#3 - Carl Robichaud: Solving the Nuclear Dilemma

31:40And the U.S. had a little bit of a head start and was able to exploit some of the vulnerabilities on the Soviet side in order to assure its NATO allies that it would come to their defense. France wasn't persuaded. And that's one of the reasons that they developed their own independent nuclear arsenal, because they didn't want to have to rely on the U.S. They thought that when push came to shove, a U.S. president wouldn't trade Washington for Paris. So you see this dynamic playing out again and again throughout the nuclear age. But there is an equilibrium, which is if both sides have a secure second strike, there's no incentive to moving first.

32:31And that is the equilibrium that most people describe as being stable within the world of deterrence. And the problem is it's always an unstable equilibrium unless both sides really accept it. And when you look at the evidence, the U.S. never fully accepted this. That their ability to do a fully annihilating second strike was sufficient? No, that they did not want to be deterred by the Soviet Union. Interesting, I see. So in effect, the Nash equilibrium didn't exist because the U.S. didn't accept the premise of a Nash equilibrium. Right. And part of it comes down to what people sometimes call the stability-instability paradox, which is if a conflict is stable at the highest order of violence, in this case, nuclear deterrence.

33:32If it's stable and if you're confident that neither side has any incentive to use nuclear weapons, then that creates opportunity to engage in all kinds of violence below that threshold. So you're actually saying that it can create more conflict, sort of constrained conflict up to a certain size. There would be more incidences of those than if there was no risk of mutually assured destruction above that. So if the Soviet Union believed that the U.S. would never use nuclear weapons because the U.S. would be deterred by the Soviet nuclear arsenal, it could be free to invade Germany. Right. That was what the U.S.

34:19feared. I mean, I've heard people make the sort of bull argument that actually nuclear weapons have been a good thing.

34:27that the major powers living under the threat of mutually assured destruction has made war less likely. And to be fair, the world, since World War II, has been in a relative period of seeming peace when you compare to all of history before that, where it just seemed like wars every now and then, it was just always wars going on everywhere. Would you agree with that statement? I would be interested to hear your sort of like steel man for the case of having nuclear weapons and for keeping the peace. So we've only had this history once. We've only have one Cold War. And I think that the fact that we didn't have active conflict between the US and the Soviet Union, a direct conflict, was overdetermined by a few different variables.

35:24I think nuclear weapons is part of that story. It led to mutual restraint on both sides because the stakes were so high. And even though the conflict was seen as very high, it forced the US and the Soviet Union to play out their competition in different ways. Proxy wars. Proxy wars, yeah. So like Afghanistan. Afghanistan, Vietnam, Korea. various conflicts in Greece and elsewhere. So there were, in the Middle East, you have all these proxy wars. The 20th century was peaceful relative to what it could have been if there were a third world war. But these proxy wars were serious matters and led to millions of people dying.

36:16So we didn't constrain war in every case. It just shifted away from the core of Europe to the periphery. And so you had all these conflicts, these proxy wars, as well as some of the post-colonial wars were inflected with Cold War tensions where the U.S. would back one side, Soviets would back another. Some of these became brutal civil wars throughout Africa and Latin America. So nuclear weapons kept the peace in a sense, but it was always an imperfect peace. And it's an oversimplification to only look at the Soviet Union and the U.S. as the key player. China becomes a really big player. And there are also lots of independent countries that have their own goals and aspirations that are separate from this Cold War conflict.

37:13And they actually don't want to be part of this game. They want to do their own thing. And so I think the traditional story told from Washington and Moscow oversimplifies and takes out a lot of the agency of people in other countries, from Vietnam and Korea, to lots of different states in Africa and Latin America. What breaks mutually assured destruction, national equilibria, in a sort of hypothetical situation? Are there specific types of escalations that reliably always break it? Yeah. Well, there are certain types of deployments, certain types of weapon systems, which create less stable equilibria.

38:10so for example you've heard about MIRVs these multiple independently targeted re-entry vehicles so these MIRVs allow you to place three or four nuclear warheads in some case ten nuclear warheads on one missile you can see why that would be destabilizing because a single missile that you launch could take up up to ten of an adversary's missiles In that case, it might make sense to go first. Because you could launch half your arsenal and take out 90 % or more of the adversary's arsenal, and then they are vulnerable to your nuclear blackmail. Right. So MIRVs come along, and this is a big threat to the stability, and countries try to respond.

39:06In both the U.S. and the Soviet Union, they developed sea-based deterrent. So let's put the nuclear weapons on submarines that can go and hide in the depths of the ocean silently. These are extremely survivable. At least the U.S. submarines could not be tracked and targeted. We later learned that the Soviet submarines were more vulnerable than we had thought. But the other way that the Soviet Union manages to ensure its survivability is by putting its warheads on these road mobile tells, these erector launchers that can travel by road and can also be hidden. So you don't know where they are at any point.

39:55But it's a cat and mouse game because as satellite imagery improves and as overhead surveillance improves, now you can start to locate where these tells are. So if your adversary's systems can be discovered, they become vulnerable again. So you have this constant cat and mouse race. We've now got sort of citizen intelligence as possible, Like a friend of mine who actually was just on this podcast, Bill Perkins, he's built a company that, or is creating a company that basically enables anyone to very easily order almost real-time satellite information. Which is great because it's got all those implications for, you know, you can track deforest, you can get citizen scientists to track deforestation, soil erosion, all these different kinds of environmental things.

40:51but they can also track, therefore, what's going on in countries, you know, in terms of military activities. So it really does, like, sort of change the nature of the playing field. It's essentially lifting the fog of war a lot, right? Because all previous nuclear dilemma situations, only the military had access to this kind of information, and their information was much more limited. There were fewer satellites. It was less, you couldn't necessarily see through clouds, that kind of stuff. Whereas now we've got such improved spectral technology and we've got the internet and everyone can talk about it and freely publish this information.

