Calculating the Cost of Golden Dome

Jeffrey Lewis 0:08

You are listening to the Arms Control Wonk podcast, a podcast on arms control, disarmament and nonproliferation. I'm Dr. Jeffrey Lewis, a senior research scholar and William J. Perry lecturer at Stanford University and a distinguished research fellow at the Foreign Policy Research Institute.

Aaron Stein 0:24

And I'm Aaron Stein, president of the Foreign Policy Research Institute. And Jeffrey, we have a guest today.

Sam Lair 0:30

Hello, Sam. Hi. I'm Sam Lair. I'm a research analyst and fellow at the Foreign Policy Research Institute.

Jeffrey 0:40

And a graduate of the Middlebury Institute of International Studies.

Aaron 0:44

Along with me. Hey. Two alumni, where Jeffrey used to teach before you moved north, good friend, north. Ah, loving the Palo Alto life. Well, you were even further west. You were in Hawaii. So what cocktail of choice did you find out there?

Jeffrey 1:05

Just a lot of tiki drinks. And there are some really spectacular craft cocktail places. In Honolulu. I had no idea. I had a milk clarified Mai Tai that was just extraordinary. I dream about that thing.

Aaron 1:29

For anybody in the alcohol industry, we are not above being bought off here. If you want to get in touch with us for alcohol sponsorships, we're open to it, right?

Sam 1:39

FPRI is very open to bribes in the form of sophisticated cocktails.

Jeffrey 1:42

We used to shill for some bitters.

Aaron 1:44

I do. I found a pallet or case down in my basement. They had made the move. Oh, I used mine. They were great. I'm sure you did. Well, I'm trying to come up with a way of pallets, containers. Perhaps my bitters were in a life jacket?

Sam 2:00

Well, you know what else is in Hawaii is the Pacific Missile Tracking Range, or words to that effect.

Aaron 2:07

Yeah. So how are we going to land the plane here on the work that we've been doing at FPRI about Golden Dome?

Jeffrey 2:15

It sounds like the plane just crashed there. Which maybe is a metaphor for the future of Golden Dome. That's right.

Aaron 2:24

A while ago, we actually did do a podcast about this, Jeffrey. Sam reminded me about this, where we sort of did a loose calculation about Golden Dome. But we really worked through it over the past year. With some generous support from the New Land Foundation. So we thank the New Land Foundation for that. Who wants to take the lead in summarizing what we've been doing?

Perhaps, Sam, you want to go first?

Sam 2:49

Sure, I can talk about this. So broadly, the way that we conceived of this project was when news about Golden Dome started coming out and people were talking about space-based interceptors again, we thought about how previously people had gone about doing the work of trying to understand what SBI constellations would look like. And the closest equivalent to the kind of ideas that were being thrown around is still brilliant pebbles, right?

The idea that you would have an interceptor with a sensor in a life jacket up in space Then when the interceptor itself detects an incoming launch, it kind of ejects from the life jacket and goes off to chase down the missile. And there are a bunch of different ways that people had put together to calculate how large a constellation of brilliant pebbles equivalents would need to be in order to be able to handle certain threats.

And one that we found most salient, though perhaps not always the most accurate, was a model put together by Edward Teller and his colleague at Livermore, Gregory Canavan.

Jeffrey 4:00

You know what I really liked about that, Aaron, that Sam and I kind of really enjoyed was The whole idea with the space-based interceptors is because the satellites are in motion, right, you have to sort of cover everywhere between certain latitude points if you want to cover anywhere. And In the 80s, without the ability to really model the movement of big constellations, which we could like totally do an STK, this was like a really beautiful, analytic, just pure numerical simulation.

I don't know, Sam, it gets within like what, like 10% of the real number?

Sam 4:41

Yeah, something like that. And again, I think the math is a little bit fuzzy at the edge. And there's a particular variable because the argument that Teller and Canavan make is, you know, depending on what you're sizing the constellation against, you don't actually need to cover the entire world, right? You're not trying to prevent every missile launch everywhere.

Maybe you're only trying to size it to, in their case, they were really upset about the SS-18, which fair. And so they basically said, we're only going to defend up to the highest latitude that SS-18 silos are at. And because of that, that means that you actually get some benefits in terms of constellation sizing. And that's where the fuzziness comes into play.

And they said that was about 10%. I'm not sure how true that is, but we ended up finding it to be a very useful tool for making estimates.

Aaron 5:29

When was the first paper published?

Sam 5:31

I can't remember, Jeffrey.

Aaron 5:34

Wait, the first Canavan and Teller paper? Yeah, it was 88 or 89. Do I have that right?

Jeffrey 5:39

Yeah, it's like the mid-80s. They get in a big fight with Dick Garwin, who roughs them up pretty good.

Aaron 5:45

Yeah, so I'm trying to get the sequential dates. Canavan and Teller are first, and then Dick... So it's 1990. Right, and then Dick Garwin's also in 1990. But they're modeling things... You know, basically at the turn of 1989 to 1990.

Jeffrey 6:01

So a different era, although... Yeah, and they're really, I mean, I, you know, don't want to speak ill of the departed Dr. Teller, but... Well, we'll get to that.

Sam 6:12

Screw that guy.

Jeffrey 6:13

Yeah, yeah. They, you know, I'll tell you what, the numerical model is fine. Some of the inputs, I have questions.

Aaron 6:20

So that's actually really important. So who wants to take a stab at explaining math over the podcast? Without the use of it simulated in the paper.

