many wins against the bureaucracy as you want. The bureaucracy will probably come back and reimpose itself and create new battles for you to fight. And again, you know, I, I left this sort of, I left, I left this book thinking both that I was inspired by the fact that, that this small team was able to overcome many of the constraints imposed on it by bureaucracy, but also a little bit grateful that, more than a little bit grateful that we have this bureaucracy because it does play an important role in virtually everything that. That we do. [01:16:49] Speaker B: Right. Because there are costs to behaving the way that Project Maven behaved. And in the aggregate, of course, you put down the book, probably grateful that some of the effort or many of the efforts Colonel Cukor undertook succeeded. But it's hard to run the counterfactual. We can't run the counterfactual, which is to say, had the Project Maven team been a little less abrasive or a little less violent or a little more tactful about its interorganizational or intraorganizational conduct, you know, would we still end up with the product that many people praise? It's. It's hard to know that. [01:17:32] Speaker A: So I think that that transitions nicely into a second leadership challenge that I. That I want to discuss probably quite briefly, and that's one of team building. Yeah. One of the features of military service is as a leader, in almost all cases you're given your team, you don't select your team. And I think that there's a distinction between leadership in that context and leadership when you select your team. Colonel Cukor was given the opportunity to select his team in part because in standing up a new organization, there weren't the systems and processes for, you know, branches to sort of align bodies against requirements. He had money, he had a mandate, and had this challenge of building a team. And so what he did was the quickest thing that he knew how to do, which was basically mobilize a number of reservists. [01:18:20] Speaker B: Yeah. [01:18:20] Speaker A: And I think this is something that maybe a lot of people who even are familiar with Project Maven maybe don't understand. Is that. Or don't know. Is that early on, it was almost all reservists who were assigned to Project Maven because that's a pretty quick and easy way of getting bodies into the. Into the organization. Again, a small team at this time. Cukor had spent his career kind of keeping an eye on other Marine officers that he thought were talented. And so a lot of them were people that he handpicked in many cases, again, reservists, and said, do you want to come on board? There's very, very illuminating references to incredibly short interviews. It seemed AI expertise was not at the top of the list of qualifications required for this. He wanted certain personalities, strong personalities, personalities that could be aggressive, because I think he recognized that's what was going to be required to overcome some of the bureaucratic challenges that we've already discussed. So he had the benefit of hand Picking his team. You know, I've seen MWI is a good example of this. Like the. The team was largely hand selected when it first stood up under the leadership of then Colonel Liam Collins. The dynamic in a team like that is very different than one that the bureaucracy places the organizational systems and processes in place, align bodies against the requirements. The dynamic is very different. It does feel more like a startup culture, which is what he was trying to create. Now you referred to the behavior that the team's behavior. We don't have to get into a lot of details, but suffice it to say Colonel Cukor picked strong willed, strong personality figures, perhaps, you know, not overly concerned with rank. A willing, you know, some of these people had a willingness to break with some of the traditions and norms of deference to rank that is typical of the US Military. He would eventually, as we see later in the book, he would receive a letter of reprimand for it. There was an, an IG investigation, actually I think two IG investigations opened about Project Maven, the behavior of individuals within it, a number of issues. He maintains that he was cleared by both those IG investigations. But the upshot is he did receive a, a general officer letter of reprimand, [01:21:00] Speaker B: which is to say nothing of the personal costs that this sort of workplace imposes on the teammates. Again, we've neither the information nor the interest in parsing the personal lives. But as anybody who is listening knows, these high intensity organizations put a great strain on families too. And those who help stand up Project Maven and get it out to the warfighter oftentimes paid those costs and their families paid those costs. And one, not be frank about that. If, you know, we're trying to parse lessons to be learned from the project's history [01:21:39] Speaker A: and it becomes, you know, it becomes a question of perspective. Because there's one way to look at this and say that willingness to work, what I think they said, you know, 17, 18 hours a day and demand that your subordinates do it as well, was really the only way that Project Maven could have accomplished what it accomplished over this time like that. From that perspective, this is an incredible achievement, a success from the perspective of, you know, people are our first priority, taking care of people and families. It has to be a priority. I, you can't, you can't count this as a success given the burnout that some of the team members, as documented in the book, the burnout that some of them achieved in the, and the personal costs that that were paid because of it. So this ultimately did come with a letter of reprimand. But Kukor clearly adopts the prior of those, the first of those two perspectives because he kind of wears it as a