41:27Intuitively, it feels like that's a good thing because there's less cloak and dagger-ness going on and there's less unwarranted suspicion. Let's put it that way. And it makes it harder for individual nations to essentially defect. by saying, they're like, oh no, we're adhering to this treaty, we're not doing anything. Then the fog of war has been lifted and they can't just go and do that secretly. I think that's where it's been the biggest game changer is our ability to track proliferation developments. And by our, I mean citizens, right? You don't need to rely on the story that governments are telling because the commercial satellite imagery you can buy is as good as the stuff that the high-tech expensive satellites from the Cold War era, that's now stuff you can go to your computer and you can purchase it for a couple hundred bucks.

42:26And so you have different researchers, some at the Middlebury Institute for International Studies or at the Center for Strategic and International Studies or the Federation of American Scientists who basically conduct this citizen investigation into the state of the world with regard to nuclear weapons. They can find a new missile field in China that shows that China is increasing the number of missiles it's deploying. They can find suspicious facilities in Iran and in Syria. Now, we can assume that governments can all find this stuff too. They have even more sophisticated means of gathering information.

43:13But now that information is publicly held. It's not just the domain of governments. In general, I think that's a good thing, although it could have some unintended consequences as well. It certainly makes things easier to drum up political outrage, for example. Yes. You know, not just international actors doing bad things, but also if your own country is doing something that is sort of essentially defecting from an agreement, you know, doing something escalatory, then their own citizens can apply pressure to them, which historically didn't happen before. So it seems like on net that's a good thing.

43:54But it can also be used to generate false pressures. You can find a facility that's actually an industrial site, and you can say, this is a nuclear weapons facility. Look at these anti-aircraft guns around it. And unless you are skilled at discerning satellite imagery, people can tell a very plausible story that's totally false and that leads you down the wrong path. And by the time that lie is countered, it's already spread around the world six times. Yeah. The moral panic has already started and whatever knee-jerk reactions come from that. Yeah. None of these have a clear answer. There's always pros and cons.

44:37That's why it's such an interesting topic. It's like, how do we factor human irrationality into sort of these game theoretic models, right? Because usually whenever you read about anything that's to do with game theory, it's like, well, assuming a rational actor, then these are their payoffs. This is how we would model their decisions. But humans are very, very rarely rational. And presumably, again, like in all games, there are a plethora of cognitive biases that come in and cloud actions on both sides. And how do we model that degree of clouding? How do we model that irrationality? Yeah. So as a poker player, how do you become more effective and more rational with your game?

45:25The way I would go about it is I would just play lots and lots and lots until I felt like I'd seen enough scenarios and I was able to sort of discern between, you know, What is an example of a, when I have a run of results, were those results down to my good strategies or down to variance and luck? And over time, because I would then develop a good sort of feeling of when it's luck versus skill, I would be able to notice when my emotions were latching onto one of those, was I feeling good because I got lucky or was I feeling good because I was playing well? Am I being overconfident, underconfident?

46:07So generally it's just like practice, essentially. It's playing the same game over and over again and learning to discern the signals and separating that from the noise and refocusing your mental state. And I think humans approach more effective and rational behavior the more they practice decision-making. The problem with nuclear crises is, well, the good thing about nuclear crises is that they don't occur very often. The problem is that decision makers, if they find themselves in a nuclear crisis, it's going to be the first nuclear crisis they've been in. And they don't have years of experience.

46:50Now, there are ways to play war games and to try to simulate crises to give an artificial feel for what it would be like in that decision-making room. but it's not the same. And any war game you devise will be based on synthetic data and on simplified and false scenarios because you have to make assumptions about what the adversary is going to do. Well, what if they don't do what you think they're going to do? What if they deploy submarines with nuclear torpedoes on them which you didn't anticipate? What if they gave local aircraft, anti-aircraft gunners, the green light to shoot down your aircraft?

47:39What if they deploy tactical weapons to the Cuban mainland without telling you about it? These are the types of things that happen during crises. So you could imagine sitting down at a poker table where you're playing the highest stakes game of your life and someone has just changed the rules. And you're sitting down with eight players who you've never met before and you're all trying to figure out what the rules are together. And some of you are genuinely more rational in the conventional sense than others. Some of you really care about long-term consequences and do actually care about the health of your nation over the next hundred years.

48:22And some of them are just wanting to have bravado and want to appear like a strong man. Right, and all of this is happening in a really loud room with everybody shouting. You've got your buddies leaning over your shoulder and saying, raise them, raise them. And it's a chaotic environment because none of these games are played just with two players. There are allies who are involved. There are third parties. And it's going to be a very chaotic environment. In your line of work, do you actually run models that try to simulate these kind of things and add different weightings in terms of personality types of leaders and that kind of thing?

49:12There is a whole practice of war games within the U.S. military, within the Russian and Chinese military. It's a field in and of itself. And how do you create realistic simulations? The problem is that a lot of these are generated with a goal in mind. So, for example, they will generate a war game to test the capabilities of a new missile defense system. or of a new deployment strategy for a naval fleet. And so they're trying to hold everything else constant and change this one variable. In many cases, if you are trying to simulate a military crisis in the Western Pacific, you might even take nuclear weapons off the table because that's not the variable that you're interested in.

50:05So I've heard of some of these war games that just assume that neither the US nor China will resort to nuclear threats, which seems unrealistic to me if there ever is a crisis over the Taiwan Straits, for example. So you can practice, you can simulate, but it's going to be a... You might even get a false sense of security. You might take away the wrong lessons if you're putting in the wrong inputs into your war game or your simulation. I think that the key thing we can do for decision makers is instilling a sense of uncertainty and humility and try to understand where our own cognitive biases might be to really try to understand how you personally respond during a crisis and how the bureaucracy you're a part of might respond during a crisis?

