 I'm Scott Sagan, professor of political science and senior fellow at the Center for International Security and Cooperation at Stanford University. I'm really pleased to be chairing this session, a dialogue with James Acton, the Jessica T. Matthews Chair and the co-director of the Nuclear Policy Program here at the Carnegie Endowment. Are we all set? Is that better? OK. I want to begin with a full disclosure. As James knows, but as you should also know, I was a reviewer, an outside reviewer, for this very important international security article before publication and pushed James trying to make what I thought was also was a very good paper, even better. And I think he's done that. To me, this is a very important article published in international security. I should mention that we don't have copies for everyone here, but there's no firewall, no paywall in the international security website, so you can download a full copy of it. There's also a policy brief outside where James and some of his international colleagues talk about this problem of entanglement. I admire this work greatly because James has a very unusual capability to bring technical knowledge and political knowledge together in his work. And the questions of entanglement of command and control and the risk of nuclear escalation require analysis that is both technical in nature and political in nature. I would say that the analysis is entangled, but that is the nature of the article. Escalation to entanglement, how the vulnerability of command and control systems raises the risk of inadvertent nuclear war. I'm going to give James 10 minutes to outline the central arguments. Then he and I are going to be in dialogue about some of the theoretical conceptual frameworks, one or two technical issues, and then we're going to be focusing at the end on ideas, some of which are in the paper and some which have been developed subsequently, about what can be done to reduce, or at least mitigate, some of the risks that he identifies in this piece, and then I'll be opening it up for Q&A. So James, five to 10 minutes, and look forward to you giving the summary of the main arguments. Thanks, Scott, and many, many, many thanks to you for agreeing to do this today. The paper discusses this idea of entanglement. And incidentally, that's not my term. Credit goes to the late great John Steinbruner for inventing that terminology. But entanglement describes what I see as growing interactions between the nuclear and the non-nuclear domains. It has lots of different manifestations, delivery systems that can have both nuclear and non-nuclear warheads, nuclear and non-nuclear forces that are co-located. But the two specific forms of entanglement that I zoom into in this paper are, firstly, the dual-use nature of command and control, many command and control capabilities. Early warning satellites, for instance, that have designed both to detect an incoming nuclear attack, but also to trigger ballistic missile defenses against non-nuclear ballistic missiles. And the growing vulnerability of these dual-use capabilities to non-nuclear weapons. Now, entanglement, this form of entanglement is on the rise. 20, 30 years ago, we didn't worry about cyber weapons. Now that we have increasingly digitized command and control, we really do have to worry about cyber weapons. The US command and control system has always had dual-use components. But it's becoming increasingly reliant on those dual-use components. And the conventional war-fighting doctrines of the United States, Russia, and China all appear to call for attacks against command and control assets as a way of trying to win a conventional war. And that, essentially, is the root of the escalation risks in this paper. And in particular, it's an idea that I term incidental attacks. So an incidental attack is when one attacks, one state attacks, another states dual-use command and control system. Potentially, purely for the purpose of winning that conventional war, at least not losing that conventional war. But in the process, degrades its opponent's nuclear command and control system. So to give you a very concrete example, imagine NATO and Russia are fighting a conventional conflict in Europe. Imagine NATO's regional ballistic missile defenses are proving effective at blunting Russian non-nuclear ballistic missiles. We have indications from the Russians that in that circumstance, they might try to shoot our nuclear early warning satellites out of the sky. Potentially exclusively because of the role of those early warning satellites in triggering our non-nuclear ballistic missile defenses. But by taking out our early warning satellites, they've had the incidental effect of degrading our nuclear early warning system, which, for reasons I'm about to go into, I think has the potential to be very escalatory. What I want to emphasize about these incidental attacks is two things that I think people haven't paid existing attention to. Firstly, a number of authors. And here I highlight, I think, the brilliant, in particular, Caitlyn Talmadge's brilliant article in IS about US-China conflict and the vulnerability of command and control. And as Caitlyn really focuses on command and control assets in the theater of operations, what I want to highlight is many of these command and control assets that might be attacked in a conflict are space-based or very distant from theaters of operations, such as land-based radar. And even more importantly, it's not just US attacks on Russia and China that could prove it escalatory. It's Chinese and Russian attacks on the US as well that could prove escalatory. There's three escalation mechanisms that I paint in the paper, which let me just explain them very briefly now. And let's go back to this US-Russia scenario. Russia attacking our US early warning satellites. Even if Russia is purely motivated by trying to win that conventional war, at least not lose that conventional war, I think there's a risk that we interpret attacks against early warning satellites as the prelude to nuclear use. If Russia was considering nuclear attacks against the US homeland, Russian strategists say if we were to do limited attacks, maybe they get shot down by the US ballistic missile defense shield. And this is optimistic, but I think it's what the Russians worry about. And if it is what the Russians worry about, then they might very well proceed limited attacks against the homeland by taking out early warning. So one possibility is what I term misinterpreted warning. We interpret what they're doing as the prelude to nuclear use. Second form of escalation is something that I call the damage limitation window. As the recent nuclear posture review illustrated, the US nuclear strategy involves something called damage limitation. In the event of a nuclear war, trying to reduce the damage that we would suffer in a nuclear war. I don't think this is a new policy, by the way. I think it's always been part of US nuclear strategy. I think the framers of this posture review chose to highlight that. I don't think we'd be successful doing that against Russia or China, but it's part of our strategy. And we would require exceptionally sophisticated commander control capabilities to spot mobile missiles, to trigger ballistic missile defenses. So even if misinterpreted warning didn't arise, even if we didn't think that Russia attacks on commander control with a prelude to nuclear use, we might still worry that if the war turned nuclear at some point, our commander control architect would be so degraded by that point that we would be unable to have any prospective damage limitation. I think that could be deeply concerning to decision-makers in Washington. And I think that could, in both cases, that could lead to escalation. The nuclear posture review actually threatens nuclear use in those scenarios. Very explicitly, the US nuclear posture review says that in the event that command and control assets associated with nuclear forces are attacked, the US would consider the use of nuclear weapons in that scenario. So if you take the posture review literally, and I