Jeffrey 6:33

Okay. I think I can do this. Right. So the number you want to get out is how many satellites do you need? Okay. And that's a really simple concept. Your satellite has to be able to fire an interceptor. So you need to know like how much the inner, how quickly the interceptor accelerates and how fast it goes.

And the window that you have to shoot is the length of time that the missile is burning. And it gets a little fuzzier than that. And it turns out you don't get to shoot the whole time the missile's burning. First, you have to wait to see it. So you lose a little time on the front end. Then you have to sense it and you have to be like, oh, yeah, that's a missile.

I think we should shoot that. And then you have to make a decision and you have to get that decision to the satellite. And then hopefully it's still in position and it fires. And then you have to have kind of a back end where... When the missile burns out, if you have multiple warheads, you have the missile sitting on a bus, you can shoot at the bus But once it starts dispensing warheads, anything it dispenses, you can't shoot at.

So you kind of, you basically have a short window and all you're doing is calculating how many satellites do I need to make sure I have at least one or, you know, actually for our calculation purposes, because sometimes you miss two shots at each one. And then, you know, you're multiplying it by like, How many missiles are there in how big of an area?

So if there's like 100 missiles all packed together in a tiny area, I need a really dense constellation. If they're spread out all over China, I don't need as dense a constellation.

Sam 8:35

So I think to sum up the math a little bit, right, by kind of establishing northern and southernmost latitudes that you want to defend, you have a certain area of the Earth, and then you have an area that an interceptor of a given speed can reach over a given time. And how much time they have is defined by that interceptor window that Jeffrey was talking about, right?

Which is that all of missile burn time, and you can also include bus deployment time, Minus detection time and then decision time around release, right? So how long to detect it, how long to make up your mind that you're going to shoot at it. It's burn time minus that amount of time. And then from, yeah.

Jeffrey 9:15

And we'll probably, we'll probably spend a lot of time going over this, but like canavan and teller pick the easiest scenario in the world. They pick the longest burning liquid propellant missile with the most warheads, which dispenses them really, really slowly, which are based on silos and the Southern part of the Soviet union. So they really do manage.

It's not that they're lying. It's just that the scenario they pick is very favorable to this calculation.

Sam 9:49

I think we should say, listen, it takes some massaging to get to the NHTSA criteria.

Aaron 9:54

I think the way we would say this is that Teller and Canavan came into this supporting space-based missile defense. I'd say that we would not be slandering them and saying, libeling slandering them and saying so. No.

Jeffrey 10:09

No, no, no.

Aaron 10:10

We're on firm legal ground.

Jeffrey 10:12

By the way, this reminds me of one of my all-time favorite moments as a young RA in DC. I don't know if anybody remembers Baker Spring at the Heritage Foundation, but he had just written a report for the Heritage Foundation on missile defense. And it was obviously everything under the sun. And I think maybe it was Larry Wurzel.

Oh, these are people just aren't around anymore. Larry Wurzel looks at him at this meeting at CSIS and says, you know, oh, Baker, do you have any money in your plan for, and then name some obscure missile defense. And Baker gets this just grin on his face like, yes.

Sam 10:46

Like, yes, of course, there's money for everything. I think something that's salient here as well is that Canavan and Teller are both housed at Livermore, which is also the institutional home of the Brilliant Pebble concept. Actually, they still have a pebble there, both a sounding rocket used to test some of the propulsion technology, but then the last mock-up model of the Brilliant Pebble is there.

I got to see it last October when I went there for a workshop. It was amazing. It's in the paper.

Aaron 11:19

It's in the paper. Spoiler, the paper is not quite formatted. It's finished, but it'll be out in September. The reason why I say this is because not to slander our fellow colleagues, particularly in this field. But the density, or I guess the number of satellites, gets you a rough estimate of cost. And you can throttle that cost up and down if you make certain assumptions.

And those assumptions are on that window you have to shoot at things. And that window is based upon the burn time, or how long the missile takes to get up into space. And so what we found in this is how susceptible that model is to major swings in terms of overall cost by how fast the missile that's being shot can and should travel.

And Dick Garwin had this, I think, in 1990, which is why I was trying to get my My year is correct because Teller and Canavan, I think it was 600 seconds on the SS-18 for a burn time, which is too long, right?

Jeffrey 12:34

Yeah. I mean, yes. It's both too long for the actual SS-18. I mean... It includes all of the deployment time. So basically they're shooting at a bus. They're giving themselves up to and including the moment when the bus has one of the like 10 or whatever it was warheads left on it, which is like, it's not the way we do this.

Sam 12:58

Like the bus has deployed nine warheads. Canavan and Teller's Constellation gets the last one and they kind of... You know, wash their hands of it and say, well, that's damage limited.

Jeffrey 13:07

Yeah, well, they refer to it as leakage. I mean, there was some leakage, you know, like, you know, Taylor and his depends.

Aaron 13:14

The point is not to sort of bash on their paper. The point is to draw out a conversation about how the model shifts based upon our threat scenarios, based upon the burn time of the missiles. So we chose three missiles, right, to sort of represent How you would conceivably size your constellation. And the three missiles were, who wants to fill it in, Sam?

Sam 13:43

Yeah, so Jeffrey put together a model of the actual SS-18, or rather the successor to it. So Jeffrey ended up putting together a sense of what the Sarmat burn time looks like, since Sarmat is based on the SS-18, which itself is based on the SS-9. You know, we have some sense of that. And then also, you know, handily, the SS-18 was also the Juniper Space Launch Vehicle.

