Editor's Note

Matthew Bunn, the James R. Schlesinger Professor of the Practice of Energy, National Security, and Foreign Policy at the Harvard Kennedy School, emphasizes that mere statements are insufficient for addressing the North Korean threat. Instead, he contends that carefully calibrated actions are essential to effectively deter North Korea without triggering provocation. In order to maintain peace and prosperity on the Korean Peninsula, Bunn argues that the United States must convey two clear messages to North Korea: (1) U.S. will retaliate if North Korea initiates an attack, but (2) it will not use the offensive capabilities if North Korea does not challenge the status quo.

The problem of deterrence on the Korean peninsula is complicated by a difficult “security dilemma.” Steps that appear to the Republic of Korea (ROK) and the United States as necessary defensive actions intended only for deterrence may appear to North Korea as offensive threats to the survival of the North Korean state, provoking North Korean actions that further undermine ROK and U.S. security.


In peacetime, every action to counter North Korea’s deterrent forces – whether to target them “left of launch” or to defend against them – may provoke additional North Korean efforts to ensure its forces can survive attack and penetrate defenses, expanding the threat. In crisis or conflict, any action that made North Korea believe a full invasion or a decapitation attack was coming, or had already begun, could lead the North to escalate – perhaps even to use nuclear weapons – to stop an attack or as a final act of revenge.


Moreover, Chinese, Japanese, and to a lesser extent Russian security are also intimately related to what happens on the Korean peninsula. What the United States and the ROK do to deter and to defend against North Korea – and what North Korea does with its own forces – affects them as well, and they are likely to react to defend their interests.


Therefore, to maximize ROK and U.S. security, every purchase of a new weapon systern, every military plan, and every military action in a crisis or conflict must be judged both by its contribution to defense and deterrence and by what effect it may have on provoking additional threatening actions by North Korea and by others. Fortunately, the ROK/U.S. Combined Forces Command has long provided a foundation for both partners in the ROK-U.S. alliance to make plans and consider such issues jointly, and the new forum included in the Washington Declaration should make it possible to extend that close joint consideration to nuclear planning as well.


This problem of deterring without provoking is not unique to the Korean peninsula. There are many things the United States and NATO could do that would help Ukraine’s effort to defend against Russian attack – from providing long-range missiles to enforcing a no-fly zone to sending their own troops to fight – that they are not doing, because they judge those actions would be too likely to provoke various forms of Russian escalation. There are many things Russia could do that would help its war against Ukraine – such as attacking weapons supplies for Ukraine while they are still on NATO soil – that it does not do because they would be too likely to provoke a NATO and U.S. response that would endanger Russian security. Similarly, there are many things the United States could do to help Taiwan that it does not do because it understands they would provoke Chinese reactions that would undermine, not contribute to, the security of Taiwan and of the United States. Avoiding undue provocation is a fundamental part of successful deterrence.


This issue of deterring without provoking creates very difficult policy dilemmas. For example, the ROK and the United States clearly want to maintain their conventional military edge over North Korea, to deter and defeat any invasion of the South. But the record of Cold War crises makes clear that if one side is conventionally inferior in the area where the crisis is taking place, they are more likely to consider the use of nuclear weapons, as they have few non-nuclear options – thereby posing a serious danger to the conventionally superior side. As another example, the ROK and the United States have every incentive to try to maintain the ability to target North Korea’s missile forces – but that effort gives North Korea incentives to expand and diversify those forces even more, to try to ensure that they can survive and penetrate.


Deterrence of major wars and smaller provocations



Avoiding major war on the Korean peninsula – particularly nuclear war – should be the preeminent focus of ROK security policy, and of U.S. extended deterrence policy to help defend the ROK. Any major war on the Korean peninsula would be absolutely devastating to the ROK – and would have impacts that would reverberate around the world (including the possibility that the conflict would lead to nuclear devastation in the ROK, in North Korea, in regional locations such as Japan and Guam, and in the United States).


Fortunately, what might be called “core” deterrence is likely to be very effective: North Korea understands that getting into a full-scale war with the United States and the ROK would be a terrible disaster for its regime, and the ROK and the United States also understand the awful destruction and loss of life a war would cause. More North Korean missile tests and nuclear weapons production will not change those realities.


Many therefore argue that deterring major war is a less immediate problem than deterring smaller-scale North Korean assaults, such as the sinking of the Cheonan or the shelling of Yeonpyeong Island. The ROK and the United States appear to have been successful at convincing North Korea not to carry out actions at that level of violence in recent years – though whether that will continue is, of course, uncertain.