think it was meant literally by the people who wrote it, we might use nuclear weapons in the event of those scenarios. I think that's fairly unlikely, but not impossible. I think we might resort to nuclear threats in that scenario. I think we might say if you continue attacking command and control, then we would use nuclear use. If they interpreted our threat as a bluff, we then might be in the position of having to use nuclear weapons to live up to the threat. Or I think we might take very escalate reactions to preserve the command and control system. Imagine at the beginning of the war in an effort to keep that war limited, we said we're not going to do deep strikes into Russia or China. And imagine that anti-satellite weapons deep within Russia and China started to take out US satellites. I think that would be a strong incentive to break the limit that we had initially imposed and conduct deep, non-nuclear strikes into Russia and China. I think those could be very escalatory. There's also a third escalation mechanism in the paper, which I'll just very briefly flag now, which is the more traditional problem of crisis instability that many others authors have written about, which is where a state becomes concerned that its nuclear forces are deeply vulnerable. I think this form of entanglement could increase that problem, but that's less new, so I'm not going to focus on that today. Let me end by just saying two or three words about two or three sentences about what to do about this problem, and then I'm sure we'll discuss that in more depth later. I absolutely think there is stuff that could be done cooperatively. And I in no way want to downplay the value of cooperative and risk mitigation between the US and Russia and the US and China. I think there are very interesting ideas about no attacks against nuclear, no cyber attacks against nuclear command and control that are well worth exploring. I also think US-Russian and US-Chinese politics right now don't permit that kind of serious intergovernmental discussions, let alone agreement. So my pessimism in the paper about cooperative measures, and it is pessimistic, is not because I don't think they're worthwhile, but because I don't see them as viable in the short term. I think the best we can do right now is unilateral measures. One of the intellectual inspirations for this piece is Barry Pozen's book, Inadvertent Escalation, I think from 1991. And Pozen made a suggestion at the end of that book that I fully acknowledge I'm basically repackaging what he said. But he said, one of the big things we could do is reorganize ourselves internally so that military and civilians charged with defense policy think more about these escalation risks that they are considered at all stages of the acquisition, war planning, crisis management cycle. That for me is the most valuable thing. It would be both valuable and I think doable. So many of the risks that I'm talking about arise from perception or rather from misperception. Really just awareness of the fact that we may be misinterpreting what the other side is doing and they may be misinterpreting what we are doing. Even just as simple as that raised awareness, I think could help mitigate some of the risks that I'm talking about. So it's not necessarily easy to do, but for me the most plausible path to risk mitigation here is by creating constituencies within both the civilian and the military parts of the Pentagon so that they are more aware of these risks and they factor them in to acquisition, war planning, crisis management, arms control planning, all of the different areas to which this would be relevant. That's great. That's a great start. Let me first push you on the conceptual framework and how you're trying to apply it for different countries because there's one conceptual framework but I think you'd acknowledge right from the get-go that Russia and China have very different capabilities. So could you help us, examples you were given were Russia focused in which I think the case is the strongest that there are concerns about damage limitation and more explicit vulnerabilities on the Russian side and capabilities against us. Can you apply it to China very, very briefly? So you do that in little less detail in the paper but to which degree do you worry about misinterpreted warnings, damage limitation windows or crisis stability concerns with respect to China given for example, they have no I think plausible case to be made that they could limit damage in a way through their escalation decisions. So I think that's exactly right. So in the case of Russia, to the extent we understand Russian nuclear doctrine, Russian planners appear to prepare for attempting damage limitation against the US. If Russia, so Russia might, if Russia might attack early warning satellites, it might attack communication satellites. Try to win the conventional war. We might interpret that as preparations for a large-scale Russian strike. There's no chance of us misinterpreting, of imposing that kind of misinterpretation or perhaps correct interpretation vis-a-vis China. As you say China's forces are a lot less, a lot smaller than Russia. China would have no capability of doing damage limitation against the US. So I'm not worried for instance if China started attacking dual-use communication satellites, I don't think it's very, very unlikely we would misinterpret that as preparations for a first strike. But this kind of different forms of misinterpretation can still arise. Okay, well what's the example in China? So for example, I would assert whether or not China's no first use doctrine is real or true, there is significant concern in the Pentagon that it's not true, that in a conflict China would use nuclear weapons in a limited way. But agree or disagree with the Pentagon's analysis, I think if you read the nuclear posture review, they pretty strongly hint in there they don't believe Chinese no first use. So in that case, we might still worry that if China were losing a conventional conflict, if our ballistic, exactly the same scenario with Russia, if our ballistic missile defenses were blunting Chinese non-nuclear missiles, and China started attacking the Sibir's constellation and there is the early warning system and there is very, very clear evidence this is considered within China. It's one of I think the strongest evidentiary point about Chinese or Russian doctrine is that China really thinks about these attacks against early warning assets. We might interpret those again as preparations for limited nuclear strikes either against targets in the region or potentially but less likely against the homeland. In both cases, I think that would be extremely escalatory. I think we would be intensely worried about that even if China doesn't have the ability to disarm the U.S. No one wants to be on the receiving end of a nuclear strike even if you retain your nuclear forces in that strike. Okay, so let me push on this question of if there is a conventional or a non-kinetic or a kinetic conventional attack on a command and control asset. You outline worries that the United States either because of the nuclear posture review or because of statements in a crisis saying don't do that or don't do that again or else we might use nuclear weapons might feel compelled through mechanisms that I've called the commitment trap in the past to go forward and potentially use a nuclear weapon not because they planned in the sense of intent but because they were caught making a deterrent threat and felt that they had to follow through less credibility be doubted in other kinds of scenarios. Help us assess how seriously you take that because on the one hand, I've worried about that that's caused crises in the past. On the other hand, this president seems to be able to wiggle out of commitments all the time. So I hate to put a silver lining on some of the back and forth statements that President Trump has made but it is stunning that he seems to be less compelled to follow through on threats that he's made than other presidents. Thoughts or? Cool. Well Scott, two points. I mean firstly