badge of honor and says, you know, look, that was the cost of doing what we set out to accomplish. [01:22:45] Speaker B: Right. And there's reference made to the Apollo Project, the effort to get Americans on the moon, and the famous divorce rate that the Apollo Project left in its wake. The implication almost being that if you want to do great things, this is the price you pay. Again, you can't run the counterfactual and see if had you reduce the strain on the marriages enough to prevent divorce, we wouldn't have gotten to the moon. But I think that, as you said, it's a. It's a matter of perspective. Certainly this podcast aspiration is to do great things without imposing catastrophic personal costs. But I don't think we're putting people on the moon [01:23:25] Speaker A: yet. [01:23:26] Speaker B: Yet. [01:23:27] Speaker A: So the third and final kind of leadership challenge that I want to want to discuss is he faced the challenge of harnessing the press, the private sector. This is not something that the US Military has traditionally, institutionally, it does quite well through its procurement systems. It doesn't do very well at any level below the institution. There isn't scope for individual organizations, individual units typically, to go out and develop the relationships and harness the power of the private sector in the way that Project Maven would clearly need to do. Do you remember the now famous letter, I believe it was in 2018, 2017 or 2018, with the Google employees who essentially said, we don't want to be involved with this Project Maven. We don't want to see the weaponization of AI of our AI products. [01:24:18] Speaker B: Yeah. A well told episode in this book. General Dunford was not amused. [01:24:23] Speaker A: No. And the narrative that emerged from that was that Google employees took a stand and forced the company's hand and forced it out of Project Maven. Well, that's one small slice of the story. The broader story is that Maven would ultimately involve 50 private sector companies, I believe was the number. It's Amazon, it's Microsoft, it's Palantir, as we discussed. It's Scale AI, it's Nvidia, and it's Google. Sure, Google sort of withdrew from the project at the time, but. But in one one part of Google withdrew. Google remains involved with Project Maven today, I believe. So how do you sort of marshal all of those private sector actors and get them on board? US Military leaders are not trained for this and they're not as I mentioned encouraged to do it. Typically it is a challenge of bridging cultures between the military culture and say, the culture of Silicon Valley, which could not be more different. You see, Project Maven, you know, it's, it's a bunch of people who, well, I guess they're not even wearing uniforms. In many cases they wear civilian clothes for, for much of the work. But adopting this culture that looks much more like a hard driving startup trying to get its minimum viable products shipped, you know, scrambling for investment dollars. It, it, it looks like the story of a Silicon Valley startup that just happens to be taking place in a US Government building. [01:25:51] Speaker B: And there's some important decisions that Kucor makes along the way about the extent to which this should remain private. There's this distinction drawn between Fademist and Maven Smart Systems. Without going into too much detail here, Fademist belongs to the government and the software underlying Maven Smart Systems does not. That was the decision that Kukor thought was important. And it seems that in certain respects, Project Maven has won out. I think that as we continue as a military to mature technologically, one lesson that might be useful to draw from Colonel Cukor's example is that it's good for senior leaders to be literate in the ownership structures and the technology and the contract details of, of these companies, because those procedural matters have determinant effects on the battlefield outcomes. [01:26:53] Speaker A: That's a great point. I think that, you know, there was, there was a time when mastery of military arts and sciences was enough to give you a successful career as a military leader, even including up to some quite senior levels. I think that that's not the case now. I think, as you said, it's important that senior leaders are conversant in this. It's important that junior leaders are conversant in this as well and have an understanding of not only the technology, but all of these different layers that, that get added in when you're employing tools that are not developed by the US Military. They're developed in partnership with an organization like Palantir, for instance. One of the other things that Kukor had to navigate was look, at one point they needed basically an interface for the Maven Smart system, right? He wanted Palantir to create it. Palantir wasn't even really that interested in creating it. They had kind of struck out getting a bespoke version of Google Earth for it. He was convinced that Palantir should be the ones to do it. He had to convince this private sector company that they should do it. He also had to Convince offices, individuals across the US Military, who at the time really did not like Palantir very much. Its reputation was. I mean, within the halls of the Pentagon, its reputation was lower than probably any other company in Silicon Valley. So Palantir doesn't really want to build this interface. Nobody else in the Defense Department wants Palantir to. To be the. The centerpiece of this. They're not the prime. At the time, I believe the prime contractor on it. That's ECS Federal. And yet imagine the audacity of what Colonel Cukor said. You know, you, Palantir, you don't want to do this. Nobody else wants you to do it. But I think you're the right company to build this. So I'm gonna, you know, I'm going to lean forward on it. If that