The problem is that small-scale assaults and the risk of full-scale war are intimately linked. The responses that the U.S.-ROK alliance might make to a North Korean provocation, and the responses North Korea might take in turn, could escalate to full-scale war. Even where full-scale war is in no one’s interest, managing escalation can be quite difficult.


The difficulties of managing crises and conflicts with nuclear-armed states



Deterring North Korean outrages, and responding to them if they occur, without provoking North Korea to escalate to higher levels of violence, is not easy. It does not depend only on the strength of ROK and U.S. forces or our willingness to use them. Given the many challenges to human and governmental decision-making amid the anger, fear, and time pressure of a major crisis; the particular peculiarities of decision-making in personalist dictatorships like North Korea; the accidents and unintended actions that often occur in crises; North Korea’s constant misperception and paranoia about U.S. and ROK actions and intentions; and the limited understanding in the ROK and the United States of what the North Korean regime is thinking in the heat of the moment, managing escalation from a low-level conflict to higher levels is difficult and involves enormous risk.


Indeed, the overwhelming lesson of the U.S.-Soviet crises of the Cold War is that in crises with nuclear-armed states, things can get out of control. During the Cuban Missile Crisis, U.S. naval forces were dropping “practice depth charges” on Soviet submarines to force them to the surface – not knowing that those subs were nuclear-armed and had full capability to use their nuclear weapons if attacked; Soviet forces in Cuba shot down a U.S. U-2 aircraft against Khrushchev’s direct orders, leading President John F. Kennedy’s team to believe the Soviet leadership was intentionally escalating the crisis; a U.S. U-2 blundered into Soviet airspace accidentally; a hastily rearranged U.S. early warning radar received a false alarm of an attack underway; and more. As President John F. Kennedy pithily put it during the crisis: “There’s always some son-of-a-bitch who doesn’t get the word.”


With events occurring that no leader intended, and misperception rampant on all sides, the problem of avoiding provocations that lead to escalation is a difficult one. No one should believe that they can send careful, “tailored” deterrent “messages” to the other side that will be interpreted in the way the sender meant them.


Further complications: other parties and more



The situation is made even more complicated by other elements of the overall structure in which it is embedded: other parties in the region, other national interests in addition to military security, domestic politics in each of the relevant countries, and the dynamics of the ROK-U.S. alliance.


First, actions the ROK and the United States take together for deterrence and defense, and North Korean actions, will also affect, and be affected by, China, Japan, and to a lesser extent Russia. China’s perception of ROK and U.S. positions, for example, will affect its willingness to put pressure on North Korea to restrain its nuclear activities. China may come to perceive ROK missiles as threats to China, not just to North Korea; if there are ever talks over new accords to limit intermediate-range missiles in Asia, China or Russia may insist that ROK missiles be included.


Moreover, the ROK has to think not only about its military security but is broader national interests. China’s economic pressure after the deployment of THAAD and the collapse of the investments in Kaesong or in preparing to build reactors under the Agreed Framework are obvious examples of these linkages to economics and other issues. The value of the Seoul Olympics to achieving a period of reduced North-South tensions also makes clear that many things beyond strictly military issues affect the security situation on the peninsula – so that planning military activities to maximize national well-being requires taking broader considerations into account as well.


Meanwhile, domestic politics in the ROK, in North Korea, in the United States, and elsewhere, inevitably has its effect on decisions, particularly in peacetime but also in crisis. During the Cuban Missile Crisis, President Kennedy remarked to his brother that if he had not acted to address the Soviet missiles, he would have been impeached. The internal dynamics each leader faces – whether in the dictatorship of North Korea or in the polarized democracies of the ROK and the United States – will inevitably have their impact on escalation decisions.


Alliance dynamics between the ROK and the United States will also affect decisions, in peacetime and in crisis. As with any other alliance, the national interests of the allies overlap substantially but are not identical. The ROK wants the United States to be tough in standing up for ROK interests – but not so tough that it pulls the ROK into a war it never wanted. The United States, similarly, wants the ROK not to be too “soft” on North Korea, but not to respond to North Korean provocations in ways that might lead to open war. Many in the ROK worried U.S. policy was too warlike during the “fire and fury” crisis of 2017, for example, and many worried the United States might give away too much when President Trump turned to summitry and said he had fallen “in love” with Kim Jong Un.


Deterrence and reassurance



As Thomas Schelling noted, “To say, ‘One more step and I shoot,’ can be a deterrent threat only if accompanied by the implicit assurance, ‘And if you stop I won’t’’’ (Schelling 1966, 74).[1] For the fear of retaliation to stop the adversary from attacking, the adversary has to believe that the “retaliation” will not occur if they do not attack. Hence, reassurance of the adversary is a fundamental part of deterrence.