is not just a commitment trap the escalation that could go on ahead. I mean partly it's, I think a lot of what worries me is robust conventional operations to protect command and control assets that rises the level of violence and drives us closer to the nuclear threshold. So the escalation mechanisms don't just revolve around the commitment trap here. In terms of Trump specifically, I'm not gonna try and play amateur psychologist with the president. I have no idea in a crisis whether or not he would feel obliged to uphold US declaratory policy. All I would say is this is not just about Trump. I mean, I think the mechanism, the nuclear posture of you creates a kind of threshold in shelling terms now that future administrations may choose to keep that language in place. And so it's not just about Trump, it's about how future presidents are likely to behave in this scenario. As I say, I worry more about us making a threat in a crisis. If you continue to attack these satellites, then we will use the nuclear weapons. That threat being interpreted as a bluff and us feeling committed to live up to that threat. I worry more about that than I do our saying because we said it in the nuclear posture review, now we have to use nuclear weapons straight away. And I think a threat uttered in a crisis probably has more likelihood of being followed through than a threat made years previously in peacetime. But if it's not just about the president, you could also imagine DoD officials because of it is our posture, therefore we have to follow through. And it would shape plans, it would shape briefings, it would shape options presented. So I think that's an important thing to have identified. I would certainly agree with that. So a couple of technical questions. One, you have a very interesting section in which you talk about cyber attacks against command and control, in particular against warning or communications. Citing the 2007 reported Israeli use of cyber operations against the Syrians just prior to or during the attack against the reactor in the Syrian desert. This gets into very technical work right away but give us a sense or convince us that the problem of cyber attacks against warning systems doesn't have to be as vulnerable as the Syrian Air Force was. But obviously you're going against much more capable forces 10 years later. So how do you assess the vulnerabilities to a cyber attack against a warning system? I don't assess the vulnerability. Like I don't claim to know about the vulnerabilities in the US early warning system. What one can do is point to things like the Defense Science Board. Now the Defense Science Board who have received briefings I haven't and have clearances that I don't have assessed that it is quote, impossible for the Department of Defense to fully defend its networks. And their point is that the scale of the threat depends on the sophistication of the adversary. In the case of lower level non-state actors they're kind of more bullish about the possibility of defending absolutely critical IT systems. In the case of highly sophisticated, very competent state-based actors they're not massively confident about the possibility of defending IT systems. The other thing that I would point out here is I think and this is somewhat speculative, it's kind of hard to prove. But my guess is that the dual useness of a number and this is not something I say in the paper because I can't prove it but if you kind of say it verbally then and it's not on paper then you can get away with it. But the, it has been recorded so I was joking. The dual, I suspect the dual useness of many of these commander control capabilities contributes to their cyber vulnerabilities. How so? If you take a system like advanced extremely high frequency which a communication satellite used for everything from special forces operations through to nuclear operations. And you think about the number of users that system has and the number of ways there must be at interfacing those systems and the speed with which users demand access to those systems. You're creating intuitively, for me at least, more potential points of vulnerability. If we really had a nuclear only system we can discuss the pros and cons of going down that route later. But if we really had a nuclear only system that had just a small number of users you could imagine having extremely rigid access controls and stuff like that. If you expand the number of users I think inevitably you're likely to introduce challenges to controlling access into that system. And challenges to insider threat problems and all sorts of facilities. If you want to penetrate somebody's system the more people you have using it for multiple purposes the different kinds of security clearances involved the different opportunities one might have in that sense. You could see a compromise that could do this as well as a technical. Indeed, absolutely. One last technical question. You talk about the reduction in redundancy in US command and control and you just mentioned the AEHF satellites. So we've gone down from multiple systems to not these aren't warning these are to give emergency action messages to nuclear forces to four systems that are the same. How independent are they? Because it strikes me that for a system to go wrong it's not the actual number. It's not redundancy in that narrow sense. It's redundancy in the sense of how independent of the systems are they overlapping so that one going down won't make that much of a difference or are they common mode errors or are they not overlapping in their coverage so that it could be a real problem one vulnerability creates massive vulnerability. Convince me and convince the rest of the audience that that's a big problem. So let me underscore your first point which is the reduction on redundancy within the nuclear command and control system. By the end of the Cold War the US simply had many more command and control assets and many separate systems. So it's your right to highlight there's kind of two different forms of redundancy. There's just redundancy in pure number of assets and redundancy in number of systems. And I think this has big implications. It depends on the threat. I think this kind of redundancy becomes important. If we were only worried about kinetic attacks if we weren't worried about cyber if we were only worried about ground-based anti-satellite weapons blowing up satellites building more and more of the same satellite would potentially be a useful response to this. We don't have the money to do it but if we put up 100 identical advanced extremely high-frequency satellites into orbit which would cost like $500 billion or something but if we were to do that that would, you'd have huge challenges in shooting down that number of satellites. However, if all of those satellites rely on the same software and they have the same then potentially they have a common mode software vulnerability. So in theory one could imagine a single cyber attack being able to disable multiple different satellites because they all run on the same software. So this distinction between redundancy of numbers of assets within the same system and redundancy of different kinds of systems I think is an important distinction. And building more of the same guards you against some threats it doesn't guard you against others. So before we talk about solutions I want to do one creative thing which is paper and some of their earlier work with co-authors really focuses on Russia first then China second but this problem is present in spades with North Korea. It strikes me that there's some unusual asymmetries here far more asymmetric capabilities between the US and DPRK than between the US and Russia and even US China much more of a hair trigger posture on both sides in the region. I thought something I've coming out in the very near future about imagine that the Hawaii incident occurred in North Korea. We had that false warning in Hawaii it didn't cause big panic here in Washington because people had redundant systems that could say no there's no attack coming against Hawaii second because there were