I'm sure I butchered that in arms control wonk tradition.

Aaron 14:15

You got the pronunciation right. We have to boot you off the podcast.

Sam 14:21

Exactly. But so thankfully there's a bunch of data for the Space Launch Vehicle and Jeffrey was able to use that to kind of reverse engineer what that launch window looks like. And I believe that was about 245 seconds plus another 100 seconds Of bus deployment. So that was one missile we considered. The second missile was a 180-second burning solid propellant ICBM, and that was derived from the Topol, which was also a space launch vehicle.

So the Russians gave us a bunch of data on that, and that's about 180 seconds. And so we assumed that the Chinese are probably putting a missile with similar burn time like that into the silos across China. And so we sized a constellation against the Chinese silo fields. And then something that Garwin and a bunch of other people in the late 80s and early 90s suggested was that you can make a much faster burning solid propellant missile.

Garwin said something on the order of 100 seconds is certainly feasible. One of the early studies into boost phase defenses that was commissioned after the STI speech in 83, the Fletcher panel, they concluded that actually you could, a report made to the Fletcher panel Trying to be specific here. But that report suggested that you could make a 60-second burning solid propellant ICBM with only about a 20% reduction in payload.

So we didn't quite go that far, but we assessed a 100-second burning ICBM as well.

Jeffrey 16:04

Yeah, I mean, I was just going to say, I mean, there are just some really neat things that come out of this analysis, but I'm sure you're going to ask about them.

Aaron 16:12

Well, no, it's the Goldilocks, right? You have the longest one, right? You have the medium one, and then you have the fastest burning one to get a range of costs. And we pegged the burn time the best we could to real data, right? I think one of the things that's changed since Perhaps Teller and Canavan would have access to classified material.

Perhaps Garwin would as well. I can't remember his exact position at the time. But we just know a lot more because the Russians shared a lot in that glorious time of the 1990s.

Jeffrey 16:47

The fact that they used a Topol as a space launcher and then gave us all the data is just incredible.

Aaron 16:51

Yeah, and then they reused these things for space launchers. So we have, we think, very accurate information about how long they burned. And then we made a series of other assumptions that, and I don't know where we got this, I forgot, because the paper is not coming directly to my brain right now, that you also have to think about replacing one fifth of the constellation every single year.

And I think that's based upon our experience with Starlink, but also just some reported conversations about this.

Jeffrey 17:19

Yeah. I mean, Aaron, that's the thing I really wanted to bring up, which is one thing that's kind of neat is In the 80s and 90s, the killer argument against these constellations was that they were so big that you could, you just did not have the space launch to put them up. They would be ruinously expensive if you had it, and, and in fact, we didn't have anything like that level of space launch. And so I think, I think people were right to say, hey, now that Starlink exists, that's a brilliant pebble-sized constellation.

So at least in theory, we can make smaller satellites And launch costs have come down. And so you're saving on both ends. And so we really can do small satellite constellations. And so that's what triggered us to want to go back and ask, could you do Starlink but for Brilliant Pebbles? And I think the answer is still no.

But it's no for a different reason than it was in the 80s and 90s. And that, to me, is pretty interesting.

Sam 18:29

Yeah. I think another thing that's worth mentioning as well is that launch cost per kilo basically has come down really, really significantly. And that's kind of what enables things like Starlink. But the number of space launches is actually still a bottleneck, right? And there's kind of a long wait time for a ton of these national security relevant satellites.

And one of the difficulties is that since there's no serious competition with SpaceX, that means that you're still reliant mostly on SpaceX, right? The US government doesn't have a serious launch program anymore. And so even though launch costs are much cheaper, it would take a really long time to get an SBI Constellation in orbit. And that would probably come at the cost of other space national security programs.

Aaron 19:12

And then you're replacing the entire constellation, one-fifth of it, I believe, every single year. And so your costs accrue every year in addition to just sort of general maintenance. You're talking about building a one-fifth of the overall size of the constellation and then launching it every single year. It's a huge endeavor.

Jeffrey 19:32

Yeah, I mean, part of the things that I... This is a really great business model if you're SpaceX.

Aaron 19:43

Yeah, it's huge. But maybe we should talk about sort of the variation in sizes because if you make very generous assumptions, this is why I started with the Canavan and Teller, you can sort of size a constellation if you squint at it, like makes economic sense. It's like, oh, that's not so bad. But that's such generous assumptions.

But if you make a sort of more aggressive assumption that an adversary is not going to sit there and say, oh, wait a minute, I'm going to let the Americans destroy my deterrent by building space-based missile defenses. They can shorten the burn time without trading a lot of payload in ways that could make the constellation, forget about leakage, very expensive.

And I think there's also one of the things that we tried to capture in addition to the cost of space launch, which was the required number of interceptors. Based upon how fast we think they can fly, right? Fly is not the right word, I get it, you're in space, but like, you know, how fast they can burn out of those life jackets to intercept the missile that they're shooting at.

And we also made another assumption, I think we should say, is that we tried to minimize leakage. And so in a way that Canavan and Teller were kind of cool with maybe nine to 10 getting through, perhaps, you know, to quote perhaps that heritage paper you were talking about, Jeffrey, this can be dealt with with other missile defenses behind them.

We decided that you probably don't want leakage because those warheads will continue on to their targets. Even if they fall a little short, they still impact somewhere in the continental United States or North America. So maybe our Canadian friends.