This is not as easy to accomplish as it sounds. Robust U.S. and ROK conventional military forces, designed to deter North Korea and defend against it if necessary, can look to North Korea like forces capable of destroying the North Korean regime – and in their paranoia, they might conclude, in a moment of crisis, that the ROK and the United States are prepared to undertake such a regime-change conflict. Large-scale military exercises, seen by both the U.S. and ROK militaries as both deterrent signals and efforts that are necessary to be ready to fight if needed, can be seen in the North as preparation for war. (After all, Russia used large-scale exercises as a cover for gathering its troops for the invasion of Ukraine, assuring the world it had no intention of invading Ukraine.) The U.S. nuclear threat – and the ongoing U.S. and ROK effort to maintain the capability to destroy and defend against as many of the North Korean missiles as possible and to threaten leadership targets in North Korea – may add to North Korea’s perception of a threat to the very existence of its regime.


Statements alone probably will not help very much in solving this reassurance problem. North Korea is unlikely to believe messages sent by either the United States or the ROK that indicate we are not planning large-scale attacks or trying to overthrow the North Korean regime – especially since, as our countries are democracies, there is always a range of voices, some sending more hawkish signals than others.


Instead, reassurance is likely to have to rest on behavior – acting, particularly with military forces, in ways that would be very difficult to misinterpret as threats of invasion or destruction of the North Korean leadership. Actions to respond to a North Korean provocation, such as missile strikes or air strikes on North Korean targets, may be needed – but they may also fuel North Korean paranoia. Each action must be carefully calibrated in the effort to provide an adequate response without provoking North Korea to escalate the crisis to higher levels of violence.


Imagining the situation from North Korea’s perspective



To understand the risk of provocation, it is helpful to try to look at things as you might if you were a North Korean leader. One always wants to avoid mirror-imaging, but still, it is helpful to imagine what you might think and how you might react if you were in their situation. This is not an argument that U.S. or ROK officials should agree with North Korean perspectives, but only that understanding the adversary and how the world looks through their eyes – a certain amount of “strategic empathy” – is important to achieving deterrence with minimum risk.


If you were a North Korean leader, you would be presiding over a desperately poor country. Your southern neighbor would be far richer and more technologically sophisticated and have far more international support. You would be in a state of constant hostility with the most powerful country on earth – the United States – and with its ally the ROK. Your conventional forces would be no match for the combined might of the ROK and the United States, though, with thousands of rockets and artillery tubes within range of Seoul, you would be confident, even without nuclear weapons, that you could do some awful damage to them before they could destroy you. Nevertheless, if you were unable to deter them from attacking and a full-scale war broke out, unless you used your nuclear weapons there is little doubt your forces would be utterly defeated and your regime destroyed.


And so, you might find yourself relying very heavily on your nuclear weapons. But you would see that the United States and the ROK were constantly trying to improve their ability to find and destroy your weapons before they were launched and defend against them after they were launched. Hence, you might invest a good deal in more weapons, more survivable weapons, and weapons better able to get around defenses. Moreover, you would likely worry about ensuring that retaliation would occur even if you were found and killed. That might lead you to think about making sure that if you were gone, others would still have the ability to fire the nuclear weapons.


All of this is at least somewhat similar to what North Korea is actually doing. This kind of putting yourself in the enemy’s place does not by any means explain everything – North Korea remains an extremely odd state in countless ways – but it does help a bit with judging what kinds of actions by the United States and the ROK might provoke unfortunate North Korean responses.


Scenarios to consider



To consider these issues of deterrence, provocation, and reassurance more closely, it is worthwhile to think through some particular scenarios.


Vipin Narang, now serving in the United States as a deputy assistant secretary of defense overseeing space and nuclear policy issues, has warned that in some circumstances, it would make sense for North Korea to use nuclear weapons first in a conflict (Narang 2017). Narang envisioned a situation in which a crisis escalated, leading North Korea to believe that the United States and the ROK were about to launch a large-scale assault designed to overthrow the North Korean regime. In that circumstance, he argued that North Korea might use a small number of nuclear weapons against Andersen Air Force Base in Guam and bases in the ROK and Japan to destroy much of the American and ROK ability to pursue the conventional fight. North Korea might then hold its long-range nuclear weapons in reserve, seeking to deter the United States from striking back. In Narang’s scenario, the North would be hoping to block an invasion and get a ceasefire that would leave Kim in power. As Narang puts it, “Kim may surmise that if he doesn’t use nuclear weapons first, he is certain to lose; if he does, he may have a fighting chance of surviving.” In this scenario, the key factor that leads to nuclear use is U.S. and ROK action that leads North Korea to believe that a full-scale assault is coming – it is the alliance’s actions that inadvertently provoked the North Korean nuclear attack.