professional people who said there's someone screwed up and it could report correctly up above and there's nothing going on. And third because we really didn't think Kim Jong-un was about to attack Hawaii but if that occurred in North Korea not the identical problem but similar kind of false warning A, they don't have much redundancy B, it's their professionalism they have yes men or cover up men who would not want to report on what it just they don't want to report a screw up because they might get killed not fired for doing it and three I think Kim Jong-un does think we might launch a surprise attack because we keep telling him we might launch a surprise attack. What I haven't seen done and I don't know if you want to outline it on the spot I'd love to hear you about how do these ideas about a potential command and control and a conventional military operation could lead Kim Jong-un to think that we're about to execute a damage limitation nuclear attack and how should we think about whether this is magnified even more in that particular context? Well, I don't claim to have massive expertise on this so what I'll just do is quote my friend and colleague Jeffrey Lewis' book on this issue some of you may have seen he published a book recently the 2020 commission and as you were talking I realized that part of his way his fictional scenario that generates a war between the US and North Korea involves North Korean command and control Absolutely So you know I would just And a false warning of an attack I will credit him Jeffrey but I will just point blank just steal his scenario here Jeffrey's belief and this is all footnoted in the book is that North Korea's cell phone network and yes there is a cell phone network in North Korea is used for everyday communications between North Koreans with cell phones which I think is about three million users conventional military operations and nuclear military operations So it's an entangled if that's true and if that's really the only nuclear system that North Korea has well if that's true it's an entangled system if it's really the only communication system North Korea has for sending nuclear execution orders that's a really entangled system and what happens is there's a conventional attack from South Korea sorry there's kind of spoilers here but I think you know how the novel ends there's a very limited conventional attack from South Korea against North Korea as always happens when there's a big incident the cell phone network gets overwhelmed because people are starting to phone one another and you know are you okay? But because there's a shared cell phone network for conventional operations nuclear operations and everyday phone calls the overwhelming of the everyday phone calls part of the network overwhelms the military part of the network and all of a sudden it looks to the North Korean leaders like there's been a cyber attack against the nuclear command and control system so I've studied the US, Russia, Chinese examples in great depth to the extent we know the Russian Chinese command and control systems and I should just preface it by saying there's big uncertainties there I don't have massive knowledge of the North Korean system but I suspect that governments wanting to do stuff cheaply create huge incentives to entangle your assets and I suspect that is true in North Korea just as it's true in other places as well Yeah and they need to do things cheaply so let's talk about the two main mitigation actions you talked a bit in your brief outline about questions about declaratory policy this idea of having risk reduction teams I think you're uncharacteristically modest for our Washington think tank guy to say that you built all this up on Posen I think you go way beyond Posen so I want to compliment you on that but how would this risk reduction team work you've got like multiple people from different agencies trying to push their agency to cooperate with to me it's a great idea I just have a hard time envisioning how it would work bureaucratically So it's a great question the first thing that I've changed my thinking about a bit since the paper was published is to be convinced into believing that you have to have somebody really senior in the bureaucracy responsible for driving this through and entailments are well in particular, yeah entailments are you actually want somebody in the existing bureaucracy and in particular the first I think ought to be made responsible for this is the undersecretary of defence for policy okay the undersecretary of defence is already legally responsible for reviewing US war plans to make sure they are consistent with civilian guidance I believe that's actually the only legal responsibility the undersecretary of defence for policy has that's a very, very serious senior position so the first thing that I would love to do is make the undersecretary of defence responsible for managing inadvertent escalation risks not just those related to entanglement by the way but all inadvertent escalation risks now clearly somebody that senior is not going to spend much of their time on that but you have somebody who's legally responsible for doing it and then can empower the people below him or her who are actually charged with doing that so in terms of how these teams would work you know I think you would want to have people with a whole group of expertise you know, intelligence experts who understand Chinese and Russian thinking technical experts, strategists, military planners so to give you one example the US at the moment is in the process of just starting to recapitalize our nuclear command and control system I worry we're just going to rebuild the same command and control system again only with more sophisticated satellites I would love people inside the bureaucracy to be able to review these plans to identify problems associated with them and to be able to alert senior decision makers and propose alternatives up to those senior decision makers you know, the secretary perhaps in some cases even the president they're presumably not on nuclear control systems but you know it would be up to those senior decision makers to decide how they're going to weigh up escalation implications against traditional military considerations, cost all the other factors they have to consider but you've got to have that voice in the bureaucracy saying hold on guys, there's things you haven't considered here in the case of war planning you know, I think that the way we conduct a convention is hardly original but the way we conduct a conventional war I think has significantly can alter the likelihood of it going nuclear in the process of while the people under the under secretary for policy are reviewing war plans you know, to make sure they comport with civilian guidance have then comment on the likelihood of escalation risks have them propose ways that you could alter those war plans to reduce those escalation risks so I don't necessarily view this as the trick is not giving them a vague amorphous thing it's identifying all of the specific processes you care about arms control planning, war planning crisis management acquisitions having them comment on all of the planning and preparations and documents that the bureaucracy is generating the whole time to kind of hardwire consideration of escalation risks into this enormous system very good, very good the discussion of declaratory policy in the paper is brief but focuses mostly on what we shouldn't do like we shouldn't have a nuclear posture review that says we might use nuclear weapons under these conditions and you have some language that would soften or at least create more ambiguity in the threat that create more wiggle room and would also make it a little bit more believable than what's the current threat but you don't go the full direction of a commitment and I'm curious about why it seems to me that while I have understood even though I've disagreed over the years with people who argue in favor of the US first use threat myself favoring a no first use declaratory policy you could get more people on board with a no first attack against