works, you look really, really good. [01:28:53] Speaker B: Right. [01:28:54] Speaker A: If it doesn't, you look really, really bad. And that's an amazing decision that he made. And there's layers of it. You know, he was accused of sort of maybe some under the table sweetheart deals, which there's no evidence of. [01:29:06] Speaker B: Right. [01:29:06] Speaker A: He never took a job with Palantir. There are other members of Project Maven that did go work in the private sector, including with [01:29:12] Speaker B: or used to work at Palantir, and then went to Project Maven, which also invited scrutiny. [01:29:16] Speaker A: Exactly. You get the sense, you know, at least my sense of this was Colonel Cukor just had a vision of this, and that included Palantir as a centerpiece of it, and it didn't matter if nobody else wanted that. It's what he wanted, and he was going to make it happen. [01:29:33] Speaker B: Right. And the insight with which the book ends up crediting him is that he, before the leadership of Palantir, saw that Palantir should be an artificial intelligence company. I can't pull up their. I'm not interested in pulling up their branding materials, their website right now, but my instinct is it would say that Palantir is an artificial intelligence company. But the. The story the book tells is that nobody at Palantir thought that before Drew Cukor thought that, [01:29:58] Speaker A: no, they were a data management company. [01:30:00] Speaker B: Right. [01:30:01] Speaker A: So that kind of brings us, I think, to the end of the. The discussion about the difficulty of harnessing the private sector and more broadly, the leadership challenges. But I thought by way of kind of closing out this part three, I found myself grappling, you know, through the 350 odd pages of this book. You know, is this. Is this an effective. Is this a. Is this a good leadership style? And I never came up with A satisfying answer because, you know, I'm curious. Your thoughts. My I. Where I landed on it, is that maverick leadership style, which Colonel Cukor clearly is. There's some, you know, there's. There's discussions that contextualize him within, you know, the space defined by John Boyd and Hyman Rickover, you know, as sort of mavericks who went against the grain and ultimately changed the US Military for the better. In hindsight, maverick leadership style only works, is only effective when it is. When you are a maverick, imagine a US Joint force where every leader adopted the sort of approaches that Colonel Cukor takes. It ceases to work. The entire machinery would seize up, I imagine. And so he was able to accomplish what he was accomplished against the bureaucracy, against the normal order of things, because that normal order of things existed. If it didn't, if everybody sort of did their own thing, you know, pursued their own vision, it would be. It would be chaos. It would be anarchy. I think [01:31:37] Speaker B: we already made the point that it's difficult to know what about a maverick style is useful and what about it is an excess that we could do away with and still retain the underlying good. It's sort of like evolution in the sense that there are a lot of random mutations and you need a lot of time to discern which are useful. These sorts of maverick personalities are necessary, but they're only useful if they exist within institutional context. And those institutions are maintained by stewards who are oftentimes a lot more boring, but in the aggregate have at least as a good an effect on the organization. You know, I think of people like George Marshall, once in a while you have the rare combination wherein one is innovative and risk tolerant enough to see through real institutional reform and then can steward the resultant institution over time. Like, I don't think that Colonel Cukor was ever going to last as long as Hyman Rickover, but in certain ways, Hyman Rickover was as innovative. And I don't think that we can aspire to create many Hyman Rickovers. I think they're a rare type, but they're useful to check our otherwise, at times strong impulse to romanticize to a fault, the maverick. [01:32:59] Speaker A: So that brings us to the end, I think, of part three. Should we move on to back matter? [01:33:02] Speaker B: Yes, let's close out. So as a reminder, Back Matter is the end of the podcast. Just as Ed end of a book. You have back matter. We've talked through the book's three parts. We dealt with front matter. John, what's the first of our back matter? [01:33:18] Speaker A: So the first piece of back matter is margin notes. This is where, you know, there are things that we sort of scribbled in the margins of the book that maybe didn't fit into the discussion, but we thought were worthy enough of note to maybe bring up at the end. Theo, did you have any margin notes? [01:33:33] Speaker B: Just that. But I thought that the book was also fascinating as a history of conflict since basically the start of the 21st century. Katrina, Manson, whether she set out to or not, I think also provides you almost an amusement park ride through the conflicts of the 21st century that captures certain aspects, their essence. Everybody should know. So, of course, Iraq makes an appearance, Afghanistan, substantial, but also Somalia, also Ukraine, also Israel. There's a great series of remarks from General Cogbill, who just left 11th Airborne, but at the time was a deputy commander in centcom, I believe, as a one star, about just how much the war that kicked off after October 7 supercharged the evolution of targeting, for better or for worse. Obviously, that's a separate book. Book. [01:34:30] Speaker A: So I had one margin note that I wanted to draw attention to, and that's. There's, you know, there's a segment, a section in the book where we're kind of given a window into the internal decision making at Google about whether