Defense, drones, and military procurement, with Bean of Naval Gazing

Patrick McKenzie Aug 28th, 2025

A pseudonymous defense industry expert explores the intellectual crossovers between military and civilian domains.

This week Patrick is joined by Bean, a pseudonymous defense industry expert, to explore the intellectual crossovers between military and civilian domains. The conversation reveals how the defense industry's fundamental constraint of having only one customer (a monopsony) creates entirely different incentives than tech, leading to conservatism and 30-50 year product lifecycles. Bean argues that drones are largely modern iterations of cruise missiles we've had since the 1950s, and explains why current anti-drone defenses make swarm attacks less threatening than headlines suggest.

[Patrick notes: As always, there are some after-the-fact observations sprinkled into the transcript, set out in this format.]

Sponsor: Mercury
This episode is brought to you by Mercury, the fintech trusted by 200K+ companies — from first milestones to running complex systems. Mercury offers banking that truly understands startups and scales with them. Start today at Mercury.com
Mercury is a financial technology company, not a bank. Banking services provided by Choice Financial Group, Column N.A., and Evolve Bank & Trust; Members FDIC.

Timestamps

(00:00) Intro
(00:29) The overlap between tech and defense
(01:35) Operations research in World War II
(02:55) Mathematical insights and military strategies
(05:28) The role of operations research in modern warfare
(16:59) Tech and defense (Part 1)
(19:48) Sponsor: Mercury
(21:00) Tech and defense (Part 2)
(26:07) Economics behind the defense industry
(32:07) SpaceX's early challenges and achievements
(33:00) The Super Hornet development story
(34:39) Military procurement lessons
(37:42) Aerospace industry retention rates
(38:42) Lockheed Martin's dominance and supply chain
(40:55) Drone technology and military applications
(46:53) Anti-drone defenses and future warfare
(48:01) Naval warfare and historical perspectives
(01:01:03) Wrap

Transcript

Patrick: Hello everybody. My name is Patrick McKenzie, better known as patio11 on the Internet, and I'm here with Bean, who uses a pseudonym. Bean is the author of the Naval Gazing blog and works in the U.S. defense establishment. Thanks very much for coming on the program today.

Bean: Thank you so much for having me.

The overlap between tech and defense

Patrick: So I guess first is a bit of an apology for why we're discussing military affairs on what is usually an infrastructure podcast. But I wear a t-shirt somewhere that says I'm on team "make things people want." [Patrick notes: I said this metaphorically rather than to invoke a literal t-shirt, and then remembered this is literally a t-shirt in YC, and is granted to people for an accomplishment they esteem. I have no such t-shirt. Apologies.]

And I have a lot of respect for people who are on team "kill people and break things." [Patrick notes: I say this mostly affectionately rather than ironically, despite the awkward humor. War is terrible, militaries are a regrettable necessity, I am thoroughly in favor of ours being good at what it does, and a respect for that means foregrounding the reality of it.]

But the professional overlap is usually not that much.

My pitch for people who are in the tech industry or in finance or similar, about why they should pay more attention to defense: there is historically and currently much more intellectual overlap and intellectual exchange of ideas than might be broadly appreciated. People might know that the Internet was originally a DARPA research project, and they might know that operations research was originally created to win wars, and even that the entire field of logistics was created to win wars. And then the rest of us got to use it anyway.

I also think that history and the study of it are intrinsically fascinating. So I'm happy to explore some of that with you.

Operations research in World War II

Bean: Yeah, of course. I mean, we are both fans of operations research, so I guess we should probably start there.

Patrick: Sounds reasonable. So operations research for people who don't have an undergrad degree in the subject—which is mostly everybody—is a particular branch of essentially applied mathematics that largely comes out of World War Two. And everybody invented it at various times in various places, relatively independently of each other, with some sharing of notes.

And then—I think this is a bit of an exaggeration—the United States entirely forgot about it after the war. But Deming, a gentleman who was instrumental in convincing the Japanese automobile industry, among others, to adopt it after the war, was considered broadly transformative to that industry. He got various medals for it, and then Japanese industry ran the table on the U.S. for a period of a couple of decades.

And sometime around the late 80s, early 90s, the United States was like, "Oh, statistical confidence intervals and manufacturing processes. That is an interesting idea. Perhaps we should think through it. Oh, did we invent that 50+ years ago? Shoot."

[Patrick notes: A relatively minor piece of the puzzle which shows the general shape of the hole: if a manufacturing process is well-controlled, then you might expect output of it to vary along certain physical dimensions randomly. A newly produced part should be the length you expect it to be, plus or minus an error bar. If you were apprised that variation in parts was suddenly non-random, you would think that process is out of control, and seek to bring it back into control.

You can teach a bright high school student quick heuristics to tell whether a series of data points is randomly distributed or not, rejecting the null hypothesis of random distribution if, for example, they see three consecutive data points which are more than 3 standard deviations out. We do this every year in Statistics 101 classes.

Japan, Inc. did this every year for people pushing brooms in factories. If your sweeping takes you past a quality control station with three consecutive Xes above the topmost line, stop sweeping and immediately push the Big Red Button. Someone will come running. You will not get in trouble.

The social situation of people who sweep floors has changed a bit in the Japanese economy, but andon cords and their modern analogues live on.]

So anyhow, that is my fun story, but I've told that in various places. I understand that you have a fun story about operations research.

Mathematical insights and military strategies

Bean: Yes, several. So, looking at one specific field in the Atlantic during World War Two—this was probably the height of operations research in that first era—in the battle against the U-boats. [Patrick notes: Submarines. (I was, a long time ago, something of a WWII history buff, and will take the liberty of supplying some background information for listeners/readers who might not have had the opportunity to reach that chapter in high school history, since it is often skipped at the end of the year.]

Actually, I will pose this as a puzzle to Patrick so that we will see what he makes of this. Originally the British, when they were setting their air-dropped depth charges—for those who don't know, a depth charge is basically a chunk of explosives set to go off at a specific depth, this is how you attack submarines if you don't have homing torpedoes—they did some analysis and they said, "On average, when we spot a U-boat, it's been out of sight for 50 seconds and it dives at two feet a second. So we'll set them at 100 feet." But the kill rate for attacks is only about 1%.

They managed to make one change that raised the kill rate sevenfold. Any idea what it was?

Patrick: Oh, this sort of thing is always an excellent intellectual puzzle, and I'm embarrassed I can't just cite this from memory, because I was something of a World War Two buff when I was in middle school. [Patrick notes: Ironically this had basically nothing to do with me deciding to immigrate to Japan after university. My intellectual fascinations tend to attack with all-consuming intensity for a season or a few years, then recede.]

All right, so they're probably downstream of some conditional probability nonsense, which means conditional on them losing sight, they are getting a greatly non-representative sample of U-boats. So obviously they're going to make it longer than that. The question is how much longer?

And this puts me in mind of the other operations research result, which is don't armor the places that you observe airplanes getting shot—armor the places you don't observe them getting shot, because those are the airplanes that didn't make it home successfully. [Patrick notes: This is the commonly-seen-on-Twitter plane with chickenpox meme to encapsulate survivorship bias. (A useful intellectual concept which, since its commoditization as a meme, has probably lost most utility in enlightening public discourse.)]