certain kinds of command and control we understand that even in a conventional war we are not going to attack certain satellites why couldn't you have a declaratory policy that could constrain leaders first unilaterally and then potentially having this turn into some kind of treating mechanism so one of the ideas that I do and again this is I think Richard Danzig was the first person to propose this at least I've seen publicly you know one of the things that I do like the idea of exploring is an agreement for no cyber attacks against nuclear command and control you could absolutely extend that to no attack of any kind no conventional kinetic action the challenge I think is defining the scope of that commitment and this is complicated because so many command and control assets are dual use you know if we lived in a world in which nuclear command and control systems were completely separate from non-nuclear command and control systems and incidentally I actually don't think it's technically possible to live in that world but put that aside if it were technically possible to be in that world I would absolutely support trying to get unilateral and cooperative commitments against non-nuclear attacks on nuclear command and control it becomes a lot harder though if all of these are entangled assets they use both for nuclear and non-nuclear operations I mean you know that you then get into cost trade off cost benefit analysis about you know this asset is only peripherally involved in nuclear command and control but it has huge benefits for conventional warfighter we've got to leave that one on the table and the you know the Russians and Chinese would be saying yeah but early warning satellites your communication satellites they're all massively dual use my gut feeling here is that making a vague declaratory statement that we wouldn't do that would just merely it would it wouldn't be credible without us defining what we meant by that I think that has to be a cooperative process but I think we have to the only way to make a system like this workable is if we all share definitions of what constitutes nuclear command and control what are legitimate targets for attacks and what aren't I think we just have to discuss that cooperatively so I'm kind of skeptical without those definitions of the value of doing it unilaterally I massively support the idea of government discussions intergovernmental discussions to try and flesh out what those definitions might look like excellent well I'd like to open this up to questions and comments from the audience I would like you to be brief if you could identify yourself and your affiliation and again be brief and add a question mark like isn't that right at the end of your statement gentlemen right here hi I'm Michael Martel from the National Security Archive I'm kind of building off that last question if there was such a commitment to not go after nuclear command and control do you see any scenario where that would actually encourage entanglement from someone like China Russia or North Korea who's at a conventional disadvantage and would then see that as a mechanism by which they could extend their nuclear deterrence over their conventional forces yes is the I mean I'll give you a slightly longer answer but yes is the sure haunts to that question you could I asked them to be brief you could actually say more than one word so there is there is there is no so one of the implications of entanglement is that it potentially deters attacks against dual use assets if you attack my dual use asset even if it's just for conventional war fighting purposes all hell could break loose because you're attacking my nuclear command and control system and that is that is an argument I think that you're starting to hear in the Pentagon against trying to disaggregate nuclear from non-nuclear command and control my colleagues Alexei Abatov and his team in Russia and Li Bin and Tong Zhao in China I edited a paper a Pekarniki report is actually out there that you can pick it up on the way out called just called entanglement that we published last year and what Li Bin and Tong Zhao say is China did not deliberately entangle their nuclear and non-nuclear forces now that they are entangled some Chinese strategists see advantages in keeping them entangled so essentially what you're doing there is you are reducing the risk of an attack by increasing the consequences of an attack ultimately it comes down to a matter of judgment about how much risk is there in the system I can't I won't disagree with the principle of the argument that some risk is good for deterrence my overriding sense is the risk is too high at the moment there is too much risk in the system the escalation the bad side of the escalation risk is out of sync with the good deterrence benefits you would get from that I don't know how to prove that that's just kind of at the end of the day what I've reached and not least because I think a lot of policy makers aren't aware of the scale of the risk obviously it's very easy for me to think that it's very difficult for me to convince even my own government that I'm the Russian and Chinese governments that that's true I think this process would have to start with internal classified assessments by each of the three governments to ask themselves what are the benefits and risks of entanglement and I think you know in principle I don't think there's any reason there's no I can see no argument against each of the governments independently of one another on a classified level assessing for their own self how the benefits and risks of entanglement compare to one another and hopefully that internal process might start people thinking oh the risks are a bit higher than we thought they might be here maybe we need to do something about it but that is obviously you know as I said I have enough trouble convincing my own government of this kind of thing convincing the Russian and Chinese governments is harder another question German right up here Hi James Eric Gomez from Cato so one thing that you mentioned before that interests me a lot in my thinking over there is how the U.S. conducts conventional operations especially in regard to China things like airsea battle you know going after deep doing deep strikes against enabling forces do you think that changes in how the U.S. conceptualizes deterrence vis-a-vis China could reduce some of these risks and namely I'm thinking of if allies had their own effective kind of A2AD zones that they could establish would that do things to sort of slow down the pace of conflict and reduce the necessity of those U.S. deep strikes or do you think that has do you agree or what's your assessment of that kind of thinking it's a great question Eric I haven't I mean I haven't thought about that question in that particular way before so let me give you a general answer and a more specific answer my general answer is I absolutely absolutely believe that the way we conduct a conventional war has huge implications for the likelihood of its turning nuclear and this is a very hard thing to say but I will be blunt we fight wars the way we do for the very noble cause of reducing casualties amongst U.S. forces prosecuting that war if doing that increases the likelihood of the war turning nuclear then obviously the benefits of that kind of strategy could massively outweigh the costs and I think we have to potentially plan to fight bloodier conventional wars that have less risk of escalating and you know like people I think there's actually an interesting similarity here I mean you know Georgetown, Dartmouth colleagues Libra and Press interestingly we disagree on a whole load of nuclear policy stuff when it comes to the risks of escalation from a conventional war 100% agree with that and Bridge Colby has written a lot about this as well so you know people with very different views about nuclear all can worry about the same risks of escalation out of a conventional war and you know that might involve things like deliberately preserving adversaries command and control assets to make it clear that we're not going after the