Jeffrey 21:26

Yeah, that's right. We ran the numbers. So there are two different issues. One is you want to run the numbers. What does it look like? Let's say I hit the bus. The warheads are very likely to survive and continue on their way. I was actually talking to somebody who's an enthusiast for this, and he just was like absolutely insistent that it would destroy the warheads.

And I found that absurd. So that's one number is at what point have they are they going fast enough that they're going to hit somewhere in the U.S.? But then the second number we used, and I could not believe the Defense Science Board declassified this, but the Defense Science Board declassified a chart. They didn't say which systems, but it was a series of US systems and adversary systems that had the timeline for warhead deployments.

Aaron 22:16

Yeah, that was a great, you pulled that. That was a you, that was a you pull. Yeah, yeah. That was a you pull, yeah.

Jeffrey 22:21

And, you know, their position was, as a practical matter, After burnout, you needed to intercept in, what was it, Sam, 100 seconds? 100 seconds. That's right.

Sam 22:31

I think 100 seconds was the kind of longest burn time, or the canonical cutoff, I think. So we just used that. And we only used that for the Sarmat estimate. We assumed that the Chinese are putting single warhead missiles in the silos and that the fast burn North Korean missile would also be only using one warhead.

Jeffrey 22:58

Yeah, we gave North Korea a fast burn missile just because we wanted some other scenario. But fast burn missiles are really wild. The idea that you would... Try to burn out. I was reading that in some cases they burn out so fast that they actually have to coast before you deploy the warheads, because they burn out in the freaking atmosphere. Yeah, it's, it's crazy. Like you, you don't have bus time, but you do need the time to, you know, get out in the atmosphere and then eject the shroud before you can dump the RV. And we had some numbers, I think, from Ankit Panda's reporting, right back when Ankit was a, was a reporter, before he transitioned to being a think tanker.

Sam 23:43

Yeah, I think Ankit got a unnamed US official, man that guy really gets around, to comment that the Hwasong-15 burned for about 290 seconds. And the North Koreans have gotten better since then, so I'm not sure how useful that number is anymore. So we ended up with a whole range of how big these constellations needed to be for these various targets.

And a bunch of it depends on how fast the interceptor moves. And so we ended up considering interceptor speeds or burnout velocities of four, eight, and six kilometers. And so that produced the variance. There's also some fuzziness with decision time. We assessed a couple of different levels of decision time,

Jeffrey 24:34

I thought we were extraordinarily generous with decision time.

Aaron 24:38

This was probably the biggest fight Sam and I had, which is... Yeah, I was about to say, if there was any tension on the paper, it was about decision time, because...

Jeffrey 24:48

I just, you know, yeah, the magic, we don't have to... Yeah, okay.

Sam 24:54

Well, and I ended up coming around to your position because I think you're right. It... You can't just fire immediately when it takes off. The APS report, American Physical Society report on this from 2004 does a really good job of explaining that the infrared sensors that we have in space, they probably aren't going to be able to detect the signature of the missile until it reaches about seven kilometers in altitude, and that's going to take some time.

Less time if it's a solid propellant missile, more time if it's a liquid propellant missile. So you're losing some of the burn time there, just waiting for the infrared sensor to detect it. And then after that, you have some decision time. And so I guess we ended up with 45 seconds of decision time, 30 seconds, and 15, right?

So assessing those three variables. And 15 is really pushing it.

Jeffrey 25:45

I think 30 is pushing it to be.

Sam 25:48

Yeah. But so some of these are, you know, I guess when we're talking about like the assumptions that we make, you know, we're we're assuming for the more aggressive numbers that some breakthrough in detection occurs and there is extensive automation. Which kind of introduces room for adversaries to mess with this, right? If it's a highly automated system, then I would like it to be shooting at my sounding rockets instead of my ICBMs.

Aaron 26:15

I was just thinking about that's one way for SpaceX to get a bad business model is that it starts losing some of its rockets to attrition.

Jeffrey 26:25

This is, this is a serious conversation that we should, we should have in a second about like what the countermeasures start to look like, because all of this is not, we're not assuming any countermeasures. I mean, other than the fast burn booster, this is just straight up like the Chinese have built a silo. What's it going to take to cover that silo? you know, and so it, it suffers from the classic, we play chess and they don't move their pieces fallacy, which you know, like.

You got to start somewhere. But, you know, I don't know what it is worth saying, by the way, about the sensors, because sometimes people are like, well, you know, like the Chinese desert, it's often very clear. The problem is those sensors are not designed to look at the ground because they the ground has clutter and noise. So even if it's clear, like the whole like waiting for it to break above the clouds, that's not Because there are clouds, I can't see through them.

I mean, sometimes that's the case. But it's also like this system is optimized to look at a certain point. And sure, you could build a different sensor system, but then we would have to cost that out. Or like Sam Wilson from aerospace. Yeah.

Aaron 27:42

Or I was going to say cloud. Your favorite, Jeffrey.

Jeffrey 27:46

I do love some quads. Quad's a powerful, powerful, powerful ally.

Sam 27:53

I think returning this to how large the constellations need to be, I think making what we think are reasonable assumptions, so a four kilometer per second burnout velocity interceptor, 45 seconds of decision time against these three things, if you're Building a constellation to delete Sarmat, you're going to need about 3,000 interceptors, and that's just having one. So you're assuming that all of your interceptors work.