Okay. So I don't have a theory of the mind for why they are getting into the wrong branch of conditional probability. Can you spoil me? [Patrick notes: Reading this after the fact it is somewhat embarrassing that I didn’t ask clarifying questions about whether the depth charges were being deployed by ships or planes.]

Bean: Yes. So what happened was this. The problem is, a typical depth charge has a lethal radius of about 25 feet. But once the submarine starts to dive, you lose track of it very quickly. And they checked, and in about 40% of the cases, the submarine didn't spot the aircraft until it was very late. And it was basically still on the surface when they dropped.

[Patrick notes: As a bit of background knowledge for you, in WWII all submarines were diesel submarines. Diesel submarines spend most of their time on the surface, submerging only to attack or briefly when threatened. This is easy to understand once you realize that they had ancient battery technology and combustion engines. Nuclear submarines, which dominated Cold War tactics and strategy (and are quite well-represented in fiction), can stay submerged for prolonged periods with the need for it. (And WWII geeks might mention “Well technically late in the war we invented snorkels for submarines, and so they could spend much more time under the water, just not under very much of it at all.”)]

Bean continues: But in that case, the depth charges were set too deep. In other cases, big ocean, small submarine, you don't really have a chance. So they cut the depth charge setting to 20-25 feet. Said if it's been out of sight more than X seconds, don't attack. And they got a sevenfold increase in kills.

Patrick: Okay, so exactly the opposite of what I predicted. And this is why we hire smart people with domain expertise.

Bean: I mean, I'm sure you would have been able to figure it out if we laid it out a bit better.

The role of operations research in modern warfare

Patrick: But I guess the—and this is a repeated refrain across the histories of World War Two—the people who are actually fighting in the war are supported by a bunch of, without loss of generality, math PhDs that are back at headquarters. And the math PhDs very frequently do come up with things that by the standards of, say, modern research papers, look like toy models. But those toy models accurately describe the world that we live in and in some cases, more accurately than the equally toy models that were developed by people who literally had their fingers on the triggers, as it were.

And then the application of those toy models in a tight loop with the end user does result in the end user being more effective at—euphemism—engaging the enemy. Killing people and breaking things. Is that what is happening?

Bean: Yeah. And I mean, you saw just a ton of different applications in different fields. Everything from—there's a story in one of the foundational texts about an analyst who shows up at a new base, notices that there are two tubs for washing dishes. Everybody had to wash their mess kits. There are two tubs for washing dishes, two tubs for rinsing. He says, "Hold on. It takes twice as long to wash as to rinse. Why don't we turn one of the wash tubs into a rinse tub?" And the line disappears. Everything from that to how best to attack submarines or where to put armor on the planes.

Another well known one was convoy size. The question they had was, okay, how big should a convoy be to go across the Atlantic? And originally they had thought, okay, we looked at the math. It turns out the number of ships we lose per convoy is basically constant with convoy size. So having more escorts—the advantage of bigger convoys theoretically is there are more escorts to attack submarines. That doesn't seem to be working. Does that make sense?

Patrick: The advantage of having more size is that you have more escorts for submarines. I find myself expressing doubt here, and I'm thinking of things like, okay, the convoy is distributed in a two dimensional space. And so there is intrinsically going to be a bit of a square law here with regards to how many escorts a submarine encounters.

So the submarine certainly doesn't know the entire size of the convoy fleet or the number of escorts they have. They don't care about that. They only see their view in two dimensional space or three dimensional space. And the perimeter gets—oh, goodness, math—linearly larger with the radius, the size gets larger with the radius squared. So you've got something like O(sqrt N) increase in required escorts in response to increasing convoy size, maybe.

[Patrick notes: This would imply “as few convoys as you can get away with (for other reasons), as big as you can make them.

I was using an intuition which I am not sure represents WWII tactical reality, but which seems plausible to me: submarines are much more vulnerable to detection by ships they are close to than by ships farther away, because the primary technology which detected submarines with the Mk 1 eyeball. Or, option number two, exploding.]

Bean: Yeah. Well, and that was the theory, and it didn't look like it was working because they said, "Well, hold on. So we have a higher escort to perimeter ratio with bigger convoys, but we're still losing—it kind of doesn't matter the size of the convoy. We're still losing three ships per convoy." And then they realized, "Hold on a minute. But if the convoys are bigger, we have fewer convoys and fewer losses overall." But that took them, I think, a year plus to figure out.

Patrick: Yep. It's interesting. I was looking for second or third order effects, like if you have more convoys, then your convoys are departing port more times and maybe it's more difficult for the enemy intelligence apparatus to guess exactly where they're going to be at a particular time. [Patrick notes: Immortalized by wartime counterintelligence propaganda, like “Loose lips sink ships.” posters, and absolutely a real operational risk.]

And maybe that gets more through and then avoids the first order effect of like, wait, if we pay a flat three ship tax per convoy, we should choose to send fewer of them.

Bean: Yeah, yeah.

Patrick: Okay. So just briefly, voicing over something you said earlier, you mentioned that there was the wash tubs versus rinsing tubs and an analysis of where the bottleneck actually is in the system. And that was kind of uniquely motivated by wartime. But also with regards to industrial production generally, one of the core intellectual insights of operations research was you don't just have to finger-to-the-wind all this stuff, you can model systems effectively. Then you can predictably find where the bottlenecks are and adequately size the system prior to turning it on to design those bottlenecks.

This has second-order implications, too.

If you know where the bottleneck in the system is, you should certainly want to saturate that bottleneck versus having built yourself into a position where you are not capable of saturating the bottleneck of your own system.

[Patrick notes: This phrasing is a little confused because, by definition, the bottleneck is the bottleneck because it is being saturated. Let me try again now that I have the benefit of a word processor: you would intelligently choose to have the resource which is scarcest or most expensive for you dictate your bottleneck, and you would maximize your usage of that scarce resource. If, for example, you have a lathe in the factory, and you can get no more lathes because there is a war on, you want to push the utilization on the lathe very high. (But perhaps not to 100%, because of increased wait times; queueing theory, it is a thing.) If your lathe is routinely not lathing because you did not hire sufficient unskilled labor to bring materials to it, one way to phrase that is “Your new bottleneck is unskilled labor, not the lathe.” Another way to phrase it is “You are very bad at your job.”

For examples of people who are bad at their job, see the discussion of bottlenecks in government service provision in Recoding America. (Out of respect to the author, I will note that she goes out of her way to disavow this conclusion. It is, to me, inescapable from facts presented about the performance of some identified public officials.)]

Patrick continues: Which that's the sort of thing where it was really worth papers and PhDs a number of decades ago. And now Factorio players understand it.

Bean: Oh, yes. I have definitely once implemented a thing in a Factorio mod that was deliberately inspired by that production system.

Patrick: So do you have another fun go-to from the history of operations research to make me look like an idiot on the spot? I'm happy doing that as many loops as possible.

Bean: I think I've got one more and then a book recommendation for you and anybody else who's interested in this.

So when they were looking at arming merchant ships—and they said originally, okay, so if we mount guns on our merchant ships, we discover they only shoot down about 4% of attacking planes. And the guns cost more than the planes we're shooting down. So this is a bad idea, right?

Obviously that is not the case given that I'm posing this to you.

Patrick: Yeah. Well, one, there's an OpEx versus CapEx distinction trivially here where you mount the gun on your merchant ship and maybe it's more expensive than the number of planes that it scores in week two. But you hope that the gun will be in service over a long period of time. Maybe you have a high discount rate because the war has an end date, and you hope to not run it until guns literally melt.