Jeffrey Lewis, a professor at the Middlebury Institute of International Studies at Monterey, has envisioned a scenario in which actions responding to accidents, combined with misinterpretations, lead to disaster (Lewis 2018). North Korean air defense units mistake an ROK airliner for a military aircraft after it mistakenly drifts into North Korean airspace without its transponder on (as a result of a temporary loss of power in the cockpit). The ROK President, without U.S. agreement, retaliates with six conventional missiles, targeted at the headquarters of the Korean People’s Air and Anti-Air Force and a compound used by Kim Jong Un and his family. Kim, with communications out, misinterprets a tweet from the U.S. President as indicating a decapitation attack is underway. As in Narang’s scenario, he uses nuclear weapons against a variety of targets in the ROK and Japan to try to stop further attacks. The United States launches a conventional-only counterattack but does not succeed in destroying all of North Korea’s missiles, and North Korea fires ICBMs at the United States that destroy several American cities. In this scenario, too, without intending to do so, the ROK takes action that leads to a North Korean misperception that provokes them to use nuclear weapons.


Public descriptions of current U.S. and ROK military plans suggest that if these plans were implemented they could lead to something like Narang’s scenario. Imagine that North Korea launches yet another provocation – shelling, sinking a ship, shooting down an aircraft, or something else. The ROK and the United States jointly conclude they need to retaliate to punish North Korea and reestablish deterrence, and they launch modest strikes into North Korea, destroying a few non-essential military targets in the North. North Korea sees the U.S.-ROK action as an escalation and feels that it has to respond; it fires several conventional missiles at U.S. bases in the ROK, primarily as a warning. The U.S. and the ROK, seeing that North Korea is beginning to use its missiles, launch a “Kill Chain” effort to destroy the North’s remaining missile forces. This has two effects. First, it puts North Korea under pressure to “use them or lose them” – if it does not fire its nuclear-armed missiles, they may be destroyed before it can do so. Second, it creates a fear that this campaign – which would probably include attacks on North Korean air defenses, to protect the pilots carrying out the anti-missile campaign – is really intended to prepare the way for a full-scale invasion. To stop such an invasion, North Korea uses a handful of nuclear weapons against key U.S. and ROK bases and ports, seriously damaging U.S. and ROK warfighting capabilities on the peninsula, while holding its remaining nuclear weapons in reserve to deter the United States from striking back.


All of these are situations where the North initially uses nuclear weapons believing it can deter the United States and the ROK from an overwhelming response. To deter North Korea from considering such an option, the U.S. nuclear posture review states that “Any nuclear attack by North Korea against the United States or its Allies and partners is unacceptable and will result in the end of that regime. There is no scenario in which the Kim regime could employ nuclear weapons and survive” (U.S. Department of Defense 2022, 12).


But will North Korea believe that the United States would not be deterred from carrying out regime-ending attacks? It is worth making enormous efforts to avoid ever getting into a conflict so intense that such considerations would arise – and in particular, to find ways to deter without provoking.


Steps to mitigate the dilemmas



The dilemmas of deterring without provoking cannot be fully resolved. In part, they are simply inherent in the realities of deterrence on the Korean peninsula (and, often, elsewhere). But several steps could be taken to reduce the risk of provoking North Korean responses that undermine ROK and U.S. security.


Reducing the temperature. All of these issues are made worse by intense North-South hostility. Reduced hostility, if it could again be achieved, and could then be sustained, would reduce the chance of crises, the chance that crises would escalate to conflict, and the risk that conflict would escalate to the use of nuclear weapons. In general, while past agreements with the North have ultimately failed, the periods when talks were underway and North Korea believed it had a chance to get things it wanted from those talks saw fewer North Korean provocations and slower North Korean buildups than other periods. The ROK and the United States should work together to examine past successes and failures in reducing the temperature with North Korea to draw joint lessons learned about what works and what does not. They should seek to make offers to North Korea that would make it in the North’s national interest to break the current deadlock and return to talks. In particular, they should explore the circumstances in which they would be willing to declare an end to the Korean War and begin taking the first steps toward diplomatic normalization. Vipin Narang is among those who have argued that having U.S. diplomats in Pyongyang would help reassure the North that the United States was not about to attack (Narang and Panda 2020).