51:10So a thing I obviously like to talk about a lot is this concept of Moloch, which is the sort of collection of game theoretic forces that make people optimize for the short-term strategic advantage. Because if they feel like if they don't do it, then everyone else is going to do it, so they'll get left behind if they don't. But on net, that creates bad outcomes. So in what way does the Moloch dynamic apply to the nuclear dilemma? Yeah, well, I think Moloch is how you get 60 ,000 nuclear weapons deployed in the 1980s aimed at each other. Otherwise, what's the possible rationale for having this many weapons?

51:59It all starts with this security dilemma, right? In international relations, you talk about the security dilemma, which is in a state of anarchy, as you are in the state of international relations, states, in order to make themselves more secure, build up armaments to address external threats. But in doing that, they inherently threaten their neighbors and their potential adversaries, who will then build up their own armaments, which then creates new threats. And so you have this escalation. This is the dynamic. This race to the bottom, essentially. Yeah. Where more and more money gets poured into military defense or offense away from what people actually need, like healthcare and infrastructure.

52:46Both sides would prefer to spend virtually nothing on defense and be able to provide for the needs of their people. But so long as one is doing it, the other has to. And it's very hard to create confidence. And coordination. Yeah, exactly. Right. For some reason, the individual states are not able to either sufficiently coordinate or to sufficiently trust each other, even if they can verbally speak and come to agreements, but they can't sufficiently trust each other. And so this crappy thing ends up happening anyway. It's this multiple party collective action problem. The multipolar trap, there's many names for it, but Moloch is sort of a name that at least I'm trying to catch on.

53:34So that said, you mentioned that's how we ended up with 60 ,000 warheads in the mid-'80s. But actually nowadays, we're down to, what, 12 ,000? So that's, to me, a real shining example of us globally managing to break out of a Moloch trap. We've managed to reduce the number of warheads we've created seeming mechanisms in order to coordinate to do that in a way that sort of presumably from the individual's perspective didn't even seem possible at the time. So how did that happen? Because I think it's actually kind of amazing. Yeah, so a lot of people hear this term arms control and they think specifically about treaties.

54:23But I want to reframe for a moment and think of the word arms control as being anti-Moloch. Okay. Because if you think about arms control broadly, it's a variety of mechanisms. It can be transparency mechanisms, communications, hotlines. It can be unilateral restraint. Or it can be formal, bilateral, or multilateral treaties in which a side is legally committing itself to a course of action. Those are the mechanisms through which you overcome this security dilemma. and you credibly signal to your potential adversary that we're both better off not running this race. And there are times when that will work and there are times when that will not work.

55:13A lot of it depends on the broader geopolitical tensions. Some of it has to do with allies and other things that are going on in the world. So it's not 100 % effective. But when it works, that's how you escape this treadmill. and sometimes we've been pretty effective at this starting with nuclear testing is a good example so everyone's better off if there is not atmospheric nuclear testing because it leads to health effects that spread all over the world and affect everyone but it required a joint agreement by the US and other countries to no longer test above ground And then they had a secondary treaty that was a comprehensive test ban treaty.

55:59So we're not going to test above ground and we're not going to test underground. So that's an example of how states were able to escape from this situation. Treaties that limit the types of warheads or where you can place warheads. There's an outer space treaty that both sides have agreed not to put nuclear weapons in space. That's a good thing. So how did the first of these treaties, the first one was in 1968, right? The Non-Proliferation Treaty. That's a big one, yeah. There was actually the Limited Test Ban Treaty came before that. But the Nuclear Non-Proliferation Treaty is the core treaty that limits the spread of nuclear weapons.

56:41Right, that was the one that said, because at the time there were five states that had nuclear weapons already. and basically the world got together and said, that's it, no more states, get them, we're capping this here, this is enough of this madness. So, first of all, like how... What were the mechanisms that enabled that to emerge? Like, was there some kind of third party that was created in order for that to happen? Well, it revolved around the United Nations and these sets of negotiations. But the reason it succeeded is that it was backed by the U.S. and the Soviet Union and China, who were the dominant nuclear states at the time.

57:28and they pushed on their allies to say, we will provide you some security assurances if you sign on to this treaty. There were other parts of the treaty as well, including a promise to share the benefits of nuclear technology. So it was a technology facilitation treaty as well. So we're going to allow you nuclear power and all the other benefits that come from this technology. We're going to actually facilitate that. And we agree to this Article 6, which states that we will work towards the cessation of the nuclear arms race and the ultimate elimination of nuclear weapons from the world, even though there was no timeline given for that.

58:11But it was this combination of real bilateral pressure plus this sense of international norms. I think that sounds like the key ingredient, The fact that the three or so major players, US, China, and Russia, all had the political will to say, look, we need to make this thing happen. Why did they have that political will? Was it simply just like sheer fear? Or was it because it was the fortunate situation where they also benefited? They were all essentially being rational and could see that it would be good for them if this stayed only like three or five. That's good for them because they could maintain something of a monopoly on this technology.

59:00And you may initially think that you would want nuclear weapons in the hands of your allies because that would make them more powerful, but it would also make them more independent from you and you don't know if they're going to always be your allies and your friends. And given how powerful these weapons are, they are a great equalizer. But there was very much a sense that these weapons are different, and we run real risks if we allow them to spread everywhere because of their potency. And if you look at the projections right after the Manhattan Project, initially they thought that nuclear weapons would not spread quickly just a few years later.

59:50The Soviet Union tested their first nuclear bomb. And at that point, everybody started freaking out because they realized these technical barriers are much lower than anticipated. And you had dozens of countries that were exploring nuclear technology. And that was also facilitated by this U.S. Atoms for Peace program, which was working with U.S. allies to develop nuclear reactors and share in the benefits of atomic energy. It was sort of this Cold War scheme to draw countries closer to the U.S. and to enmesh them in the alliance. And by virtue of that, there were lots of states that had the start of a nuclear weapons program, including countries like Switzerland, Sweden, South Korea, Japan, as well as a number of other countries.