And then some deterrence effects. But having learned that, it's always the first order answer and it's the first order answer that I'm trivially missing.

[Bean interjects, after the fact: Interestingly the ongoing OpEx cost of the gun likely dominates the total cost of ownership, because you need trained crew to man it constantly for the service life. The search term is “naval armed guard.”]

Bean: You actually already said it—it was the deterrence effect. Basically, they discovered it cut the chances of being sunk from 25% to 10% for level bombing. People don't like being shot at. It turns out the point of the gun isn't to shoot down the planes. The point of the gun is to make them not want to attack you.

Patrick: Makes sense. And there's human factors there where, given an infinite amount of time to engage the target, you pick the best possible way to engage it. [Patrick notes: Angle of attack, flight speed, number of passes to make, and so forth.] And maybe you have more success as a result of that. Given the same set of orders, plus your desire to live to fight another day, if there is flak or similar coming from the ship, you don't have the ideal angle of engagement. You have one which you think will give you a chance and allow you to live again, and you accept the lower possibility of succeeding there.

The other combatant would certainly like to force you into that.

[Patrick notes: There are many, many practices in the military (and perhaps other competitive domains) which are not about achieving your aims so much as frustrating the ability of your opponent to achieve theirs on their own terms. One trivial example: suppressive fire, which is when you shoot at someone with the explicit intent that you do not care if you hit them. Hitting them is bonus points. The thing you are doing is denying them the ability to direct aimed fire at you (and to maneuver around you, and so forth).]

Bean: Yes. Very much so. Although, actually, speaking of this just reminded me of another fun operations research story. It turned out when they analyzed the kamikaze campaign at the end of—towards the end of the war, the U.S. Navy's anti-aircraft guns—these were on warships, not on merchant ships—had gotten so good, they discovered that the deaths per bomb hit were actually lower for kamikazes than they were for conventional attacks.

Patrick: I will be careful about what I say here because this is politically consequential in many places.

[Patrick notes: I am greatly indebted to a well-governed democratic nation who is one of our truest friends. Also, I am specifically indebted to the immigration authorities there, who zealously pursue their statutory authority to, and this is a quote, forcibly expel undesirable foreigners from the nation. Now guess what government post sounds particularly appealing to young, talented people of a certain, say, ultranationalist persuasion?]

Patrick notes: But it wouldn't surprise anyone. Though tough to know this through the fog of war at the time, we have the benefit of history afterwards. And given the operational dynamics of that campaign, recruiting people who were in many cases not very trained as pilots of the planes, because you didn't need to educate them all that much to successfully get the bomb to the target.

And so they had much less of the sort of knowledge base that more experienced pilots were doing with conventional bombing. Also self-destructive choices were made all over the command structure back in the day. But convincing the most successful pilots that rather than bringing up the next generation of pilots or continuing to fly missions, they should immediately go on a one way mission, was not an effective way to maintain their long term ability to fight the war.

Bean: So the funny thing is, they largely didn't do that. [Patrick notes: I think we might have a failure to communicate here but, with appropriate deference to an expert in their element, I do not think what I said is inaccurate on this point in particular.]

And this is actually another quasi-O.R. thing that your comment reminded me of.

One of the things the U.S. discovered during that campaign was that if they could get fighters to attack the kamikazes when they came in formation, and they discovered that they could get fighters to intercept before that formation was in range of the fleet, it was basically ineffective—nothing happened. If they got within visual range of the fleet, there was a problem because it turned out there was one guy in that formation who knew how to navigate, and his job was to turn around and go home after he got everybody else in visual range of the fleet.

Everybody else, they knew how to take the plane off. They knew roughly how to control it in the air. They probably weren't that great at landing, but they didn't need to be.

Yeah. I mean, the mismanagement of the Japanese pilot training pipeline is a whole different story. [Patrick notes: See above regarding “We might have been talking about two slightly different takes on the elephant.”] That was largely just done in by the fact that they didn't train people and they didn't have a rotation system. The U.S. would rotate pilots—by the time you flew for six months or a year, then you go home to train people. And the Japanese just never did that. You stayed in the frontline till you died.

Patrick: History never repeats, but it rhymes. There have been recent conflicts where that has been part of the reporting, although difficult to tell given fog of war and similar. But the conflict in Ukraine has different doctrine between Ukraine and Russia, and perhaps different attitudes with respect to who makes decisions, how you train those people, etc.

[Patrick notes: I trailed off before coming to my point:

It is my impression that the U.S. stacks against favorably against many armed forces in many times and many places on the specific issue of “Who is capable of maintaining human capital in the armed forces, including at the officer, non-commissioned officer, and line troop levels?” And while some servicemembers might roll their eyes at this, knowing what they’ve seen with them, there are very many armed services which are not designed or operated to win wars. Then they get into wars and, well.

I will avoid identifying any patronage networks masquerading as armed services because it is socially untenable, but while the Chicago Public Schools has many sterling qualities, they would be singularly ineffective at armed conflict… with a small list of near-peer adversaries, that is.]

Bean: Yeah, very much so. And that theatre is not an area I'm not a specialist in.

Oh, right. The book recommendation I mentioned earlier and forgot—the book is called "Blackett's War." It's a popular account of the air campaign in the Atlantic. Recommended for those for whom the discussion of depth charge setting is interesting, and if you want to learn more.

Patrick: And I will drop some classical links in the comments to various books that talk about operations research and the Deming stuff and similar. [Patrick notes: Out of the Crisis, etc etc.]

Bean: Yeah. And I have a blog post on—I was basically just reading out of my blog post on operations research on the Atlantic for those examples, except the kamikaze stuff.

Tech and defense

Patrick: So transitioning to a larger topic, every time two communities of practice meet, they tear their hair out about each other. Tech and journalism: "Why don't these journalists understand what they're doing? Why do they keep making stuff about us? Tech and policy: Why are those idiots in Washington trying to regulate something they don't even understand? Have you ever even met a financial regulator who can diagram an order book?"

Spoiler alert: I can't say with certainty that I've ever met a financial regulator that could diagram an order book. (I absolutely can diagram an order book. Different conversation.)

But that doesn't mean—and having talked to a few people over the years in the tech industry who have probably populated government, and a few people over the years in the military and civilian spaces that are adjacent to the military, there's a little bit of "Why don't the civilians understand what we're doing?" going on.

So why are we idiots? Why don't we understand what you guys do for a living? Or, rather, in what way don't we?

Bean: I mean, yeah. And in fairness, I have similar feelings about journalism with regard to defense issues, too.

Patrick: I think this is the human condition.

Bean: Yes. Everyone dislikes journalists. No, that's not right.

Patrick: I was going to go with “Everyone thinks that every other group of smart people doesn’t understand their local concerns as well as they understand their local concerns,” which I think broadly rounds to true.

Bean: Okay. So some of it is the defense industry is big and specialized and is operating under constraints that tech very much isn't and vice versa.

The big thing we have around—we have one customer. We only sell to the federal government. We don't have—we just do not have other people we can go sell to. Which—the technical term is monopsony. This just changes the whole structure of the industry because, among other things, it means, for instance, firms tend to be pretty reluctant to do things on spec. It'll happen if it makes sense. I have an example of that, actually. But if the government wants to buy a fighter jet, you're going to ask them to pay to develop it because, well, they're the ones who want it and they're probably going to change their mind a bunch.