If you wanted to double up, then you would need to double that. That'd be about 6,000 interceptors. If you wanted to do the Chinese silo fields, you would need 33,000 interceptors, And then to deal with a North Korean fast burn ICBM, you'd need about 9,000 interceptors. One of the interesting things about this as well is that the more dense your silo field or missile deployment area gets, the more difficult it is for the constellation to intercept.

Maybe another reason we should think about dense pack again is that it'd be very difficult to resolve with a space-based interceptor constellation. But you can see that these numbers get really, really wacky.

Aaron 29:08

Based on basic assumptions. I mean, we're assuming... Again, that's why I'm trying to lead the conversation is that the number can vacillate wildly depending on... The assumptions you make, but even at the lower end, we're talking about huge constellations here. You're talking about Starlink-sized constellations in its current configuration at the lower end, making very generous assumptions to the defender, pegged to the threat as it is today, right?

With no countermeasures, nothing. That's right. And then... Go ahead. Oh, and then you have to replace one fifth of them every single year. And the difference between it and Starlink is that it's a lot more expensive per satellite than it is for Starlink because you're actually carrying your interceptors and your sensors. Yeah.

Jeffrey 29:56

And I think it's the kill vehicle. I think it's the kill vehicle. And that for us was like the hardest part because we were looking at Enfire. You know, we're looking at like historical space-based interceptor programs where we actually flew hardware and they were like, Grotesquely expensive. And we were kind of like, well, that can't be representative.

And the answer is we just don't make a lot of kill vehicles. They're real expensive. And maybe there's some alternative universe where we're churning out all kinds of kill vehicles at a low cost, but that almost certainly means reduced performance.

Sam 30:41

Yeah, so I think a good way to think about this is the proposals for Brilliant Pebbles, they assumed that the interceptor cost, this includes kill vehicle cost, it would be from about $1.3 to $3.5 million Then when we actually integrated this into a spacecraft, so Clementine for example, which to be fair did go to the Moon and so maybe there was some fancy stuff on there, but that cost almost $100 million.

Then Enfire, it cost almost $120 million. Then if you think about just the cost of say the EKV on the ground-based interceptor for the GMD program, that's like 40 million dollars each for those kill vehicles, and that's using the cheaper sensors, right? This is a really interesting story actually that I, I found recently was there was a question about what type of infrared sensor you would use on the, on the exo-atmospheric kill vehicle for those.

And it was a competition between two different firms. One was using a much cooler silicon-based sensor, and then another was using a mercury cadmium telluride sensor that was a little bit cheaper and didn't get quite as cold. So its discrimination abilities weren't as sophisticated. And then the company that was suggesting the silicon sensor got caught doing some bad behavior during the bidding process and got excluded.

So the US government ended up defaulting into the poor performance yet cheaper sensor, which is incredible.

Jeffrey 32:15

Yeah. And so you end up robbing Peter to pay Paul, right? Because you save on the kill vehicle, but then you need... More interceptors with kill vehicles, and it's like I just think the math is really unforgiving. If, if I just wrote an article and I, I think I called it the, the, the mundane technical disappointments of, of any defense program. In this case, they're very unforgiving for constellation size and.

Aaron 32:48

Yeah, I mean, so what did we actually come up with the ultimate cost benefits at the high end versus the low end and everything in between?

Sam 32:59

So I think against Sarmat, you're looking at upfront costs of at least like... Tens of billions of dollars, maybe $20 billion, and then you're spending several billion each year to replace that.

Aaron 33:16

Just to cut you off again, we did not factor in R&D either. We did not factor in R&D either.

Sam 33:21

No, we didn't.

Aaron 33:21

No, these just magically work.

Sam 33:24

And we assumed also that these were really cheap interceptors. We were assuming that the interceptor just cost a million dollars, which is not going to be the case. Against the Chinese silo fields, it's probably in the hundreds of billions of dollars upfront, and then maybe $20, $30 billion each year to sustain it. And then against the, the fast burn North Korean one, it's going to be 20 to $40 billion upfront.

And then, you know five to 10 billion each year to sustain it.

Aaron 33:55

And again, this is assuming that the interceptors are very cheap, cheap interceptor, again, not R and D. Okay. We're just not talking about R and D here because it's, it's hard to capture that. Right. And the reason why I keep bringing up R&D is not just the upfront costs that we're spending now, right? But you assume that you need to make upgrades to your system as adversaries make upgrades as well.

And so you're funding new research and development as you go through on improving the kill vehicle. Tests, we also didn't factor in a couple of year tests. We're just talking about the cost of the size of the constellations.

Sam 34:33

Yeah, and this doesn't even include things like command and control infrastructure, the software required to do that, right? All of these kind of other costs that end up creeping in and pushing programs upward.

Jeffrey 34:44

Right, because you're going to end up, I think the tendency, it's going to be really hard not to gold plate this. Because because otherwise, I mean, have you seen the Oval Office there, Jeffrey? Yeah. You know, well, because you're stuck like you can either you can either go with basically a stunt, which they could do. You know, they could put up 20 interceptors and, you know, declare that they have a you know, we did that in Alaska, right?

Yeah. They can do that. But if you start to, I mean, if you really start to take that, the dome part of Golden Dome, seriously, the gold, no problem. The dome part, that's where it gets hairy because, you know, if you're like, well, you know, there was one of the, one of the MDA things is they want endo-atmospheric space-based interceptors.

They want to be able to reach down into the atmosphere, which, you know, if you're dealing with a fast burn booster, it gives you a little extra time. But like, Boy, that's going to add some mass, I think. That's going to require a different sensor architecture, I would think. There are all kinds of questions about concepts of operations, because presumably you don't want to be like whacking space launches on the regular.