1:00:44And so the intelligence community projected there could be 10 or 15 nuclear weapon states. John F. Kennedy famously said we could have 20 nuclear weapon states. He said that in 1962. So there was a real fear that the cat was out of the bag and this technology would spread. And that mobilized these countries because they didn't want to live in a world in which every country had the capability to destroy their neighbors. This treaty, it was sort of this anti-Moloch moment where we can see the path we're going to go down if we have all these programs. And it started as a very weak treaty, but over the course of several years, there were 81 countries who signed on to it.

1:01:32Today, it is the most subscribed treaty in the world, I think. And the vast number of countries have agreed never to acquire nuclear weapons, which is kind of a remarkable thing. And some of them even gave them up, right? South Africa had nuclear weapons, and they were like, you know what? We don't want this. And willingly gave them up, which is a real testament to the fact that even something as sort of large and amorphous as a country can make such a selfless act, essentially. It seems like a key thing of treaties is the ability to be able to verify that others are adhering to it, the agreements that have been made.

1:02:17So how does the verification process work? So there is this International Atomic Energy Agency, and their role is to ensure the peaceful use of nuclear technology. This is the IAEA. The IAEA, exactly. You can just call it the agency if you want. No, it'll be great. So they were established before the NPT, actually, as part of this Atoms for Peace program. And countries sign comprehensive safeguards agreements with the IAEA, which is a way of saying, here's what you can monitor in our country, and we're going to be transparent with you about what's coming in and what's coming out in terms of nuclear materials and technology.

1:03:02And this is a way for countries to assure their neighbors and others that their programs are entirely peaceful. Now, we know that there are some holes and flaws in the system. It's not 100%. And North Korea, for example, gained technology under the Nuclear Nonproliferation Treaty, then left the treaty and built a nuclear bomb. In 1991, after the first Gulf War, they discovered that Iraq had a secret nuclear weapons program that they were hiding from inspectors. They had disclosed some of their nuclear facilities, but they had a separate set of secret facilities right across the hill. and that led to a new approach to IAEA safeguards embodied in this thing called the additional protocol in which now you had to have really complete access for nuclear inspectors.

1:04:02And these technical agencies gained a lot of competence and they can take environmental samples, they can investigate and look at all kinds of data and logs and now they have access to satellite imagery. And so it's pretty hard to avoid detection from the IAEA, but it's not impossible. It's not a foolproof system. And it does rely on the willingness of states to enforce the nonproliferation treaty and states enforce it unequally. And so in Iraq, the second Iraq war was driven primarily by concerns that turned out to be false, that Saddam Hussein was seeking to acquire nuclear weapons. It's possible that a more effective verification system could have assured everyone that this nuclear weapons program was no more.

1:05:07I mean, we knew Iraq had a nuclear weapons program. They said they had dismantled it, but they continued to hide some information. When you looked at all of the facts, it became clear that their nuclear weapons program was over. But the U.S. intelligence community didn't believe that. Well, some would argue that that's because they had incentive to not believe it, right? And they wanted to invade anyway, and it was a handy excuse. I mean, do you have thoughts on that? I think that the US thought that Saddam Hussein had a nuclear weapons program, but exaggerated the extent of it and the seriousness of it because Bush wanted to invade Iraq because he saw this as an opportunity to reshape the Middle East and to solve this problem for once and for all.

1:05:57So the alternative was to continue imposing a sanctions regime, which was messy and incomplete, and let this problem linger. So he said, I just want to solve it. And they embellished and exaggerated the data in order to go in and conduct that war. It was the most recent treaty, it looks like in 1996, the Comprehensive Nuclear Test Ban Treaty, right? The CTBT. I love their acronyms. which bans all nuclear explosions, whether for military or civilian purposes. How well has that one been going? It has a really strong technical basis. They've expanded their monitoring network and have a lot of cooperation around the world.

1:06:38And most experts believe that if a nuclear device were detonated, it would be detected. that treaty is not fully in force because it requires not only the treaty to be signed by a majority of states but also there are specific countries which are mentioned that need to be parties to the treaty the US never ratified the treaty this was a big controversy during the Clinton years neither has Israel, Pakistan, India So there are a number of countries that would need to sign on to this treaty in order for it to be in force. But even absent it being fully in force, there is this organization with a lot of technical expertise that can provide assurances that people aren't testing nuclear weapons.

1:07:36And that is a benefit to all of us. Right, because they're pretty hard things to do on the quiet. But even if you do them underground, even with North Korea's clear defection from that. Although, did they actually ever sign that treaty, North Korea? No, they didn't sign it. Okay, so they're technically defecting. But they're defecting from the norm. The organization still discovered their tests. Right, exactly, because they were doing them underground, right? But it creates essentially an artificial earthquake, and we can detect the signature of that. And we can determine the ways in which it's different from a natural earthquake.

1:08:10What does it look like? I don't know. It's a very technical analysis. They break down the vibrations and everything. It's super cool. We've had these treaties. We've still got the IAEA, which has been incredibly successful. Actually, with the IAEA, who are the key individuals who helped create? Because that's another example of a third party, like a kind of impartial overlord, so to speak. that managed to at least maintain a sufficient appearance of impartiality such that the major players would trust it and opt in and follow it. So I want to understand what the mindset and, I guess, appearance of those individuals were in order for that trust to exist in the first place.

1:09:07Because we're going to get into talking about how the lessons we can learn from the successful things in defeating Moloch in nuclear apply to new emerging technologies like AI and biotech and so on, which are arguably even more dangerous. So I want to understand, what were the sort of personas of these people in such that they enabled these other parties to trust them? Well, I think that I don't know the actual people who were involved in this. I'm sure there's some interesting characters along the way. I would like to know more about that. The interesting thing about any treaty is that they tend to start off very weak, and they start off as a lowest common denominator.