Requirements development is hard, and especially for something as complex as a jet fighter. And you also have things like you don't know how many they're going to buy. Maybe the Chinese economy—but maybe China has a revolution and becomes our friend. [Patrick notes: I’m aware that this presently sounds unlikely, but a lot can happen in a few decades. Two ago it was the declared goal of the Western political class to integrate Russia into the global financial order. A few more than that, several people who attended our wedding probably bet against the possibility of there ever being that sort of wedding.]

And suddenly all this—you spent billions and billions of dollars developing your fancy new fighter. And now they're like, "Sure, that's cool. We'll take 12."

Patrick: Can I bring up a factor which is probably so obvious to you that you wouldn't mention it, but which probably surprises people in other domains. The cost to develop a new line of aircraft carriers, line of submarines, line of etc.—there is some amount to that which is front loaded into R&D. And then there's a marginal cost of +1 unit that you buy. But the lifetimes of those units are 30 years, 50 years, possibly longer than that. Do you have a representative ship or airplane or similar off the top of your head, and how long has it been working for us?

Bean: Yes. I mean, yeah, lifetimes—30 to 50 years is a pretty good representative lifetime. To take an example, the USS Nimitz is an aircraft carrier, the lead ship of the Nimitz class. For those who don't know, "class" is basically a bunch of ships built to the same design. There are details there that we don't need to get into.

Yeah. She was commissioned in 1975, is currently on her last deployment out somewhere in the Pacific, or possibly the Indian Ocean.

Patrick: I think the Nimitz was a star character in a science fiction movie which I remember from my childhood days. It was 1989, and it was already the globally recognizable ship that was capable of being a character in a science fiction movie in 1989. [Patrick notes: The Final Countdown (1980) was old enough to be in syndication on television, which is how I saw it, in 1989. The Nimitz itself was not old, as of the time of theatrical release.]

And then it's been a few years since 1989.

Bean: It has. Yes.

Patrick: And I think some second order consequences of that are that if you are attached to the Nimitz project—perhaps not as crew on one ship that is in the Nimitz, but you're one of the naval engineers that is responsible for the blueprints that make up the intellectual property that is the Nimitz class—that would be the only thing you work on for an entire career, right?

Bean: Yes. Very much so.

I mean, the last—the Nimitz class was the George H.W. Bush commissioned in 2009, so expected to serve through 2060. There are probably people starting to work on supporting her today who plausibly will retire with her.

The turnover is much lower. And there's a lot of conservatism just because the industry broadly has to deal with—if Google screws up and wastes $1 billion, nobody cares except Google shareholders. If the military screws up and wastes $1 billion, Congress asks questions. And that's not a place you want to be.

Patrick: Also, to voice the obvious point, but the obvious point is that you can screw up in ways that a lot of people who are sympathetic to us get killed. In doing so, you fail the primary mission and also fail at secondary missions, like, "Oh, an administration is now extremely embarrassed," etc.

[Patrick notes: I am aware that one distinction between the military and civilians is people who came up in the military will want to correct me now and say “Force protection is not, has never been, and can never be the primary mission of an effective fighting force.” I concede the point.]

Patrick notes:And so the old saw is that generals are great at fighting the last war. But the procurement bureaucracy also is fighting the last crisis, and adopts regulations and scar tissue to avoid previous crises.

Bean: Yeah, I actually sort of disagree with the line about generals fighting the last war. I don't think that's been true for at least probably a century, broadly speaking. Sometime around—I was probably before World War One. I'm speaking specifically in the naval context, which I have a lot more information on that far back. There were a lot of people in various navies before World War One who saw that things were changing and were very interested in how they were going to change and thinking about it a lot. [Patrick notes: This does not comport with my understanding of the early history of naval aviation, which I note with some surprise, but perhaps Bean and I have a difference in emphasis. We briefly return to this later in the episode.]

And that has only gotten more prominent since then. And to be clear, I'm not saying that they always get it right. Often they don't. Predicting the future is hard. I just—these days I tend to find "fighting the last war" being a thing said by people who want you to adopt their version of what the next war will be instead of someone else's.

But again, as you point out, and this is kind of one of those things where I didn't even really bring it up because it's just ingrained in the culture. Yeah, ultimately, we have people's lives on the line in a way that most people don't. The medical industry does. People who build airliners do, maybe some automotive people. I guess most civil engineers, but not tech and finance generally.

Patrick: It's not generally top of mind when you're creating a web app. Though: I think there's been more awareness in the tech industry in the last couple of years that oh, shoot. We might not have set out to accumulate huge amounts of power and ability to impact the physical world. But we find ourselves having huge amounts of power and ability to impact the world. And I think there is more awareness these days that a Python script could have a body count associated with it than there was an arbitrary number of years ago.

[Patrick notes: This was absolutely a top-of-mind concern from the earliest hours of the VaccinateCA effort, and the several team members who voiced it early and often were not coming from an intellectual or corporate heritage of “nowhere in particular.”]

Economics Behind the Defense Industry

So you want to go on to the economics behind the defense industry?

Bean: Yes.

One of the problems you run into in looking at different stuff is how much does a thing cost? We'll take the F-35 program. How much does an F-35 cost? Well, I can't give you a single number. I mean, if you want to say if I took out my credit card and went to Lockheed Martin and said, "Please send me one," and they were willing to—obviously you can't do that—it'd be about $90 million for one plane this year. But you've also seen the number $2 trillion being bandied about recently. And this number is also not wrong. It's got a bunch of different numbers in it.

So that $2 trillion is complicated—for calendar year 2012, because the other thing about long programs is you run into a lot of issues with inflation. And just other things change. How many you're building a year changes how expensive they are per unit. So if you're looking at the first figures that come up on Wikipedia, those might be from a year when we were building a lot of these planes and we're not anymore. And so the price is twice what it was, especially if that was ten years ago.

So for 2012, the F-35—it was $67 billion for R&D, $221 billion for procurement. It's about a third higher if you use current dollars, not inflation adjusted—$400 billion or so for procurement and R&D. And then $1.6 trillion for sustainment, which is everything from the salaries of the pilots and the fuel to engine overhauls and spare parts, to paying junior software engineers whose grandparents are currently in high school because they're planning to fly this thing through 2088.

Patrick: I’m confused about hiring a junior software engineer whose grandparents are currently in high school.

ponders

Okay. Got it. So you're amortizing the future costs in the current projected numbers.

Bean: Yes. And there is value in doing analysis like this, especially at the start of the program, when it's a case of, "Okay, this will cost us extra dollars to do this thing that takes—doing it this way takes more money upfront but produces less maintenance. This way is the other way around. Which one's going to be better in the long run?" Because yeah, we're expecting to fly this for 50, 60, maybe years.

I mean, if you want even longer than that, there's the B-52, which looks like it is very plausibly going to be in service at the 100th anniversary of the first flight of the first one.

Patrick: Oh, boy.

Bean: Yeah.

Patrick: I guess one kind of level-setting question—I have guessed, but I don't even know. We often hear that the numbers tend to balloon over time. But are costs front weighted or back weighted with regards to this?

Bean: Yes. Broadly speaking, in theory, they're front weighted because you're going to have to do the R&D to design the plane, set up the production line and buy the plane, and then you shut down the production line and keep flying them for some number of years. And for the F-35, it's about 75-80% sustainment costs. But those sustainment costs are spread across decades. That is a reasonable fraction of the amount of money the United States is going to spend on tactical air power over the 21st century, like the entire century.

