Table of Contents

Concerns about post-launch warhead ambiguity have tended to overshadow those associated with pre-launch ambiguity—even though the escalation risks are probably greater for the latter, once the likelihood and consequences of mischaracterization have been considered. Ideally, nuclear-armed states would reassess the cost-benefit trade-off of deploying ambiguous delivery systems to determine whether they should phase out some or all of them and revise acquisition plans. Realistically, however, states are highly unlikely to reduce their reliance on ambiguous weapons (though they may be somewhat more amenable to not increasing it).

Some of this resistance may stem from a desire to exploit warhead ambiguity to enhance deterrence. Of the three states highlighted in this report, China is the most likely to be doing so already, though the evidence is far from definitive (see box 3).1 That said, if Russia and the United States were to systematically reassess the benefits and risks of ambiguous weapons, one or both might decide that ambiguity was actually a feature and not a bug. In principle, this conclusion would not necessarily be wrong; it would depend on whether those states had accurately estimated the benefits and risks of ambiguity.

More prosaically, the financial benefits of ambiguous, particularly dual-use, delivery systems are dramatic. If a state decides that it needs to deliver nuclear and nonnuclear warheads in a similar way over similar distances, it can choose between developing one dual-use delivery system or two single-use systems.2 Opting for a dual-use system could significantly reduce research and development costs, as well as operating and maintenance costs. This economic reality makes dual-use systems extremely attractive to governments and militaries, which face fiscal pressures no matter how well-resourced.

Finally, in peacetime, when acquisition decisions are made, contingency planning occurs, and risk-reduction measures are implemented, inadvertent escalation risks—including, but not limited to, those associated with warhead ambiguity—are likely to be discounted for at least five reasons. These reasons stem from both the psychology of individual decisionmakers and the behavior of organizations, such as government bureaucracies.

First, decisionmakers may understate the chances of escalation because they are likely to base their probability estimates on “availability”—that is, “the ease with which relevant instances come to mind.”3 For example, because so few people have experienced a car accident in which they were seriously injured, 90 percent of drivers believe they are better than the median driver.4 Inadvertent escalation has thus far never led to nuclear weapon use, so decisionmakers could be similarly inclined to ignore the danger. But because crises—and especially conflicts—that could have culminated in nuclear weapon use have been extremely rare, it is wrong to simply extrapolate from the past and conclude that nuclear escalation is exceedingly unlikely in a future crisis or conflict.5

Second, even if decisionmakers do not underestimate the likelihood of inadvertent escalation, they may still dismiss it as being too unlikely to worry about, even though its consequences are potentially so enormous that the risk—the product of consequence and likelihood—should demand their attention. Research suggests that people tend to round small probabilities down to zero in decisionmaking.6 Thus, they are more likely to insure against a high-probability, low-consequence event than a low-probability, high-consequence event, even when the two events have equal risk but the high-consequence event would be much more damaging. Likewise, decisionmakers may be disinclined to spend resources on reducing the risk of inadvertent escalation by, for example, buying one nuclear and one conventional delivery system, instead of a single dual-use system.

Third, the risk of nuclear escalation is effectively unquantifiable. As a result, this risk is likely to be discounted or ignored within decisionmaking processes—such as military acquisitions—that are, or at least aspire to be, driven by quantitative cost-benefit analyses. There is no credible way to assign a dollar value to all the various potential consequences of a nuclear war, which could include immediate and delayed deaths on an enormous scale, as well as deep psychological harm to the survivors.7 In decisionmaking processes, costs that have not been quantified tend to get less attention. An overreliance on quantitative data has been observed in contexts as diverse as business, medicine, and the management of natural hazards, but it was first recognized—as its name, the McNamara Fallacy, may suggest—as a cause of the U.S. escalation during the Vietnam War.8

Fourth, efforts to reduce nuclear risks—unilateral efforts, in particular—often lack bureaucratically or politically powerful proponents who can help ensure that they receive due consideration. The U.S. system provides a useful example. Within the U.S. Department of Defense, risk-reduction efforts frequently find supporters, even among the department’s most senior personnel. However, such support is ad hoc and much less influential than that enjoyed by the more traditional defense functions of, say, acquisitions or contingency planning. Meanwhile, within Congress—which has a major influence on acquisitions in particular—there are rarely enough votes to block a new capability because of its escalation risks (the refusal to fund the Conventional Trident Modification was a rare exception). Risk-reduction efforts find their most natural home in the U.S. Department of State. Indeed, key State Department officials can effectively quarterback efforts to develop and, under the right political circumstances, negotiate cooperative security arrangements. However, they are much less able to effect unilateral changes to U.S. nuclear posture or planning—which become the most viable forms of risk reduction when international politics preclude cooperation.