1:09:55And through negotiations, over time, each side pushes on the other side, can we do a little bit more? How can we be sure? So when the treaty was first signed, China and France, who were acknowledged as nuclear weapons states within the treaty, They didn't sign on. And there were many other countries that didn't sign on as well. But over time, people could see this is actually something meaningful, and it's worth supporting. And you had these forums. And the UN is much maligned, and I can understand why. Often it's just a talk shop, and it doesn't get much done. but it is a place where you can raise and deliberate over some of these big collective action problems that don't fall into the domain of any particular state.

1:10:49So I think it was Ireland actually who proposed the first meeting for the IAEA. So this is a small state that for a long time has punched above its weight, so to speak, in terms of advocating for better nuclear policies. Are there any new promising types of coordination mechanisms that you are either seeing emerge or that you think we should be trying to help with the nuclear security dilemma? I think we're in a real crisis moment because you have, for the past 50 years, you've had a treaty regime that has constrained the U.S. and the Soviet Union, later U.S. and Russia. That is going away, the final treaty, the New START treaty.

1:11:37Russia has suspended its participation. In 2026, that treaty expires. There's no negotiations right now because of the war in Ukraine. So barring some kind of unforeseen cooperation, we're going to be entering a new world. And it's a scary time because, as you mentioned, China is increasing its nuclear arsenal. And there's a risk of this new three-way arms race with really high stakes. We need to find out, we need to find some new mechanisms for cooperation and risk reduction or we're going to end up in a really dangerous place. Because right now, at very little cost, the U.S. could double the number of deployed nuclear weapons.

1:12:25Russia can almost double. And we know that China is going from 300 to 1 ,500. We don't know that. We think that China, we suspect. So this is the scenario we find ourselves in. And so what can be done? I think you can have, for example, Short of a treaty, you can have individual statements that we won't go up if you don't go up. Here are the ways that we can share information. Even absent a formal binding treaty, there are things you can do. You can also agree not to develop certain capabilities. I think one capability I would really like to see us avoid developing, first of all, cyber attacks on key nuclear command and control infrastructure.

1:13:13Let's take that off the table. Let's take off the table anti-satellite attacks on key nodes of nuclear command and control. Let's have a code of conduct in which both sides agree we're not going to cross that line. Let's have an agreement not to develop and deploy artificial intelligence systems that have access to nuclear launch authority. I think AI, it's inevitable that AI will be part of the next generation of warfare and probably will be involved in nuclear command and control at some level as a decision guide. But let's just decide now that we are not going to allow an algorithm to make a decision, a go or no-go decision on nuclear weapons.

1:14:04And these are the types of conversations we should be having now because once a technology is mature and developed, it's much harder to roll it back. Are there any novel technology solutions that's coming out with some of these things that we should be exploring? One particular one is we've seen, obviously, blockchain tech enables the idea of trustless verification. given that so much of treaties is about verifying that the other side is not doing something bad. Is there a way we can harness that technology? Potentially. So there's a particular problem in nuclear arms control which relates to trying to measure, count, and monitor nuclear warheads.

1:14:57So all the treaties so far have monitored and counted delivery systems. missiles, bombers, submarines, because these things are large. They're observable. Whereas a warhead is very small, and unless you're on site, even if you see the warhead, you don't know if it's a real warhead or if it's a dummy, right? And so some scientists have been working on what they call a zero-knowledge protocol in order to assess an object and determine whether it's a nuclear warhead from a distance without revealing any of the proprietary design information of that warhead. And this is an area in which a lot of innovation will be required.

1:15:46And I think some of the blockchain technology might be useful for this. The zero-knowledge protocol is rooted in cryptography. And so there is a basis for thinking that innovation here may help. What about space-based defense? So right now, we're at a moment where space capabilities are evolving rapidly. And especially as launch becomes less expensive, we're putting more stuff up. Right now, we continue to rely on a small number of very expensive, very sophisticated satellites that are part of the command and control. And that creates vulnerabilities. So I think we need to be moving towards an architecture in which we're using more off-the-shelf technology to provide redundancy for our nuclear command and control.

1:16:42Now, you asked specifically about space-based missile defense. Now, this is a real challenge. Missile defense in general opens up a whole can of worms. This gets back into the Mollick question, because one person's defense is another person's offense. And right now the U.S. is the leader in developing missile defense technology and especially strategic missile defense technology. And this is one of the things that's driving increases in the number of warheads in China and potentially in Russia because there's a fear that the U.S. could gain an advantage in missile defenses. The U.S. says that its missile defense is tailored towards a narrow set of circumstances regarding North Korea and Iran, but Russia and China don't believe that.

1:17:37Right. And so this will destabilize the sort of Nash equilibrium? Yeah, I think it already has. If you look at 2002, which is when the U.S. pulled out of the anti-ballistic missile treaty, the ABM treaty, that's when you start to see attitudes around nuclear sufficiency change in Russia and in China. It took a little while to play itself out, but I think that China's decision to go big with its nuclear arsenal is driven in part by a desire to be safe and secure even in the face of improved and expanded U.S. missile defenses. Because a shield is a weapon just as much as a sword is. I think there is the potential for space-based missile defenses, but there are a couple approaches to it, and all of them are really expensive and complicated.

1:18:38And so we have to decide, is that a path we want to go down? Because we'd be spending, the U.S. would be spending a lot of money to develop a capability that an adversary could overcome at lower cost. Right. That might work if you're competing against North Korea, which has such a smaller economy that you may be able to outspend them 100 to 1 and still come out on top. But you can't outspend China or Russia 100 to 1. So you would be entering into a new arms race that's not going to end well for either side. Are there any other types of tech? We touched on civilian intelligence, commercial satellite imagery.