But it just gets into how much sustainment are we doing, because the B-52, again, just to use that as an example, is getting new engines and new radar at 75 years old. The actual air frames we have are about ten years younger—they built the last ones in the early 60s. But yeah, there are new—that program is getting new radar, new engines. And the current retirement date is, I believe, 2050 or so, but it's been extended many times. You'll occasionally see jokes about B-52s, like the Star Trek warp nacelles—they're still flying those.

Patrick: This is—again, I feel like I'm greatly uninformed. So there is some mission profile or collection of mission profiles which the B-52 currently serves. We don't have another platform to slot in for that mission profile. We model needing that mission profile in the future. And so we essentially can't cancel our existing plane without replacement.

Or is there some other alchemy happening?

Bean: No, that's pretty much it.

The broad strokes are we currently have three bombers: the B-1, the B-2, and the B-52. The B-52 is oldest. There's a long story there, and the answer is the B-52 is cheaper to run than the other two because it isn't trying to do fancy things with stealth and high speed and everything.

And for a lot of missions, you need a plane that has a big payload and is reasonably cheap to fly. At the same time, performance wise, it is basically an airliner. And one of the serious flaws of our current system is it only builds the best—it can't do "good enough." It is very bad at good enough. It is really good at building the absolute best. And the stuff we build is genuinely amazing. It works well. It's very—amazing performance, good reliability, things work. But that comes at a very high cost. And I can think of a couple programs where they said, "No, let's just do what we have now, but cheaper or 10% better." And they worked well. But that is not how the system works.

SpaceX's early challenges and achievements

Patrick: I think there's some analogous situations happening in space development where—and space development and the military are quite tight at the hip, at least until recently. But for example, SpaceX machines were derided for a long time as not being up to NASA's gold plated standards. [Patrick notes: See previous podcast guest Casey Handmer, in his own spaces, on how much those standards are worth.] And there is a bit of "quantity has a quality all its own" going on there and a bit of, perhaps some of the standards were less than strictly speaking necessary. [Patrick notes: I award some salaryman points for understatement here.]

Bean: Yes. Very much. That is a good way of describing what happened to SpaceX, because a lot of the early stuff was, "Where are we blowing up rockets left and right?" Looking at that—sorry, my background is actually as an aerospace engineer, so I'm reasonably familiar with how all of that happened.

Yes. Originally, all of the standards were developed in an era where we were converting ballistic missiles, and then it sort of got into a high reliability, high cost place. And full credit to Elon Musk—and I'm not necessarily the biggest booster of SpaceX—but full credit, they managed to get us to a different equilibrium that is better. And that has been very impressive.

The Super Hornet development story

And is there room for doing things like the Super Hornet program? The other one would be the Super Hornet, which was basically a new airplane that looks like an older airplane. Because it looks like an old airplane, it was easy to tell the Good Idea Brigade "No, we're not going to do that. We're just going to build a thing that is 10% better than what we have now."

Patrick: So I have less complete context on the Super Hornet than you do, and so don't quite understand the analogy. Can you spin it out in a few more words?

Bean: Yes. Of course.

So what happened was, at the end of the Cold War, budgets were being slashed, and the Navy realized it was going to have a big hole in its tactical air fleet. Programs they had been counting on were going to go away. Stuff was going to get retired, and they were going to have a gap from the mid-2000s onward as stuff got older and just had to be retired. And so they went to McDonnell Douglas at the time and said, "Hey, we like this Hornet—F-18 Hornet—you have here. Can you make us a better version of it?"

And the plane that resulted is, I want to say, 20 or 30% heavier, structural commonality 10%, maybe. But originally, at least, avionics were the same. They redid a lot of stuff because it is basically a new airframe, but they called it the F-18F Super Hornet. And that is basically what is on all of the carriers now. And it came in on time, on budget. They got it into squadron service in about ten years from program start, which in military aviation in the last 30-40 years is just insanely fast.

Military procurement lessons

Patrick: Are there any takeaways that we should learn? Not just a military procurement project, but there was a complex technological artifact produced by the government and contractors in less than a ten year time frame, which arrived on time and on budget. Stop the presses.

What should we be copy-pasting into everything else?

Bean: I think a lot of it was just the conservatism of the project, the fact that they weren't shooting for the moon. They were going, "Okay, we have to build—" I mean, the Super Hornet is better than the Hornet. It's longer range, has a higher payload, etc. But everything it was doing was pretty much well within the state of the art. It wasn't getting stealth—they did a little bit. It was a case of, "Okay, hey, if you can reduce the radar signature at a reasonable cost, do that, but we're not going to blow up the program over stealth. We're not going to introduce a fancy new radar system or sensor fusion, all that stuff—no, we're just going to take the electronics out of the old plane, put it in a new airframe, and then we'll upgrade the electronics later."

I do not know how the electronics upgrades went. I doubt they were quite as smooth. But obviously I just haven't read the budget documents on those.

Patrick: But trying to level set appropriately for myself, this isn't simply incremental innovation. This is not "we have an existing program, but if you can adjust some of these parts for better machinability, etc., we could bring down the cost by 10% and make it 2% faster." This is "no, we're actually going back to a drawing board here."

But we're not inventing new science to go back to the drawing board.

[Patrick notes: In the engineering community “we’re not inventing new science” is sometimes used to say “the scope of this problem is bounded and manageable”, and it is useful to reflect that sometimes the defense industry very literally is inventing new science. The Manhattan Project is the most obvious example, or the field of rocketry, or the field of computers. Fun fact: the military was the primary consumer of computers when that word was a person’s job description.]

Bean: Pretty much. It was what I sometimes refer to as "designation engineering" in the sense of it is fundamentally a new airframe. The structural commonality, I think it depends—I forget the exact number. It depends on how you define it. But I think 10% is not too far off, which is basically a new airplane. But yeah, it was technology off the drawing board. It was a case of "hey, you did this ten, 15 years ago. Go do it again. Let's see if we can't make it—" Well, okay. They did have slightly different requirements. It needs to be bigger. It needed to have a higher payload. Go try it again and make it look like the old one. Because making it look like the old one makes it easy to sell to Congress as an upgrade.

Patrick: That's a fascinating part of the story. There's some amount of learning effect in that if you're doing it only 10 or 15 years later, the people who were on the original project are probably still in the building, and we haven't had enough time to lose organizational memory of something, versus when you're doing new bombers every 50 years, whether you need them or not, maybe every hundred years, whether you need them or not.

It's unlikely that there are many people on the project team who have designed a bomber before. [Patrick notes: Why emerging adversaries invest so much in industrial espionage, chapter one: everyone in the world experienced at doing the thing you want to do can’t work with you, because their day job is trying to kill you. So offer them a side gig instead.]

Bean: Yes. That definitely was probably part of it. Yeah. I hadn't thought about it in those terms, but yes. And I mean, aerospace definitely has higher retention than tech.

Aerospace industry retention rates

Patrick: As a finger to the wind, or if it's something that you can say—if you and 25 people join the company today at the start of your career in aerospace, how many of those people do you expect to be there five years later? How many of those people do you expect to be there at the end of your career?

Bean: Probably for five years, I would say at least 12 to 15 and probably at least five at the end of career.