or not they want to be involved in this. And you see the perspectives of a couple of senior Google Google executives, one of whom is very sort of cautious and wonders if they want to be involved with this. The other one is, I believe, the head of maybe like the federal services division of Google. [01:35:03] Speaker B: Yeah, that's right. [01:35:04] Speaker A: Who you get this biographical detail about is from a military family, didn't serve in the military herself, but, you know, saw this as kind of an opportunity to contribute and do a service in, you know, insert. Perform a service for her country, the United States, and she says, we're an American company. [01:35:24] Speaker B: Yeah. [01:35:25] Speaker A: As if that. That alone should be like, you know, a strong enough argument that they should be involving themselves with the U.S. defense Department. And this is a question that I've grappled with for a long time. Like, is Google an American company? I mean, yes, but not in the sense that everything it owns, all of its customers, everything that it involves with itself with, is in the US it has equities elsewhere, obviously. It's the largest search company in the world. It's one of the largest tech companies in the world. It's used in every country in the world. Perhaps maybe not in China. I'm not sure if they're blacklisted there. But when you have a private Sector company that is huge, multinational, perhaps has large investors from different countries who maybe have their own national interests in mind. You know, we talk about World War II, the total mobilization of the US economy being effectively one of the things that, that, that won the war for the Allies because there was so much latent strength within the economy that if it was just pivoted and put on a war footing, would overpower what Germany and the Axis powers could, could produce. You know, that looks different in a globalized world. Mobilizing. We see this mobilizing the economic power of the. The main private sector actors within your country. Looks different now where, you know, where a company like Google perhaps has some more reservations where we're an American company, ergo, we should do this is not as strong an argument as maybe it would have been 80 years ago. [01:37:04] Speaker B: Totally. Shall we now consider our saved rounds? These are fun facts, little details that we didn't have any opportunity to share but struck us as worth mentioning. John, did you have a saved round you wanted to share? [01:37:21] Speaker A: Yeah. We're also introduced in the latter part of the book. The bulk of our discussion so far has talked about the Project Maven, Colonel Cukor's team developing this system. One of the biggest moments for it to bridge. We talked about the gap between intelligence and operations and that they were really focused on the intelligence side of the house for the. In the initial years, the first opportunity to sort of migrate it over to the operational side was when the army got involved, specifically the 18th Airborne Corps. And if. If Colonel Drew Cukor was a key figure in creating the Maven Smart System and standing up Project Maven, making it happen and accomplishing what it did. There's an army officer, Joe o', Callaghan, who was a key individual in bringing it over to the operations side. He's a field artillery officer in the army, was assigned to 18th Airborne Corps, and in 2022, I believe, kind of brought Maven Smart system into what 18th Airborne Corps does from a. From. From his. In his lane, from a fire's perspective. In the book when, when. When Katrina Manson meets him, she's describing him and she mentioned. Mentions that he owns a bike shop. Shop. Because he uses kind of references to cycling and F1. He brings cycling and F1 cyclists and F1 drivers into 18th Airborne Corps to kind of meet the team. And anyway, they mentioned that he owns a bike shop. And I went down the rabbit hole looking for it and I found out that it's in Fayetteville, North Carolina. It's called Irish Cycles, [01:38:55] Speaker B: and [01:38:55] Speaker A: it [01:38:56] Speaker B: Just, [01:38:56] Speaker A: it struck me that I'm into biking, so I found it kind of interesting. I guess that's why I went down the rabbit hole. But, but Joel Callahan was also a very hard charging, hard working, you know, person behind like that was necessary to bring this into the army, the army fires community. It just, it was sort of a reminder that you see somebody working these long hours and you know, seemingly devoting their life to something, but that every member of the US Military is also a person. And then outside of that, no matter how much how dedicated they are to the work, work, there are sort of layers to that personality. And I actually think that's just, it's so interesting. It's one of my favorite things is when you know, you know somebody professionally in the US Military and you work with them maybe every day and then you have an off site or then you're at, you know, your TDY with them and so you're out to dinner and you started get to meet them as a different individual and you sort of, kind of understand what makes them good at their job because of these other layers of, of their, of their, their personality and who they are. [01:40:00] Speaker B: First of all, that is fascinating. I'll have to stop by next time we're in Fayetteville. It's still open. Second, the saved round I have actually kind of dovetails with the point you just made. There's an anecdote in the book about an early use of the technology associated with Project Maven to conduct this strike on a. A route by which explosives are moving through a corner of Afghanistan by donkey. And basically the automated computer vision detects the pattern before the human does, and it ends up in what appears to be a successful strike. John, you already mentioned Bob work. There are