1:19:23There's a lot of exciting technology that's coming out related to nonproliferation. Just being able to remotely monitor things, our ability to pinpoint objects anywhere in the world and track them is increasing each year. One of the problems we face is that the International Atomic Energy Agency is slow to adopt the latest technology. It's a conservative organization. It's governed by its member states, many of whom are really keen to maintain their sovereignty. And they don't want very intrusive technology being placed into their nuclear infrastructure. What are the main constraints acting on the IEA?

1:20:15On the agency? One is budget and cost. So it's zero real growth. I think it's about$300 million, and about half of that is safeguards. So small. Yeah. But what it does. We're going to do a fact check on that, and I'm going to make sure I get the right number. But it is very small. It's like the safeguards budget is like the police department of a small city. Yeah, I saw that, actually. Yeah, it's like the police department. The same size as the police department of Washington, D.C., is the budget that the agency spends on safeguards, which is just mind-blowing, seeing as it's arguably the best coordination mechanism we have against the Q &A.

1:21:03Dollar for dollar, this is some of the best money we spend on international security. And if people lose confidence in it, that's a real problem. Can outside funders donate to it? Yes. They now have a policy of accepting outside contributions so philanthropists can contribute. I will say that a lot of member states are skeptical of outside money and wonder what's motivating this. Right, that's the thing you're always having to deal with because there's always some allegiance, presumably, even if they don't, people are going to assume they do, etc., etc. So it's just so difficult. When I was at Carnegie Corporation of New York, we partnered with the IAEA on a couple of projects, but it was challenging because there was a sense well, what's their agenda?

1:21:45Why are they doing this? And we're based in the US, so there are a lot of countries who like the safeguards regime the way it is because it allows them some optionality. So are there any lessons we can use from the sort of coordination mechanisms that have been successful in reducing the nuclear war risk can apply to the AI dilemma that we're now facing, the biosecurity dilemma. You know, all these novel dual-use technologies, and by dual-use, you know, that means technologies that can be used for incredible good, but also incredible damage and harm. Because, I mean, that's what nuclear is as well, right?

1:22:30It's like one of the best things we've ever discovered in terms of like extremely cheap, non-climate change causing energy. But then obviously you can create nuclear weapons out of it. You unlock that key to that technology and you get this whole Pandora's box of stuff. Same goes with AI, same goes with synthetic biology and beyond. So what lessons can we draw from the nuclear situation that we could apply to these new dilemmas? I think it's challenging because AI and biotechnology are these really powerful multiple-use technologies. They're kind of general purpose in many ways, whereas nuclear weapons were born out of a military program.

1:23:15They were controlled by states from the start. AI was born from the private sector. And so it's a national regulatory problem. It's an international regulatory problem. And the landscape is moving so quickly that there are very few people right now who are on the frontier and understand what's possible and what would you actually limit, monitor, and regulate if you wanted to do that. There are also real benefits to moving forward quickly and deploying AI in all kinds of different use cases. So there's going to be a lot of pushback against any attempts to limit this technology. Right, like how do we manage that tension between the peaceful, wonderful uses and managing the risk?

1:24:05Yeah, we're kind of in 1939 with AI. There has been no Hiroshima. There's been no Nagasaki. We haven't seen how devastating this technology can be. Instead, you have a group of scientists who are speculating based on being highly informed about physics to understand, I can see where this technology can go. and we ought to be doing something now to prevent it from going everywhere and having these really negative consequences. But the rest of the world has not yet acknowledged these harms and I think you can see that. There was a hearing today in the Senate. You could see a lot of senators were having a hard time wrapping their heads around the implications of this powerful technology that we're bringing into the world.

1:25:05And the cone of uncertainty is pretty broad. So if we look at the nuclear era and think about what we might learn, I think one lesson is that developing and listening to highly skilled technical people is key because you've got to be monitoring and tracking the right things or you're going to miss out. You're not going to build an effective system. Technical competence is key. Second, you've got to start somewhere. As we saw with the birth of the IAEA and the NPT, when these things first started, they were weak, they were incomplete. A critic or a skeptic would look at it and say, this doesn't do anything.

1:25:59And in fact, even after the NPT, there were a couple prominent failures. You know, just a couple years later, India tested what it called a peaceful nuclear explosion. And a lot of people at that point said, the game is up. This thing doesn't work. The NPT is a failure. Fifty years later, we have only nine nuclear weapon states. And in fact, the vast majority of countries in the world, the overwhelming world's population, has said, we don't need these weapons for our security. And they've signed the treaty, but they've also enacted domestic law. And they have stated that this is not part of their national identity.

1:26:45It's not irrevocable. It's possible they could go back and decide at some point to build nuclear weapons. but it would create so much friction and opposition to do that. Yeah, no, it's actually a really interesting point because with the AI situation, we've had this letter that was just signed proposing a six-month pause in terms of deploying any systems greater than GPT-4. And I have to say, I in the end didn't sign it. I was asked to sign it and I decided not to because I felt one of the main reasons was it felt sort of insufficient. I was like, this isn't actually going to really do anything.

1:27:30It might even let some actors catch up, et cetera, et cetera. But your point is actually counter to that, which is like, it's okay if the sort of first coordination agreements are seemingly insufficient. If we look at the lessons from history, they all started out a little bit insufficient, but actually worked out. And so, you know, don't let perfection be the enemy of the good, essentially. That's an interesting counterpoint that I hadn't considered sufficiently. Now, there is a problem to that, which is that some of these agreements can be kind of sticky. So that the initial safeguards arrangements that the IAEA agreed on, they changed over time.

1:28:11But then they got kind of sticky and it was really hard to update them. And it took the shock to the system in 1991, where Iraq's nuclear weapons program was discovered, to lead to an update. So there are certainly concerns. And the IAEA is not and has not been a perfect institution. And there are some ways in which it has facilitated the transfer of sensitive information. International organizations can be leaky. But overall, it created a framework for countries to sign on to to say this is what we aspire to. And that would allow you to separate out the sheep from the wolves because the countries that adhered to their agreements consistently, you kind of stop worrying about those and focus your attention on those countries that appeared to be doing dangerous things.