Patrick: Yeah. I could not give those numbers finger to the wind for people who work in the tech majors. They would say, "Well, we probably have 10%, maybe, regretted attrition rate per year. And so, amortized over a 40 year career, you do not get to 20%, even though there is a bit of a self-selection effect by the out years."

[Patrick notes: I personally do not like this HR-ism but will repeat it because knowing of the model may be useful to you: it is common in some HR places to classify attrition (people separating from the company, either of their own volition (quitting) or not (firing, layoffs)) as regretted or non-regretted. “Regretted” means the company would happily hire this person for their current job knowing what they know now. Non-regretted is… nope, pass, we think the marginal new person we hire in this role is substantially better.

How is this assessed? Some combination of performance monitoring plus asking a very small number of people, often just one’s manager, “So: crying any tears that Bob quit?”]

Lockheed Martin dominance and supply chain

Bean: Yeah. And some of that is probably geographic, frankly, that if you are in fighters and working for Lockheed Martin, there's not somebody else in Dallas doing military aviation, really. At least not that I can think of offhand. Nobody big. I would estimate that Lockheed probably has 75% of the people who work in the private sector. There is some government stuff there, too.

Patrick: Is there a bit of a supply chain for this? Coming from me—I used to live near Japan's manufacturing hub. And in Japan, there's a large automobile manufacturer with T in the name, but then there might be several hundred firms in the associated group. [Patrick notes: Or more. The edges are… tactically amorphous.]

Then there are some companies which are theoretically distinct, but in practice, they've got one client and have had that one client for the last 30 years.

[Patrick notes: This is somewhat analogous to some relationships in e.g. the German mittlestand, where a particular component might be only sourceable from one very specialized manufacturing SMB. In the Japanese case, that component just happens to be a thing that your firm, and perhaps only your firm, needs for the foreseeable future, and so you have a by-global-standards unusually close relationship with a company despite having no explicit cross-ownership nor explicit shared management.

These firms are also, historically, the shock absorber for manufacturing employment in an economy which offers honored-in-the-breach-now lifetime employment: we haven’t laid off anyone. We told our suppliers that we were cutting next quarter’s orders. I suppose they might have laid off someone, but that seems to be a rude question to someone I’ve taken to drinks weekly for the last twenty years, doesn't it. Yeah, working for small companies is rough, isn’t it. That is why you should value working for us.]

Patrick continues: Is there a similar supply chain around, without loss of generality, Lockheed Martin, or is it that everyone has been bought by this point?

Bean: Yes and no. I don't think it's quite to the same degree. I mean, there are definitely many areas where there are specialists, like the engine makers are separate from the airframe makers. No airframe maker makes engines, no engine maker makes airframes. Used to there was—Sikorsky and Pratt and Whitney were under the same corporate parent, but even then, they were kind of separately managed.

So yes, there are a lot of cases where there are suppliers. I'm not aware of anything where there's a case where there's just a company that just has one supplier, and it's been that way for 30 years. That may be a Japan thing. But like landing gear—there are people who specialize in landing gear. You don't know their names. I probably could name 1 or 2 if I had to. And there's probably another one somewhere.

So yeah, there is a supply chain, and you get into—that can get practically deep. I mean, if you want a quick sketch for the F-35, things that are going to be big enough to have notable separate suppliers: the engine, the radar. A lot of the electronics are often done by different companies. And it'll be you'll have a Northrop radar on a Lockheed jet, dropping a Boeing weapon, with a Pratt Whitney engine.

Drone technology and military application

Patrick: So switching gears for a moment, the tech industry has thought a lot recently about drones. They are in the news given the Ukraine/Russia war. Specific topics include the future for doing drone manufacturing in the United States and maybe owning more of the supply chain domestically than previously, because—as I believe it's true, and I assume you're aware if it is true—a lot of the critical components are currently manufactured in a nation which many people assume will be an adversary at some point. [Patrick notes: I pray very sincerely that China and America do not sleepwalk into making that decision simply because of it being the attractive state after aiming two military-industrial complexes at each other for sufficient years.]

So you had an arresting line for us before the interview started around drones not being as novel as people think they are. Do you want to repeat that line for the audience?

Bean: “Drones have been around for a while; we just called them cruise missiles.”

Drones have been around for a long time. We tell them, for instance, more broadly, "drone" is a term that covers just an insanely wide array of things. A cruise missile—technology that we've had since the 50s and in wide use since the late 80s—is a drone by most definitions. So is a FPV quadcopter with a grenade strapped to it. So is something like Global Hawk, which is basically a small airliner-sized plane that doesn't have a person in it and flies over a place for a long time with a camera. These are all drones. If you want to talk about drones, you will need to be more specific.

Patrick: So the smaller form factor, quadcopter ones that have been—my impression is “quite effective” in the conflict currently happening in Europe. That has been sort of forecast for a while. And then we're kind of seeing it now.

This idiot [Patrick notes: referring to myself], writing in 2011 in the comment section of a news story said:

[The existence of bird-sized drones] is not going to be a happy adjustment for people who have to maintain physical security of e.g. a base in Iraq or a community center in a small town in Iowa outside which the president is speaking. At the moment the man-portable drones mostly carry cameras, but the US is proving in spades that cameras are really lethal if you do them correctly. The smart money is not on ‘a camera will be the most lethal thing ever attached to a $100 electronic device the size of a seagull.’

So fast forward 15 years and—all right, I'll give myself some Bayes points for that one. [Patrick notes: This is a meme in the rationalist community and among adjacent people for “I award myself a notional currency for being right about a prediction.”]

Obviously, I was not the only person who understood that was going to happen. The United States has been using the larger form factor ones very actively for a while. [Patrick notes: I wonder whether Bob Woodward, who IIRC once wrote that he had seen the future of war but couldn’t say what it was, was describing armed Predator drone strikes with that phrase. That would have been, oh, 2001ish.]

And the defense establishment has smart people in it. The shape of the future was very obvious by 2011.

Bean: So there's actually some—the stuff I've been hearing, I've not been following—I mean, I'm primarily an air/naval guy. I don't—it's not that I don't know anything about land warfare, but 90-plus percent of my time is spent on either air or naval stuff.

I have heard some stuff that is a little bit more drone skeptical than you might hear in a lot of the news stories, and I think a lot of that's just it makes a good story. "Ooh, look at this cool new tech" is always good press relative to "Nope. Artillery still kills most people," which I understand remains the case.

Patrick: I think a terrible side note on how we perceive reality here is legibility. Drones get disproportionate coverage because literally they provide clips that you can upload to Twitter, etc. And people have seen those clips where they have not seen the analogous bit of targeting and operations that would happen with regards to artillery. [Patrick notes: I regret that, as a heavy Twitter user who desires to never watch people die, that is a thing you can routinely see on Twitter without making anything like a determined search to find it.]

Bean: No, that's probably very true as well. Stuff they can get on the news always gets more than it probably should. The smart bomb usage in Desert Storm was something like 10%, but they got all the press because you have cool camera footage and a B-52 dropping 15 tons of bombs in the middle of Kuwait is not videoing itself.

[Patrick notes: In addition, a Hollywood VFX house can do a Patriot Games style visualization of that for cheaper, beginning years earlier, than it can show a B-52 (unless it simply uses file footage). Media depictions of military reality mediate much understanding of it, including in government. (And I would extend that to saying “... and also in the military”, hopefully uncontroversially.)]