nuclear forces the specific question of whether allies having better A2AD capabilities mitigates that risk or not I'm not sure I can give you a great well thought out answer here and now I think it's a very good idea from the general extended deterrence perspective for allies to have better conventional capabilities because that really leads us to lean less heavily on nuclear weapons but that's a slightly different question from the one you asked you always have this risk of course that from a Chinese or Russian perspective I think they look upon the US and our allies as homogeneous entities you know fundamentally allies are just doing whatever we told them to which is I think not true but I think is fairly deeply held belief in Russia and China so there's always a risk that you know if you try to empower allies it's all just a big US plot anyway I'm going to have to think more about your specific question about whether ally capabilities reduce these escalation risks resulting from entanglement more I don't have a great answer for you here and now go in the fourth row here yeah that's you thank you so much for such a great presentation my name is Valera Lozava I'm currently affiliated with CNS as a Fulbright visiting researcher and my question is you mentioned the thresholds and generally deterrence theory points out that setting out clear red lines is helpful against variations in possible interpretations and obviously now those lines are blurred in both Russian and US strategies concerning entanglement is there any way to make them clear yes kind of it's a great question Valera I'm good to see you there's kind of a two-part question I think let me just say in general terms I think red lines are helpful sometimes and they're not helpful other times I mean I wouldn't I wouldn't commit myself to any kind of blanket statement about the value or not of having very clear red lines I think what is true is entanglement in some sense blurs the distinction between conventional nuclear conflict and at some fundamental level that's where all of the escalation mechanisms that I discussed stem from I think the nuclear posture review tried to make this distinction clearer you know the threat to use nuclear weapons against attacks on nuclear command and control was a way of highlighting you know setting out a red line against attacks on nuclear command and control I think it was a mistake to threaten nuclear use I don't think it was a mistake to highlight the importance of these assets to the US I mean if it was up to me I would have said something like dual use assets are integral to the US command and control system and we will respond to attacks on them accordingly and I wouldn't be specific about how we would respond but I would make it very very clear that these are special assets technically there are in some cases there are things you could do and not we could build a completely separate communication system for nuclear forces and non nuclear forces there are pros and cons of doing so I could talk for hours about that subject but from a technical possibility we could just build two completely distinct communication systems might be a challenge to convince the Russians and the Chinese that our nuclear communication system wasn't a dual use communication system but you know we there's no reason we couldn't do that and that would create some clarity I actually think that kind of clarity is not possible with early warning for a very very specific reason which is Russian and Chinese dual capable missiles incidentally I mean we have dual capable systems as well it's not just them but if but let me look at this from the US perspective you look at a Chinese missile like the DF-26 which is reportedly can be equipped with both the same missile body can have nuclear or conventional warheads if we can't distinguish from the outside whether a missile is nuclear or conventionally armed even if we wanted to have two completely separate early warning systems we just couldn't do that so I think there is and there's other technical challenges as well but I think like dual capable systems are the clearest one and as I say you know we have those as well like you know the B-2 the B-52 the B-21 in future it's not just them so there is stuff there is some stuff you can do to demarcate the boundary between the conventional and nuclear domains but I think that boundary you could make the boundary somewhat less blurred than it is but I don't think you can ever make it a clear red or black and white whatever colors you want to choose well let me add one point and see whether you agree with this seems to me that our analysis of these kinds of problems often doesn't disentangle two phenomena one is the clarity or lack thereof of the red line and the lack is the clarity or lack thereof of the threatened response we call that also the red line problem but that's not right so think about the 2010 nuclear posture review which changed U.S. declaratory policy with respect to chemical or biological weapons use it didn't get rid of a threat to respond to that red line someone who uses chemical or biological weapons against us or our allies but it changed the response to be not a ambiguous we hold open the right to use nuclear weapons George W. Bush's posture review but rather we will have a devastating response and we'll hold anyone both the leadership who ordered an attack and any military officer who conducts the attack responsible for the consequences in this area I would think that you'd want to change from the we hold open the right to use nuclear weapons if someone attacks the command and control asset to something more like that I don't know whether it'd be I don't like the term proportionate response because that's actually has a different meaning in the laws of armed conflict but an appropriate response a in kind response or a similar response or what kind of declaratory policy would you make deter someone from attacking a command and control nuclear command and control asset so in the paper all I was so you know in the paper all I was suggesting doing was kind of drawing an intellectual asterisk over these assets you're saying these are special and you have to be aware that there will be severe consequences for attacking them I was very vague I wasn't even vague I didn't even try to indicate what those consequences would be one of the comments and it's an issue that I'm thinking about you know one of the comments you wrote when privately to me when you reviewed the early draft of the paper which I didn't address in the paper that has never left me is how should the US respond to attacks on the command and control assets and I'm kind of trying to think through the answer to that and I can't tell you what this is a really difficult problem I can't tell you I have a great answer to that um I would point out it may and this is it may very well be scenario dependent you know if we're pretty sure we're going to win this war and it's just a matter of like a few days longer and you know there's frankly not much they can do maybe we'll just prosecute the war and we win the war and we don't respond in any particular way if the war hangs much more in the balance and you know this is a huge fight and what they're doing is enormously important to us maybe there has to be a much more kinetic response I think it's a challenge to work out what that response is that doesn't create huge escalation risks I can't tell you I have a great answer to that but all I point out is if your response to something is massively scenario dependent it's kind of difficult to threaten something specific in response to it here Anna here right in the front thank you Mr. Stott's garden I'm so pleased to to attend this because I think identify yourself first oh yes of course Veronica Cartier I'm an active in think tank in nuclear policy I think entanglement is more we have to understand in very deep because we are because it is very significant which is could lead into proxy of war it is to stop spread for the non-nuclear proliferation to the nuclear proliferation as both leader from China and Russia