Including provocation risk in planning. The ROK and the United States, as they plan defense investments, make military plans, and consider military actions in a crisis, should always ask: “How will North Korea and others in the region respond to this?” Indeed, the ROK and the United States should institutionalize this consideration, creating a group whose job is to think through possible North Korean responses to major decisions and include those possibilities in the decision-making process.


Reassurance in peacetime. As noted earlier, North Korea is not likely to believe reassuring statements from the ROK and the United States. Nevertheless, it would be worthwhile for the ROK and the United States to reiterate frequently that they commit never to attack North Korea unless North Korea attacks them first, and that they will not try to overthrow the North Korean regime unless North Korea launches a full-scale war against South Korea or uses nuclear weapons. Acting in ways consistent with those assurances – for example, designing exercises that are clearly more appropriate for defense than for offense, not conducting exercises exploring options for decapitation attacks – would help increase the credibility of such statements.


Conventional confidence-building measures (CBMs). It is worth considering whether some of the measures the ROK and North Korea have agreed to in the past to reduce tensions along the DMZ could be renewed – and what further CBMs might be possible if relations improved. Narang, for example, is among those who have argued for ending flights of nuclear-capable bombers toward North Korea (Narang and Panda 2020). Military-to-military contacts should be expanded and communication lines established and exercised. In the longer term, might it be possible to have reciprocal notification and observation of major military exercises?


Nuclear restraints. The ROK and the United States should continue seeking talks with North Korea on nuclear restraints. Rather than simply waiting for the North to agree to talks, they should develop proposals designed to convince North Korea there is enough that it might gain that it is in its national interests to resume talks. While keeping full denuclearization as an eventual objective, the ROK and the United States should recognize that objective will not be achieved in the near term, and should focus instead on nuclear restraints that could reduce risks along the way.


For many years, various forms of freeze – on testing, on production at Yongbyon, or on broader production if verification issues could be resolved – have been the main proposal for North Korean nuclear restraint. It should be recognized, though, that with the level of capability North Korea has already reached, a freeze today would be worth much less than a freeze a decade or two ago. Other measures should also be explored, despite the verification challenges they pose, such as reductions in North Korea’s ICBM force, an end to tritium production, or a ban on short-range tactical nuclear weapons. Discussions of nuclear doctrine, the dangers of predelegation, nuclear weapons safety and security, and other matters affecting nuclear risk should be pursued.





Otto von Bismarck is said to have remarked that it is very dangerous to play chess one move at a time. You have to think hard about the move the adversary will make in response. To maximize their security, the ROK and the United States need to take that lesson to heart as they play a game with North Korea and others that is far more complex than chess. With adequate consideration of the risk of provocation, and actions based on that understanding of risk, the ROK and the United States have an excellent chance of continuing to enjoy peace and prosperity on the Korean peninsula for decades to come. ■




Allison, Graham T. and Philip Zelikow. 1999. Essence of Decision: Explaining the Cuban Missile Crisis. New York: Pearson.


Lewis, Jeffrey. 2018. The 2020 Commission Report on the North Korean Nuclear Attacks Against the U.S. New York: Mariner Books.


Narang, Vipin. 2017. “Why Kim Jong Un Wouldn’t Be Irrational to Use Nuclear Weapons First.” Washington Post, September 8.


Narang, Vipin and Ankit Panda. 2020. “North Korea: Risks of Escalation.” In Survival 62, 1: 47-54.


Sagan, Scott. 1995. The Limits of Safety: Organizations, Accidents, and Nuclear Weapons. Princeton: N.J.: Princeton University Press.


Schelling, Thomas. 1966. Arms and Influence. Yale University Press.


Sherwin, Martin. 2020. Gambling with Armageddon: Nuclear Roulette from Hiroshima to the Cuban Missile Crisis. New York: Vintage Books.


U.S. Department of Defense. 2022. “Nuclear Posture Review.”


[1] I am grateful to Reid Pauly for pointing out this Schelling quote to me. For an extended discussion, mostly in the context of coercion rather than deterrence, see Reid Pauly, The Assurance Dilemma in Coercive International Politics, forthcoming.



Matthew Bunn is the James R. Schlesinger Professor of the Practice of Energy, National Security, and Foreign Policy at the Harvard University’s John F. Kennedy School of Government.



Copyedited by Jisoo Park, Research Associate
    For Inquiries: 02 2277 1683 (ext. 208) | jspark@eai.or.kr

Major Project

Center for National Security Studies

Detailed Business

U.S.- China Strategic Competition


Related Publications