1:29:02And then, of course, there's the issue of having an incomplete agreement or an insufficient agreement in that it might create a sense of false security. Exactly. Yeah. So how to mitigate that? I think it's challenging. Something like AI, we don't know whether there will be advanced signs that things are going wrong. Because these systems are inherently hard to interpret. At least the current cutting edge systems that people are deploying, they're relatively opaque. So you could easily come to a false sense of security. I think if you're really attentive, you can probably pick up on warning signs.

1:29:48I think as well in the biosecurity space, if you're attentive, you can pick up on warning signs and you can devise a system that's going to provide that kind of early warning. But it's a really challenging problem. One that I always thought was promising is like wastewater detection. Yeah. detection methods for looking for for like exponentially growing novel bits of dna in waste water of like metropolitan areas for example would be one way of spotting yeah there's something weird growing is there something going on here that we haven't seen before um yeah fairly uh and from what i understand fairly fairly cheap mechanisms but anyway well i'll save that for when i get the bio risk expert on um what's a more general question then like how to manage the tension between having these sort of centralized control mechanisms, such as giving the UN more power for example, which a lot of people are concerned would create the risk of global tyranny.

1:30:55Or global incompetence. I mean that's the other risk, right? Exactly. The WHO is not viewed as the go-to agency for this stuff because I wasn't that impressed with them. They didn't really, I don't think they got a particularly good score for COVID. Yeah, so like, you know, if you give more power to like a centralized international force, essentially, which in theory makes it easier to coordinate and punish defectors from agreements, you then run the risk of tyranny. but on the flip side of that if you don't create some kind of centralized coordination mechanism then you're in a world where it's you're getting more and more democratized access to increasingly deadly weapons um you know you're essentially giving the powers to end the world to more and more people um of which a percentage will always be essentially insane out and or malicious or just completely incompetent you know in that world then you're incredibly vulnerable um and and that kind of world wouldn't last very long basically so like how do we manage that tension um the daniel schmachtenberger who i had on a sort of prior to win-win um in a conversation he often talks about this like need for a third attractor that we seemingly is like not on that dimensionality between control and anarchy.

1:32:31It's like something separate to both. It gets the best bits of both without being vulnerable to the worst parts. Do you have any thoughts on how we can do that? There are some examples of corporate self-governance in which you have groups of companies that realize that their livelihood and well-being and ability to continue doing business self-regulate and can demonstrably show to international and domestic regulators that they're doing the right thing. I think the airline industry is an example of this. The nuclear power industry, there are lots of them. And one of the benefits of that is you have people at companies who understand deeply the technical dimensions of the problem.

1:33:14Now, there are limits to that. If we do live in a really fragile world, that's probably not going to be enough. In general, countries respond to warning signs and to harms. I think one challenge we might look at is climate change, in which over the past 30 years, we have very slowly started to bend this arc away from some of the worst climate change scenarios. it's far more costly to do it now and there's going to be all kinds of harms and suffering especially by the world's poorest people because we didn't act sooner but we are going to bend that curve and we're going to avoid some of those worst outcomes according to the current projections of the IPCC and the technology is emerging that will allow us to have a carbon-free and abundant lifestyle.

1:34:20So for some of these complex collective action problems, if you were to look at the problem 10 years ago, you might say we're all screwed. But over time, people start to take a risk seriously, they take action, and they're willing to make sacrifices if they understand what's at stake. I think with AI and bio, it's likely to be messy and there's likely to be some real harms that are inflicted upon all of us, there's a chance that we get it right. Going back to nuclear in particular, what can people who are watching do, right? Like, I think I want to be careful of, you know, if I'm going to be having these kind of conversations, you know, they're intellectually interesting, but at the same time they're absolutely terrifying in many ways, and I often myself feel quite helpless.

1:35:14And it's overwhelming, basically. It's like, oh, it's this giant geopolitical game and I'm just a little pawn. I'm not even a pawn. I'm like a piece of dust stuck to the chessboard that might just break at any minute. What can ordinary citizens do to help? Well, I think we can engage with the political process because these are threats that affect everyone alive today as well as future generations. By default, we will not end up in the best outcomes because of short-term political incentives, because of bureaucratic pressures. We have forgotten collectively about nuclear risk. We've had this 20 years of a relatively benign security environment where we could worry about marginal nuclear threats.

1:36:12We're headed back into a world in which nuclear weapons are going to be increasing, I believe, in the world, and in which those nuclear risks are going to be increasing. and the solution is to exercise mutual restraint and to engage in diplomacy. But that is really politically hard and costly. In the US right now, there is a tremendous amount of anti-Russia sentiment and I empathize with that. It's really awful what's happening in Ukraine right now, But we've got to find a way to cohabitate with a country that has all these nuclear weapons and in which we actually have a lot of shared interests, similarly with China.

1:37:09that's not going to happen unless we create political space and unless we rein in some of the worst case planning some of the rampant spending and we think about what we need in order to have sufficiency of security rather than always seeking more power and more control over the world you're saying basically create political pressure to incentivize politicians to factor in the nuclear risk? Well, support candidates who don't feel like we need to have as many nuclear weapons as Russia and China combined, for example. That we can achieve security with a nuclear arsenal that's sized to our national security needs.

1:38:04Like the status quo? Yeah. You're saying that the status quo is sufficient? Let's hold the line. I think the status quo is sufficient. So look for politicians who actually accept that the status quo is sufficient, and even if Russia and China start tripling or whatever in their... Don't play that game. Because it's not actually necessary. And in the end, we're all going to be in a worse place if we go down that road. And let's ensure that we are okay with our politicians talking with Russia and China and trying to find solutions to the crisis in Ukraine instead of pushing for maximalist and appealing, emotionally appealing positions that don't actually lead to a resolution of this war.