have stated that China-Russia relationship is better than ever in that respect I will bring to the Cold War era when US against Russia USSR and how Russia approaching developing country and as my region is Indo-Pacific region how Russia approaching and encourage like Indonesia and other country to oppose the United States and China who does the logistic providing the the military weapons and guns but now in my research that is the opposite China who did spread the through the belt silk belt all over the world and Russia approaching the small country in Indo-Pacific so could you turn this into a question I will just in one second but so then it is the the dangerous then as right now Russia is approaching like Indonesia to provide nuclear so therefore the concept of the entanglement into proxy of war and the non-nuclear proliferation to nuclear proliferation is become very significant so my question is do you think to navigate sign or Russian relationship it's it's significant and also to enforce entanglement and unilateral measures probably is the core in provide in a in a prevention for entanglement and proxy of war thank you that's true yeah just I mean just is an interesting thought just I mean very very briefly I think that I mean entanglement in a fairly narrow technical sense I mean there's this one can talk about entanglement in a much broader political sense as well in the narrow technical sense in which I mean it I'm less worried about what's going on between Russia and China I mean I think we would fight a war against Russia we would fight a war against China I don't really worry about us fighting joint Sino-Russian forces together though maybe I should worry about that so I think I think the dynamics that you're pointing to have have big political influence generally I think less so in the kind of more niche technical area that I'm looking at question there's someone yes George Beebe Center for the National Interest James I have a question about another area where lines are blurred and that's in the cyber realm the difference between a computer network attack and cyber espionage and the difficulty I think that that arises when you start talking about commitments not to engage in computer network attack against nuclear command and control is that it's difficult for assistance administrator to tell the difference in intention between an intrusion meant to collect information to understand the system espionage and an intrusion that's meant to damage or disable or disrupt computer network attack how would you handle that ambiguity so I it's a great point it is actually one that I address on the paper firstly I think if we agreed we would have to to make this kind of arrangement workable we would have to say no cyber interference with nuclear command and control where interference includes espionage as well as attack like I think trying to say we're allowed to do espionage we're not allowed to do attacks is unworkable for exactly the reason you say incidentally you know there are costs to this like you know if we don't plan to attack Russia and China with nuclear weapons in a conflict then being in our networks seeing that we don't plan to attack them could actually be a benefit so there are disadvantages to saying no espionage but I think if you could have a workable arrangement that there was no cyber interference let me rephrase that the only workable arrangement would be saying no cyber interference at all with nuclear command and control precisely because of what you say this there's something that I term in the paper double ambiguity which is again this thing you're pointing to here associated with cyber attacks against associated with cyber interference against dual use command and control assets if I launch a kinetic attack against dual use command the control asset that's ambiguous because of the dual use nature of the thing I'm attacking that form of ambiguity is still inherent with cyber interference with dual use assets but there's the other form of ambiguity as well that you say which is the possibility that espionage is mistaken for being an attack so cyber is kind of different from kinetic it's not impossible to launch cyber attacks that never get noticed by the victim so on the one hand you could alter huge cyber attack against your victim your victim mistakes it for a problem with their own system and it's not escalatory at all on the other hand you could do minor cyber espionage with no intention of attacking your adversary the adversary mistakes that has preparations from attack and kind of all hell breaks loose so there are interesting and important differences between the cyber realm and the kinetic realm and in a fast moving conflict the difficulty of distinguishing between espionage and attack I think is absolutely a critical difference between those two realms and he's a very very important observation Terry Terry Taylor from the International Council for Life Sciences thank you for your excellent remarks both of you actually for good questions as well as the points I find it I think one has to think about this issue from what you've said that entanglement is something we have to live with I don't think there's any technical or financial way we can get around that so having accepted that it seems I have a hard job thinking about the risks involved or mitigating the risks involved without making it dependent as you said earlier on a scenario or the context so aren't we left with that situation and whatever I don't know if you see what prospects there are for bureaucratic rearrangements for decision making I mean if there's something specific could say about that maybe this idea of space command does that increase the risk that we have in front of us at the moment or not isolating that perhaps may make it more risky I'm saying that off the top of my head so I would help us think about can we separate it from a specific context like for example a military confrontation in the South China Sea would be one example and there are many others your thoughts would be welcome thank you great question so firstly I would take slight issue with your premise I absolutely agree that entanglements is something we have to live with we do have choice over the degree of entanglements like we can choose to live in a more or less entangled world so there is some choice over the extent of the entanglements I'm not sure I want to I think I want the bureaucracy to consider this in a fairly context dependent way I mean you know the U.S. the DOD doesn't draw up generic war plans apart from the one against zombies but I think this kind of was a joke war plan that emerged a few years ago but you know there are here first the U.S. has a war plan against zombies so you know where's war plans for China there's war plans for Russia you know I think there is huge value to considering the very specific risks that would arise in those very specific contexts so actually in the case of like war planning or crisis management I don't have a I think it's good to think about these in a very context dependent way I think there is value to thinking about them at a more generic level and that's clearly relevant to like acquisition you know generally one is a choir on technology to be used in multiple different contexts not in any one specific context you know part of the reason I looked at Russia and China in the paper was to draw these more general lessons and you know I think we write about escalation risks in a generic way associated with technology the whole time and you know in terms of thinking about I think one can think fruitfully about how to make command and control more robust less likely to be degraded than the conventional conflict one can think about you know do hypersonic long-range weapons increase escalation risks or not so I think you know I think where necessary you can think about it in this more generic way on the space force all I have to say is this when I understand what the space force is supposed to do I will comment on whether or not I think it's a good idea from an escalation perspective I have literally no idea what