And like all that stuff, I think, could really start to add up very quickly.

Sam 36:10

I mean, you talk about whacking space launches, Jeffrey, but that's one way to maintain space superiority. No one else.

Aaron 36:14

Much cheaper just to hit the launcher with a Tomahawk, man, than it is to do this. All right. So the point is that it costs a lot. We should... And then we really did peg the time, even though we created a fast burn solid fuel for the NORCs, around real burn times based upon data. The most obvious way to overcome this thing is to compress data.

The decision time. So the missile flies faster. That window, that 30 – what did we say, 30 seconds? I can't – 45, 30, and 15.

Sam 36:52

45 is the most reasonable.

Aaron 36:54

The most reasonable is 45 seconds, right? That means 45 seconds from seeing this thing, a decision is made to shoot at it.

Jeffrey 37:04

No, it's 45 seconds from launching because you don't even see it for maybe 30 seconds. Right.

Aaron 37:11

And so you're actually compressing the decision time basically down into 10 seconds. And if we were to sit here in silence for 10 seconds, it would be awkward because people would think that the podcast had stopped. But 10 seconds is not a lot of time.

Jeffrey 37:25

Yeah, I think that's right. And Now that then starts to bring us to all the things the Russians and the Chinese might do to- Because they wouldn't just do fast burn boosters.

Sam 37:41

I mean, it'd be great if they did, right?

Aaron 37:43

That's because you want to work out thrusting. We know what your ulterior motive is here, Geoffrey. But the point is, is that you, and then if you compress the decision time, right? So let's say you take from 45 seconds, right? And let's say you compress it. We did mess around with that in the model. It sends the number of satellites just through the roof, right?

And so that cost maybe goes from like say 20 to 40 billion to like – I don't want to throw out a number. You basically need to double or triple the size of the constellation.

Jeffrey 38:21

Yeah. And so – And again, that's before the Russians and the Chinese do anything heroic.

Sam 38:29

So I guess since Aaron has kind of teased this enough, right? The Russians and the Chinese are not going to stop it at passive defensive measures of just kind of – Shortening burn time, but what's everyone's favorite SBI countermeasure? Oh, I know.

Jeffrey 38:45

Okay. Well, I'm going to let everyone else go because there are so many, but I think Sam and I have the same one.

Aaron 38:51

I think we all have the same one, so you just say it, Jeffrey. I think we all have the same one because we've all talked about this individually a million times.

Jeffrey 38:59

I would have a Sprint or Spartan-like pop-up booster. That just burns out in very, very few seconds. And I would have a really large yield warhead on it. And the minute it got within line of sight and our sensors saw it, I would pop that and I would vaporize, well, blind vaporize it, you know, depends on how hardened the satellites are, which again, like I was talking to an enthusiast, he was like, well, we can harden satellites for the radiation environment.

It's like, yeah, but now you're adding cost. And I think it's going to be pretty hard to harden it against the thing I'm going to pop in low Earth orbit. That's what I would do. And then, you know, That's the one I like, but the Russians are already looking at this, right? They're looking to put nuclear weapons in orbit.

Aaron 40:01

I mean, my favorite is the volley fire. You have the Sprint or Spartan actually with a nuclear weapon on it, and you punch a hole and then you just follow behind it, even with fast and slow. I mean, space based nuclear weapons is the fun one. And spoiler alert, there'll be future podcasts on that for other work that we've been doing on this.

I mean, the other one is just to go all fun like the Russians have been doing. I mean, why just stop at one? Put it on an underwater torpedo that just kind of floats around out there. Put it on Bill Borovetshnik, the world's dumbest sort of environmental disaster, but it'll get the job done in a pinch. You know?

I mean, there's all sorts of fun stuff you could do. Sam, what's yours? What's yours?

Sam 40:49

Yeah, so as the kind of resident sprint enthusiast, I'm hard-pressed to come up with anything better than that. There are two others that I quite like. One is independent post-boost vehicles. Oh, yeah. Which the Russians and the Chinese are doing. Right, exactly. And so instead of having everything on one bus, and then if the interceptor hits the bus early, then it takes out all the warheads, each warhead has its own bus.

So you get all of the accuracy and independent targeting benefits of MIRVs, but you don't have the vulnerability that a single bus has. So I like that. I think that's fun. The other thing is, and Jeffrey, we've talked about this, is pre-positioning mobile launchers downstream of how the SBI constellation moves so that when you launch your silos and the interceptors gobble all those up, it's created this window in the constellation that's moving.

And so you have your mobile launchers downstream and you have them shoot through the window or down orbit, I guess.

Jeffrey 41:55

But so those are, those are, those are my favorite. Yeah, in theory, the, if you, if you spread the orbits, the constellation is supposed to heal, but you know, if you force us to expend a lot of those interceptors, you know, it, it, it can't fully heal very quickly.. The other one that I really like, which actually I think should scare the crap out of SpaceX and other space operators, is since this is a constellation that's flying in formation, the Russians and the Chinese don't need to get out of the way to allow that constellation to maintain formation.

And I'm not saying that they sneakily do it. I'm saying put debris up or put satellites up that just constantly are in the way and make them big enough for us to see. And then we got to move stuff. And that both will shorten the lifetime of the satellites. And it will disrupt the operational pattern, which by the way, if they start doing, that also has real downstream effects for Starlink and Planet and all of the companies that have small satellite constellations because they're going to be, the Chinese and the Russians, I think they're going to just be barging their way through in order to open up gaps and just generally interfere with these systems.