Bean continues: A lot of the stuff that drones are enabling that they couldn't enable before is largely different ways of doing stuff we can already do. There's very little in the way of genuinely new capabilities. And it might be that somebody can access the capability who couldn't before.

We'll take surveillance drones, for example. Traditionally—this is the first use of airplanes in warfare was fly over the enemy and see what's going on. It followed very rapidly by "oh, and take a gun up so you can stop the other guy doing that." And over time, the Army has pretty much always had some capability to put a person up and go look at stuff. They used little observation planes and Pipers in World War Two. For a long time they had helicopters whose job was just have a couple guys who could go look at stuff.

Now it's drones. And the fact that it is drones does mean it's pushed down. Previously, you might have had to be a brigade commander to have that capability—just be able to call and say, "Hey, go do this for me." Now, you can have that down at a company or platoon, which is 30 to 200 people instead of a couple thousand. For those who don't know what military units are.

Patrick: It also presumably pushes it down the ladder of forces of which size and sophistication do we assume will have this capability to use against us. There were only a few peer nations that could put anything in the air that rivaled the U.S. Air Force. [Patrick notes: This is being quite generous to our peers for most of recent history.]

And now you would assume a middle class teenager can have a quadcopter if they want it. Now, I don't know how long that middle class teenager will continue to have that quadcopter if the United States Air Force decides they probably shouldn't have it. But if we look at, say, the Houthis and similar, there is some amount of adversarial use by forces that do not have an Air Force, traditionally understood.

Anti-drone defenses and future warfare

Bean: Yes. And part of it also is we kind of are in a spot where anti-drone defenses are still catching up with the offense there. It's an area of very active research. My money there is probably on the biggest counter being lasers, directed energy. The systems exist. I believe they've been used in operation. I don't know any details on that.

And with a laser, it's electricity. So plausibly protecting the community center in Iowa is going to be a case of we pull up a van—there's a van with a laser in it. You pull it up next to the community center, plug it in. It has a little bit of radar and optical sensors, and it sees something and goes, "Hey, is this a bird or a drone? And if it's a drone, then you blow it up." And there's probably some extra programming where "No, I'm really sure this is a drone. I'm going to blow it up myself."

Patrick: Is the laser for drone interdiction—are you trying to blind its sensors in semi-permanent fashion, or is it "burn it"?

Bean: You are trying to melt it effectively.

Patrick: Terrifying. But nothing about war is not terrifying. Okay.

I think you mentioned before the call that “1967 called and it wants its theories back.” What were those theories?

Naval warfare and historical perspectives

Bean: So—and this is again, more on the—I get especially irritated when people bring up drones on the air/naval side because the cruise missile thing is very true. There is some potentially new things in terms of, okay, we're going to give an infantry company—100, 200 people—a quadcopter drone, and it probably won't be like a proper drone. They'll do other stuff so that you can just look at it and go "go there" and it'll go there, pretty soon. But to give them the capability to have some over the horizon, strike precision over the horizon strike—that is a thing they don't currently have. But we've had anti-ship missiles since World War Two.

Didn't get to be widely noticed until the late 60s, when the Israeli destroyer Eilat was sunk by a couple of Soviet supplied anti-ship missiles. And there was a great deal of panic that the surface ship was dead. The surface ship is still here. And based on recent performance in Yemen, it honestly seems like it's pretty hard to kill.

I was on record predicting that would happen before it happened. I was very pleased when we started to see lots and lots of missiles shot down.

Patrick: Can you give a bit more context for people who don't follow those professionally? We obviously have surface combatants that are outside the Horn of Africa. We have some less than friendly forces in Yemen. They are attempting to kill U.S. combatants with what exactly? And then what happens next?

Bean: Okay, so they're using Iranian supplied missiles and drones of various types. And I know I have been banging on about missiles and drones being the same thing, and on a very high conceptual level, they are. On a lower level, if we use "drone" to mean slower, lower powered stuff, there is actually a reasonable difference in terms of combat performance.

And yeah, so occasionally they will lob missiles at various passing ships. Sometimes these are ships full of iron ore or grain. Sometimes they're American warships. Usually if they hit a ship full of iron or grain, we're dealing with a missile that's designed, broadly speaking, to take out a 5000 tonne frigate. And it's hitting a ship that is 100,000 tons plus and mostly made of dirt or grain or something.

[Patrick notes: Military ships are very big, relative to human scale, but container ships are very big relative to military ships. A modern-ish aircraft carrier might be 100,000 tons, which is unimaginably large by the standards of e.g. premodern navies. A container ship might be twice that.]

Bean: The insurance is annoyed, but it's not going to do too much. Sometimes they lob missiles at American warships, and in that case we shoot them down. Because we've been thinking about dealing with the missile threat for 50, 60, 70 years now and have gotten really good at it.

Patrick: So the thing that has been advanced in some places, and maybe they're indexing on science fiction depictions is that, well, we might be able to engage some small numbers of missiles coming in, but due to the economics of the encounter fundamentally changing, a very large number of less powerful explosives would be, you know, troublesome for us.

And I gather you don't broadly believe that.

Bean: I mean, no, at least not on a ten year timescale. We have deployed lasers on warships. This is a thing that exists. They have been used operationally. Actually, I think the first one was close to ten years ago. Now, rolling out wider still, not fully standard, but—and the thing about a laser is it is a weapon that takes time to work. The damage mechanism for current military lasers is primarily just literally you heat stuff up enough to melt it, which means something coming in slow is a lot more vulnerable than something coming in fast, because laser weapons more or less obey the inverse square law in terms of how much damage they do—scaling with the, it's inversely proportional to the square of the range.

There's a whole bunch of technical details there that we do not need to get into here. But if I'm sending in a bunch of what I will refer to as drones here, which is the Iranian—I forget how you pronounce it, but there's an Iranian drone they've been using. It does 120 miles an hour. A typical missile might be 500 plus. You're going to have 4 or 5 times the ability to engage the drone relative to the missile. Also, the fact that the drone is that slow means you start to get really interesting tactical effects, to the point where with a conventional missile, running away with a ship which will do 35 miles an hour, isn't really practical—it does ten times your speed. It's ten, 15 times your speed. You're just kind of stuck there. If you've got a 120 mile an hour drone and a 35 mile an hour ship, you have the potential to get out of the way. And okay, so half of them just missed because they can't see you, because you got out of the way. And the other half—well, that's what the laser's for. And the laser costs a dollar in diesel per shot, as opposed to half a million to a million.

And there are other things, because a typical drone is a high performance target by the standards of World War One. And to go back to something we discussed earlier, by World War Two, we had gotten pretty good at shooting down much faster targets with guns. We can do even better now because we've got better computers and small radars. And there are systems—there's a system called Phalanx, which is a 20 millimeter Gatling gun hooked up to a couple of radars to shoot down missiles. Proper missiles. We've had it since the 80s. Maybe—there have been some programming changes to make it better at shooting down drones.

Patrick: I suppose the booster side would say, "Well, the equation probably changes when there's 100,000 of them." And that might be more or less realistic. I think less in the near term. And who knows in the long term.

[Patrick notes: As has been widely observed by many, if you’ve seen a carefully orchestrated drone light show originating in China, you’re seeing a subtle demonstration of military force via a dual use technology. “Look at us effortlessly coordinate 12,000 drones and zero high explosives! We can adjust the ratio to taste, of course.”]