probably a couple other people, you know that appear in the book either by name or not. I have a buddy who was present for that operation and I asked him, did the team realize that you were dealing with a technology that would soon come to define the way we conceive of the future of the American way of war? And he said, not at all. And Katrina Manson makes this point several times to her credit. He said, we thought it was sort of a footnote. It was just one of many systems that passed through an AO with contractors who are trying to. He didn't say this. I'm now kind of elaborating, but try to pull money out of the US Government and give soldiers stuff they don't need. Everybody has this story about being handed kit that just distracts from the mission because somebody's trying to make a buck off the Pentagon. I think that it's humbling to hear that because it's a reminder that you might be coming into contact with, with things that will change the world without you realizing it. But it was also humbling in the sense that I was coming into contact with a friend of mine, that it was about a major development in the history of war. And I don't know this person primarily as an innovator. I know him for all of his hobbies. [01:41:58] Speaker A: All right, should we move on to bonus reads, our final, final section of our back matter. Theo, what's your bonus read for this month? [01:42:06] Speaker B: Oh, a book that I'm, I'm currently reading, or rather that I just finished and very much enjoyed, is the Master and His Emissary by, I think you say, Ian McGilchrist, which is a part neuroscience, part cultural history book. It's huge. The author is trained in both neuroscience and literature and philosophy. That makes the argument basically that the development of Western culture can be understood as the competition between the, the left and right hemisphere of the brain. It's a very ambitious read. It makes some very big claims about the nature of the 18th and the 19th and 20th century and how they're faithful to our neurological composition. But I found it to be mind expanding and I recommend it to the listeners. Too long probably for this podcast, but a great read. How about you, John? What are you reading? [01:42:55] Speaker A: So my bonus read for this month is related to the discussion and Project Maven. There is an organization at Georgetown called the center for Security and Emerging Technology. They wrote a report called Building the Tech Coalition How Project Maven and the U.S. 18th Airborne Corps operationalized Software and Artificial intelligence for the Department of Defense. It's a long and wordy subtitle. I had a chance to speak with Emmy Probasco, the lead author on the report, a couple of years ago for, for the MWI podcast, we mentioned Joel Callahan. We very briefly alluded to the 18th Airborne Corps adoption of the Maven Smart system. That story in itself is fascinating. This report tells it and I guess more broadly this is an appeal to, to listeners. You know, I read short form articles all day long, basically, you know, three minutes, five minutes, eight minutes while I'm at my desk, while I'm, you know, waiting in line at, you know, ordering Chipotle. Like I'm constantly reading those. I also read books because there's a sense of gratification of completing a book and adding it to your shelves and, you know, but there's this middle ground of things like think tank reports that often I think get overlooked from a professional development perspective. And I would say don't overlook them because like this report, if you're, if you're an army officer, you will be working with this system. And this is incredible background of how it was brought into the army. Again, it's from cset, the center for Security and Emerging Technology called Building the Tech Coalition. I think it was 2022 or. No, it came out in 2024. Yeah, that's my bonus read. So that brings us to the close of this episode and our examination of Katrina Manson's excellent book Project Maven. It's been shortlisted I believe already for the 2026 best book by the Financial Times. I suspect it will be on the same economists and all the other places that you know give awards for the best non fiction books in this kind of genre of current events. It's an excellent book. I enjoyed the discussion. Theo, what do we have upcoming on the next few episodes? [01:45:06] Speaker B: Next month we're going to get into the holiday spirit with a read of John Steinbeck's forgotten novella of Nazi occupation of Scandinavia, the moon is down, there's some snow involved if no Santa, and it's a much loved classic within certain corners of the military history and national security community. [01:45:26] Speaker A: With that, listeners, before we go just one final time, if you're enjoying the MWI Podcast, we'd like you to consider leaving a rating, giving it a review at Apple Podcasts or whatever your your favorite podcast platform is. And more importantly, share it with your peers, share it with your subordinates, share it with your your battle buddies. If you're the type of person that reads books, that likes books, that talks about books, and you know other people that are too, shoot them a note, shoot them a text, shoot them an email, spread it by word of mouth and help us get the word out about the MWI Book Club. [01:46:05] Speaker B: Thanks for listening to the MWI Book Club. [01:46:08] Speaker A: If you enjoyed this episode, be sure to subscribe wherever you get your your podcast so you don't miss future episodes. [01:46:13] Speaker B: And please take just a second and leave a rating or review, which really helps new listeners to find us. [01:46:19] Speaker A: And one last thing, what you heard in this episode are our own views and don't reflect those of West Point, the Army, or any other agency of the US Government. [01:46:28] Speaker B: Thanks again.