I might also I might also just routinely do large numbers of sounding rocket launches out of my, you know, out of locations near my silo fields.

Sam 43:41

That's a good one. But so I guess one thing that I wanted to say was that- I'm also very aggressive, but go ahead. These aren't just kind of notional ideas that we've come up with, right? Like the Russians and the Chinese have talked more explicitly about this. I know, Jeffrey, do you want to talk about the Antropov solution or the Antropov plan at all?

Jeffrey 43:59

No, you can go ahead.

Sam 44:02

So when SDI was hugely controversial, the Soviets obviously hated it. And a group of Soviet academicians put together a book where they basically talked about what they would do in response. And this ended up becoming known as the Antropov Plan or the Antropov Option. And they talk more or less explicitly about... All of these things. So they talk about putting mines in space.

They talk about doing passive countermeasures like messing with the sensors, reducing burn times, expanding the size of their force. And then they also talk about more extreme active measures. So they do actually say that they could just build a sprint alike and then put a nuclear weapon on it, and then that would be a big problem. That's the most straightforward way, by the way.

Later- Straightforward way, by the way. And then actually later, I believe it was Rose Gottemoeller was in China and she was talking with some members of the Chinese nuclear policy community and they said, yeah, we have our own version of the Andropov solution.

Jeffrey 45:10

Oh, I cite, there is a publication that I cite in my dissertation. Which is like literally where they cite, what is it? I think the book is called Mirror or something. It's, you know, yeah. I have both the Soviet book and I have the Chinese article citing it.

Sam 45:33

Yeah. No, they have their own list. Their own list. Oh, actually, so let me correct myself. I think the Soviets called it the asymmetric response, and then the Chinese called it the Andropov solution or the Andropov plan or something. But yeah, so they both put in a publication and then members of the Chinese policy community tell Rose Gottemoeller this when she visits China, I think in the early 2000s, and that gets reported somewhere.

Aaron 45:55

I think the point is that if we have come up with fun ideas at the bar, maybe not with fortified milk Mai Tais. Is that what it was, Jeffrey?

Jeffrey 46:06

Is that what it was, Jeffrey? Clarified milk. Clarified. You use the milk solids to clarify. You strain it out. There's no milk in the cocktail at the end. It just makes it clear.

Aaron 46:15

You're getting levels of cocktails above my head. If we could suss this out over clarified milk Mai Tais, the Russians, the Chinese, the North Koreans, who knows, maybe one day the Iranians, have come up with similar plans themselves or stuff perhaps we haven't even thought about here because they could be creative themselves as well. They have some gray matter between those ears.

It just costs a lot of money. And I think, okay, we're talking about this all the time now. But I think, Jeffrey, this is a phrase I stole from you. How much defense are we buying? So let's say you invest, I don't know, $50 billion. We can get something to size to North Korean silos that works for $50 billion, and we can slot it into a $1.5 trillion budget request from DOD.

Is – does it actually make us any safer? Does it actually alter the basics of deterrence that we have predicated nuclear strategy on since 19 – well, a little after 1945 but like for the past 80 years?

Jeffrey 47:26

Yeah. And my answer is, unless the Russians and the Chinese just say, oh, shucks, you got us.

Aaron 47:31

They sort of roll over on their back. Rub my belly.

Jeffrey 47:35

Rub my belly. If they don't do that, then no. I think you end up with a pretty classic Tom Schelling-like matrix where you'd both be better off if you didn't do this. But neither of you can believe that the other won't. So you each do the dumb thing and then you end up with the suboptimal payoff.

Aaron 47:59

Which is a leaky system that doesn't work and a lot of fancy Russian missiles that make sure it doesn't work. And you're just locked into this cycle. What gets me is that we're spending, again, this just goes back to classic strategy, right? Forget the – I mean, one cannot improve upon the Schelling matrix, but you're just spending money in weird places.

This distorts budgets in ways that like, do you want to buy sort of a leaky – Space-based missile defense system or would you rather buy a million new Tomahawks? I'd rather buy a million new Tomahawks if that was up to me.

Sam 48:38

Whatever. I want so many LRASs.

Aaron 48:42

My favorite missile, the SM-6. I want more of those bad boys. Those things are fun.

Jeffrey 48:48

And so just... You know, this is the famous Rand thing where they would end every presentation with, by the way, buy more munitions.

Aaron 48:55

I'm quoting old Rand report from the 1970s because I thought this was settled. We did this before. It's not even new or novel. It's 54 years ago. I think it was 72 was that famous Rand report. Maybe it's 77. My point is it's almost 50 years old or greater than 50 years old in terms of these decision-making payoffs, in terms of how you allocate finite defense dollars or war dollars, I guess, in our current environment.

Jeffrey 49:25

Yeah, I was looking, I was reading Tom Schelling's Reciprocal Fear of Surprise Attack. 1958, that thing is almost 70 years old.

Aaron 49:34

We'd like to think that we're very clever on the Arabs Control Wong podcast. These things have been thought about for much longer than I've been alive.

Jeffrey 49:42

Very, very, very clear. Very clear.

Aaron 49:45

Smart. That's my point. The Russians, too, have thought about this. You know, they're not going to pull the puppy and roll over and rub my belly. They're going to say, like, no, I'm going to go beat this thing.