the proposal is in terms of who does the acquisitions you know the forces are not in the chain of command the combatant commanders report directly to DoD so if air force so if a new sorry the combatant commanders report straight to the secretary so but the way it's talking it's as if like you know who's the combatant commander going to be in this I won't go into all the things that are not clear to me about this proposal but when the proposal is clear I'll feel qualified to comment on its implications for entanglement there's reason why it's not clear but you wrote his Nancy and Nancy Gallagher from University of Maryland you started off at the very beginning talking about or crediting John Steinbrunner with the concept of entanglement and what he was most concerned about dates way back before the current situation before he was even writing about it it was really I would say US thinking that we could use nuclear weapons to limited nuclear use to end a conventional war on favorable terms so this is I would see this as a variation of that same problem and what he was arguing for really was a fundamental change in security relationships and a fundamental change in you know reliance on deterrence on damage limitation on the whole business what you're talking about are some very minor modifications at the margins and so what I'm wondering while I think you've done a great service to point out all these escalation risks if you propose risk reduction teams are you essentially trying to delude the United States into thinking that it could manage these risks of the overall security posture that it has it's a great question I mean I would the idea is even goes even further back than John I mean I always like pointing out the aircraft that dropped the bomb on Hiroshima was a B-29 which was originally developed for conventional delivery and used for nuclear delivery so you know entanglement is literally as old as the nuclear age I mean at least if you go back to Hiroshima rather than Trinity but you're right I mean what I'm doing is unashamedly kind of fatalistic about the risk and working out how best we can manage it I you know one of the one of the very first things I did in the nuclear field was wrote with my boss George Perkovich this Adelphi paper about abolishing nuclear weapons and that was kind of an attempt to think about this big geopolitical engineering exercise you know how do we what are the conditions the requirements for creating a genuinely new security system but would allow us to do without nuclear weapons and all the terrible risks associated with them as politics have got worse and as the risks have kind of risen my focus has changed to what can we do right now to mitigate those risks because that's where I feel I personally can make the most impact that's not to say that I think it's not really important to have people thinking about these much more fundamental ways of reforming security relationships to build a much more fundamental durable lasting peace but it is to say that it is my assessment that you know right now my energy can be used even if it's just on the margins most effectively kind of just trying to reduce this risk a little bit so you know that's that's that that that is you know I wouldn't dispute your map your contention I think it is fair but fundamentally what I'm doing is proposing fiddling around the edges of a of a much bigger problem and not you know not claiming to be able to make huge progress on solving it that's ultimately an assessment of basically how messed I think how international politics are right now go ahead Neto yeah there's a risk there I mean I wouldn't I wouldn't deny that there's a risk there and then I think I and actually I mean I think I think it's a good lesson you know it's a good thought for me that as I present this thing you know it is worth and I rolled it out it is worth emphasizing that what I'm saying is risk misrestigation not mis-solution risk solution one shouldn't be complacent in this space so you know I mean I take that as a as a as a as a warning to remind people about you know how far my recommendations don't go I think that's a very fair point Nancy warning taken Jim Timby if I understood correctly Jim say say for people who don't know you oh I'm sorry I'm Jim Timby at the Hoover Institution formerly the State Department if I understood you correctly in response to one of Scott's questions you were discouraging on the idea of having a conversation with other countries about what these special assets are that we call nuclear command and control I would have thought that having said in the posture of you that we will consider a nuclear initiation if there's an attack on these special assets then there would be value in talking with other countries about what it is they should avoid attacking if not nuclear means without raising the prospect of our considering of nuclear response can you say a little bit more but I mean your the first answer was in the context of you know things are all very technically complicated but having made this statement in the posture of you which was not done on your advice but it's out there doesn't that color the value of having these conversations with other countries I must have expressed myself very badly then I am strongly in favor of discussions with the Russians and discussions with the Chinese to better understand what each of us considers our nuclear command and control system to be hopefully with the eventual goal of being able to agree for no non-nuclear attacks against nuclear command and control I am 100% supportive of those conversations what I was not supportive of would be the US making a unilateral commitment not to attack another state's nuclear command and control precisely because unless you have those discussions no one knows what you mean by nuclear command and control so I'm very supportive of those discussions in order to try to get consensus some kind of mutual not even consensus some kind of mutual understanding about what you know what we consider to be nuclear command and control and what isn't and that's a really difficult subject because it's all dual use so I must have explained myself very badly before I have one last question here in the front thank you my name is Mituo Nakai Regan Foundation regarding North Korean nuclear crisis how long does it take for them to date Nuke they say they nuke but I don't know how long that's gonna I think it's gonna take a long time sorry to develop nuclear weapons or to use nuclear weapons to de-nuclearize what do they mean when they say de-nuclearize the Korean Peninsula I mean I would say 10 to 15 years I mean if North Korea was actually serious about giving up their nuclear weapons and if it was just pure North Korean giving up its nuclear weapons for nothing in return and that's not what the North Koreans are asking for but to really give everything up to dismantle it all to ship all the material out the country for us to verify it all and convince ourselves their inspections of every suspect site if we did that in 10 to 15 years I would consider that pretty impressive it's not it's not happening in the president's first term I think we should call this to a close I just wanted to respond briefly to the thought that Nancy and James were just saying about the relationship between long-term aspirations and short-term fixes one thing I admire about your work is that you have done both and as you noted the work on what would mean to live in a disarmed world was important and I fully think it's appropriate then to turn to some other issues T.S. Eliot the British poet once wrote the people try to avoid the darkness outside and within by dreaming up systems so perfect that no one will need to be good and it's one thing to dream up these systems it's important to do that but it's also important to remind people how you have to be good and try to think about some mechanisms both political and technical that could help us do that so thank you thank you yes I do you can get the whole article yes on the IS International Security website and now their brief comment is outside