1:38:51Right, justice for justice sakes type things. Avoid those. Yep. If you live in the U.S. right now, the president has sole authority to use nuclear weapons. There are no checks and balances. This runs counter to everything else, and the U.S. Constitution and the U.S. political system. It's the ultimate unilateral power. Ultimate unilateral power. And that has persisted. It's time to end that and to limit presidential authority. Now, there are some logistical reasons why under certain circumstances you might want to empower the president to act promptly and unilaterally. But we should... And there are several bills out there that would require the president to go to Congress before launching a first strike with nuclear weapons.

1:39:47Right. That seems just common sense. You kind of need to keep the unilateral power for second strike in order to maintain the threat of mutually assured destruction. But for first strike, that needs to be a more democratic process, essentially. So we should support bills that restrict presidential authority in ways that make sense. Yeah, that's a very good point. And we should continue to align our nuclear targeting strategy with the laws of armed conflict. And my view is we should work towards a world without nuclear weapons. Now, that's ambitious at the current moment when countries are building up their nuclear arsenals.

1:40:33But most countries in the world have decided they can live without nuclear weapons and don't feel like that is essential. Are there any promising philanthropic opportunities? Like if people want to donate to something, what are the best options? Yeah, there are some organizations, and I can put those in the show notes if you'd like as well. I think that there are a couple categories of action that are needed. one of which is to rebuild the analytical base at universities and at NGOs to be able to address this rapidly changing puzzle that we find ourselves in. The non-governmental sector has played a key role.

1:41:15If you look at the 80 years of nuclear history, I've talked about this as something of a success story. You haven't had any nuclear weapons used since 1945, and they've only spread to nine countries. That's in part due to the engagement of citizens and civil society experts in academia and at think tanks. But we've lost that. It's$20 million a year that we're spending on this. It's a tiny fraction. Given the scale of the problem. That's it? Yeah, it's pretty sad. $20 million a year for nuclear security. Yeah, non-governmental work, yeah. Yeah, compare that to, I've got some stats here to compare that to.

1:41:58There's 10 billion a year spent on climate philanthropy. So that's, what is that, 400 times more spent on climate, which obviously climate is a huge problem that affects the whole world, but arguably it's less urgent and probably going to kill fewer people than if nuclear war happened. And then 60 billion is donated each year to higher education. These are people giving to their alma mater. Right. But going back to what you might actually spend some of this money on, what are specific solutions? Yeah, so one of these are to increase dialogues, citizen to citizen, scientist to scientist dialogues.

1:42:38So these played a really key role during the Cold War and over the past 30 years to bring together U.S. and Russian scientists and influencers to discuss what would the solution space look like. Is this like an organization that makes this happen? What's it called? For example, the U.S. National Academy of Sciences does this, and there are a couple other organizations. They're kind of like conventions that bring scientists from academics, basically from these different sort of competing countries together to like brainstorm. Yeah. Okay. Very knowledgeable and experienced scientists in some cases who have worked in government on these very problems.

1:43:18Another piece of this is rebuilding this infrastructure to advocate for saner nuclear policies and to counteract some of the beltway pull towards developing and deploying more weapon systems and more types of nuclear weapons, some of which would actually make us less safe. So an example of this is nuclear-armed cruise missiles. a sea-launched cruise missile can be armed with a conventional or a nuclear warhead. When it's launched and it's detected by an adversary, they don't know where it's headed or whether it has a nuclear or conventional payload. So you could see why that might be really destabilizing.

1:44:08You could be firing just a standard cruise missile that inadvertently trigger nuclear war because they don't know that it's not that. You need to paint something on it, like a smiley face to be like, this isn't a sad face, this is a bad one. At least make it... I think it's better not to deploy those, not to develop and deploy those. So does the Biden administration, and they've come out and taken this out through their nuclear posture review. They did an extensive review. They decided we don't need this weapon and it makes us less safe. Congress has gone ahead and put it back in to the budget.

1:44:42So that's an example of the type of action that's needed by citizens and groups to keep this system in check. Are there any other organizations that people should donate to? Well, I'm a big fan of the Nuclear Threat Initiative, the Plowshares Fund, the Federation of American Scientists, the Council on Strategic Risks. There are a lot of organizations in this space that are doing really good work, and most of them are pretty strapped for cash right now. And so counterfactually, if you're in a position to make a donation to one of these organizations, you will get a high return on your investment because they're able to do a lot with a little right now.

1:45:27Pie in the sky question, but if we did actually manage to solve scarcity through technology or through AI or some other emerging tech that we can't even imagine yet, would nuclear war risk disappear? I think we should try it. Think about where the world has come. There is less competition over scarce resources than there were before. Less competition over land and natural resources. The biggest source of wealth today is human talent. and that's an area in which we don't need to embrace scarcity in which we can lift people up and continue to make progress. And if that's the case, there is less need for geopolitical conflict.

1:46:29And I don't think we want to live in a world in which all of us hang under this nuclear shadow in which the decisions of a leader can lead to our death and to the death of everyone we care about in the course of an afternoon. That's not a world we should want to live in. And we can choose our way out of this problem. Humans invented nuclear weapons. We design the systems through which they are directed at each other. and in a world with less scarcity and more empathy and understanding, I think that we can leave this trap that we've made for ourselves.

1:47:21Thanks so much for watching, everyone. And thank you to Carl for letting me pick his amazing brain on this. Lots of further reading on this one. So a bunch of links in the show notes below. Feel free to dig in. Also, if you wouldn't mind, please you know like share and subscribe particularly with such an important topic like this uh you know the the algorithm is a bit of a fickle mistress uh so any boost that this can get would be much appreciated thank you so much may your next couple of weeks be full of win-wins and i'll see you next time

1:48:03Thank you.

1:48:33Thank you.