Sam 49:58

Right. And it's going to be a worse security environment where everything is more precarious. Each side is kind of more willing to do something dangerous. And, you know, there's all these trade-offs, right? We will have compromised a lot of space launch capability, right? So we don't get to put up other cool things. Like ground-based target trackers or more GSAP or something.

And instead, we've built this kind of really silly thing that has only produced negative benefits.

Aaron 50:30

And SpaceX has already gone public. We don't even have time to get in. It's too late. So we can't even get in early and benefit off of this. We're at the fringes eating away at grant dollars to talk about this. We could have been on the SpaceX IPO.

Sam 50:49

The arms control wonk, get rich quick scheme.

Aaron 50:51

Jeffrey and I have never made any secret that we're open to advertisers on this podcast. Nobody's called. Nobody's called except for the bidders guy.

Jeffrey 50:59

I mean, the one thing I would say is we can build it now.

Aaron 51:02

We could put up 2,000 satellites. That is ultimately what spurred this project to steer this thing back on the rails before we all sign off.

Jeffrey 51:10

I'm not sure how much defense 2,000 satellites will buy us. But the Booz Allen guys, I think it was, what, $20 billion for 2,000 satellites was what they said they could do. And I think it will end up being a lot more than $20 billion now.

Sam 51:24

You need a lot more than 2,000 satellites, too.

Jeffrey 51:27

But the problem ultimately is 2,000 satellites won't provide much in the way of... Again, it's like buying a house.

Aaron 51:35

You put down your down payment, you move into the house, then every year or every month you get a pay mortgage, basically. And then on top of the mortgage, which is the replacement of your satellites in orbit, you have all of your maintenance costs when the air conditioning breaks, when you get a water leak in the shower.

And so there's all sorts of... Someone sounds like an embittered landlord, actually. Landlord. And so, yes, there's all sorts of craziness. Like perhaps the tile floor in a brand new shower needs to be replaced. Who knows where that comes from? But it happens. Does that mean we're off the rails when I'm telling bad stories about my landlord experience?

Jeffrey 52:15

Yeah, no one feels bad for you being a property magnate, Aaron.

Sam 52:19

Yeah, it's a very specific set of grievances that I'm not sure are super relatable. Yeah.

Aaron 52:26

All right.

Sam 52:27

It's so hard to be you.

Aaron 52:31

Now I feel like an idiot, which means we've gone off the rails. Anything else we need to add to this conversation as we wrap up here?

Sam 52:40

As another note, I just don't think it's going to happen. You look at the budget, you look at what isn't happening, you look at what is in the real budget and what's in reconciliation. I don't think it's happening. I think it's a bad idea, but I also don't think it's happening.

Aaron 52:52

Yeah, we should shout out the work of... Sam over, not you, Sam, the other Sam over at the Aerospace Corporation, who looked, Sam Wilson. Sam Wilson. Who looked at the budget numbers. Numbers. Although.

Jeffrey 53:07

Yeah, it's like the biggest single year increase for any service, you know, like. Yeah. Space Force.

Aaron 53:14

Yeah, they're not going public. They said, you know, there's been reports that they've run two tests of this system. I mean, it was – but I think we all assume that those are software tests.

Sam 53:28

Well, so I guess this is another thing I want to complain about very briefly before we wrap up is they said that there have been two tests of Golden Dome systems and they have not elaborated – Because somehow this has more secrecy on it than like anything that I've ever seen from a public program.

Aaron 53:44

And again, they're asking for hundreds of billions of dollars, but please go ahead and say it.

Sam 53:50

I wrote something small about this for FPRI members, but it's a really bad look when the Nixon administration was way, way more forthcoming with information than you are. But so yeah, we have no idea what's going on with Golden Dome. They won't talk about it. This is something that Tom Carrico actually wrote a really great op-ed about not too long ago, where this is a problem for both critics and boosters of missile defense, where you can't defend what you don't know.

And you know, Goodline keeps talking about how you know people aren't estimating the architecture he's building, which like I guess that's fine that your architecture goes to a different school in Canada, but I, you wouldn't know. It's a very difficult public policy problem and we should be talking about it in an open and democratic manner. so allegedly there have been tests, but it could be for like cruise missile defense or something. I mean, also, I mean, I hate to be look.

Jeffrey 54:41

I understand that you have to script tests because they are tests and that's just the nature of tests. But every time we get the details of a test, the scripting ends up doing a lot of work. And so I think it becomes really, really hard to know. Without those details, it's just so hard to know how capable a system is.

My teeny tiny little nitpick is we don't do missile defense tests in the rain. I know why, and I get it. Sure, you should test in a clear optimum environment, but also you don't have any sense of how reliable your system is in inclement weather. That's a problem.

Sam 55:37

And we don't even know if they're testing Golden Dome in the rain or otherwise, because they won't tell us. So anyways, yes, we anticipate your criticism. We have assessed the cost of something that you aren't building. So please tell us what you're building.

Aaron 55:48

And if you'd like to fund a follow on Sunday, again. That is a joke, everybody. All right. I think we've done it. We've hit the hour mark. This is one of our longer podcasts here. But it's good work. And the paper should come out on the FPRI website in September. Lo and behold, a lot of FPRI staff go on vacation in August.

I did last week, as did Jeffrey. And so things take a little longer to get out when we finish these things, but we're out. And so if you want to know more about this, join the Slack channel, patreon.com slash ACW podcast. $3 if you're a student, $5 if you're not. It is the only place on the internet where you can have conversations about SDI boost phase intercepts with Sam.

So with that, thanks, everybody, for listening.