Bean: This is the point at which I start demanding numbers from the booster side, like, "Okay, how big are these things? What are they? What sort of engine does this thing have? How far out to sea can you get it?" Because if the answer is "I'm going to buy 100,000 quadcopters"—well, actually, if it's quadcopters in particular, the other thing you have is the radar on which a modern Aegis—Aegis is the name of the system that we use on our destroyers and cruisers—is megawatts of power radiated. I don't know the number. But it has a—let's put it this way. It has a substantial impact on the ship's range when the radar is running. That is how much energy it's using.

One potential counter to "I'm trying to kill you with quadcopters" is "I'm going to just look at them really hard, and they will burst into flames and fall out of the sky."

The other counter is, "Hold on a minute. Why in the world am I within 20 miles of your coast? I'm just not going to go there." Because a typical quadcopter is battery operated, has a range of 10 or 20 miles. I'm just not going to do that. And once you start to talk about systems that are capable of engaging a ship operating in a reasonable manner, 100,000 just does not look like a plausible number.

The missile is likely to be cheaper than the thing that is trying to attack because it can be smaller and shorter range. And so, yeah, at some point the thing you do is you design a system that takes a little bitty battery powered drone that you can build for—I mean, it's mil-spec, so $1,000. And there are reasons you do want mil-spec we can get into that if you want. It's $1,000 to shoot this thing down and I can pack dozens—I can take a shipping container and pack a couple thousand of those in a shipping container size. I'm trying to—I don't want to get too deep into how you deliver it to a ship. That might be a little bit tricky, but it's a solvable problem.

Patrick: Sorry I'm blanking in understanding in which direction that is solvable.

Bean: I think it is solvable to protect against drones broadly. A lot of the use cases just sort of physically don't make sense—100,000 quadcopters. I'm just not going to go to the place the quadcopters are. I don't need to go within 20 miles of the shore. I'll just be over there shooting cruise missiles that I can shoot from 500 miles away.

Patrick: Yeah. I think one that comes to mind for me—and again this is far, far away from my field—is the Strait of Taiwan is a small bit of water relative to an ocean. And things there might get pretty hairy if either side decides to make it very contested.

Bean: And there are generally, as best I understand current U.S. doctrine [Patrick notes: doctrine: “plans we make in advance”, military style], we would not expect to be operating ships in the Straits of Taiwan themselves. That would just be too close to China. And I mean, again, I am largely bullish on the survivability of aircraft carriers and surface warships, but there are some things that are just too dangerous and that would be one of them. If there's actual shooting, obviously freedom of navigation operations are a completely different thing.

And as you start to get into things where it's like, "Okay, I can game out a scenario where this is a plausible engagement." And the answer isn't just, "No, you're being silly. Please stop stacking the deck in your favor." We are still in the early days of anti-drone systems. They're going to get better. And I can think of a lot of different counter drone systems that move the economics much more in our favor.

Patrick: And it's worth saying that over in military history, there's been iterated games of this. You mentioned earlier that you disagreed with the contention that generals are frequently fighting the wrong war, and that but there was some forward thinking.

But I do remember that at least for a while, the strategic thinking of ships was: just put bigger guns and more armor. That is definitely what we need, and all we need.

And there was the age of the dreadnought, and then air power brought the age of the dreadnought mostly to the end, with a little bit of overhang into World War Two.

Bean: Oh, do you want to go there?

Patrick: I know we've been talking for a bit, but—

Bean: That is one of my very most core areas of interest. I have a very nuanced view on that, which this podcast is undoubtedly too small to contain.

Patrick: How about we give people the three minute version of your nuanced view, and then we can link to your other writing for people who enjoy geeking out about things that civilians don't understand about World War Two naval history.

Bean: Yes. So there were always alternatives to the big gun warship. It started off with the torpedo being developed in the 1870s where I think they would have called it locomotive torpedo in the 1870s, 1880s. Originally "torpedo" just meant any underwater explosive, for instance. "Damn the torpedoes. Full speed ahead" at Mobile Bay in the Civil War. That was referring to what we would now call mines.

Patrick interjects: Darn I wanted to say that and prove I know something about this subject. Continue.

Bean continues: Even before that it was ramming. The theory was the way to sink a ship is to build a thing that's armored and has a spike on the front, and we'll run it into other ships. Didn't really work. Torpedo comes around and changes happen. They developed this thing called the torpedo boat destroyer, which eventually evolved into the thing we now call the destroyer through a lot of different stuff.

Submarines came, airplanes came, and—The thing is, there were always areas where the battleship made sense. And the role the battleship filled through World War Two is if the enemy manages—if things go wrong and the enemy manages to get ships through the carriers, the carriers are looking the wrong way, doing something else. The weather's bad, whatever, and gets in range. You need something to stop them. And that is kind of the role the battleship played through World War Two.

After World War Two, the U.S. and its allies have close to a monopoly on sea power. The Soviets mostly are doing submarines and don't—the Soviets don't have any battleships, or they do, but they're extremely old and not really a threat. And so that sort of starts to kill it off. And eventually the thing that actually kills it off is nuclear surface-to-air missiles, because those also work at night and in bad weather. And you have those anyway, because it's the 50s and everything must be nuclear. And they also work on surface targets.

Patrick: And so the potential for a nuclear engagement makes it less and less likely that people want to have large surface fleets for end of the world adjacent scenarios.

Bean: No, it's more just that if I'm going to have a missile cruiser that has nuclear surface to air missiles, that can also do the "oh, shoot an enemy battleship just came over the horizon" mission.

Patrick: Okay. All righty. Well, I will drop some links to your previous work. And people who want to follow along in more detail are welcome to read it. I'm sure there are many other sub areas of naval history and U.S. doctrine that you could educate me on for a very long time. Sadly, there are a finite number of minutes in the day. Can you plug for station identification—where can people find your writing on these things?

Bean: I write at my blog, Naval Gazing. I don't do Twitter or anything.

Patrick: You're probably better off. [Patrick notes: Twitter is one of the most important products on the Internet, but I would specifically anti-recommend it if one works in a field where people commonly have security clearances and one does not need to e.g. manage a PR- or electorally-driven process.]

Bean: I definitely tend to think I am. Stuff I'm writing about currently is about different types of sort of nonstandard hull forms and how they get used. But I've covered everything from fairly in-depth history of the battleship. I got my start as a tour guide on the USS Iowa in Los Angeles. Museum ship. If you're in that area, you should go. I mean, if you're near a museum ship, you should go. Museum ships are great.

And then I've sort of drifted into defense policy. Sometimes I take apart battles. Sometimes I talk about airplanes and stuff. I've been doing it for a long time, so there's a bunch of stuff there.

Patrick: Well, thanks very much for coming on Complex Systems today. And for the rest of you folks, thanks very much for listening. And we'll see you next week.

Bean: Thank you for having me.

[Patrick notes: Bonus links from Bean! I’ve included some of these at appropriate points in the transcript, but given that he rounded them up, thought I would include them as a list.

Operations Research and Anti-Submarine Warfare:

https://www.navalgazing.net/OR-in-the-Atlantic

People and Perspectives:

https://www.navalgazing.net/Jackie-Fisher

Aircraft and Systems:

Drones and Modern Warfare:

Naval Systems:

Naval History:

Defense Economics and Analysis:

]

podcast