Fifth, even though the potential dangers of nuclear escalation could become an overriding concern during a war, they are generally not apparent in peacetime and are therefore easy to ignore. Because governments have limited bandwidths, “problems whose consequences have not yet emerged are . . . at a disadvantage in the competition for elite cognitive investment.”9 Thus, most defense planning focuses on trying to prevent readily apparent dangers from metastasizing. The arguments for, say, improving U.S. capabilities to fight a war in the Western Pacific are regularly bolstered by Chinese military activities, such as weapon testing and exercises. By contrast, escalation dangers are essentially invisible on a day-to-day basis, and thus efforts to reduce them are likely to suffer from a lack of intellectual effort and high-level attention within overstretched bureaucracies.

Taken together, these factors greatly reduce the possibility of any state’s conducting a comprehensive reassessment of the role of ambiguous weapons in its military posture. That said, less ambitious risk-mitigation measures—both unilateral and cooperative—may be more feasible. Recommended actions for Washington to consider and adopt are presented below. This framing does not imply, however, that risk mitigation is solely Washington’s responsibility; rather, this charge should be shared with all nuclear-armed states that possess ambiguous weapons, particularly China and Russia. To this end, Beijing and Moscow should engage constructively with any good-faith proposals for cooperation that Washington may offer and develop their own unilateral risk-reduction measures in parallel. Chinese and Russian scholars will hopefully respond to this report by making their own proposals, tailored to their states’ political, military, and bureaucratic structures.

Exercise Restraint in Acquisitions

While existing categories of ambiguous weapons are here to stay, the United States may be somewhat more amenable to exercising restraint in introducing new ones. To this end, procedures should be established to ensure that escalation risks are factored into relevant acquisition decisions. Even if such risks end up being underweighted, as seems likely, giving them some attention should still promote better policy outcomes. The U.S. secretary of defense should, therefore, require relevant Department of Defense decisionmakers to consider any potential escalation risks resulting from warhead ambiguity when deciding whether to acquire new categories of ambiguous weapons. To this end, those decisionmakers should be presented with a formal assessment of such risks.10 Meanwhile, China, Russia, and other nuclear-armed states can and should create their own parallel requirements.

The United States should propose to China and Russia that they jointly agree not to acquire ambiguous intercontinental ballistic, cruise, or hypersonic boost-glide missiles.

Risk assessments should not be prepared by acquisition personnel (who generally lack the relevant expertise), but by a civilian-led team reporting to the under secretary of defense for policy. This team should draw on broad-based strategic and country-specific expertise, including from nondefense intelligence agencies. The first such assessment should be conducted for the planned nuclear-armed sea-launched cruise missile. Even though this weapon may not be dual-use itself, it will almost certainly be deployed on platforms—such as attack submarines or surface ships—that currently carry only conventional weapons.

Ambiguous intercontinental missiles, whether ballistic, cruise, or hypersonic boost-glide, are especially significant because the escalation risks associated with ambiguous weapons probably increase as their ranges do. Fortunately, no nuclear-armed state appears to have acquired such a weapon yet, largely because of the technical challenges to ensuring sufficient accuracy for a nonnuclear warhead to be effective. This barrier, however, is unlikely to hold for much longer. For this reason, the United States should propose to China and Russia that they jointly agree not to acquire ambiguous intercontinental ballistic, cruise, or hypersonic boost-glide missiles. In theory, the same outcome could be achieved through a series of parallel unilateral policy decisions, but a cooperative approach would allow for discussions about definitions and implementation and potentially even verification.

If a joint prohibition were agreed upon, China, Russia, and the United States would surely choose to keep all their ICBMs and SLBMs nuclear-armed. Only Russia has shown an interest in developing an intercontinental cruise missile, the nuclear-powered Burevestnik, on which there would be little point deploying a nonnuclear warhead. However, while Russia would likely choose to arm intercontinental boost-glide missiles with nuclear warheads, and China might do so, too, the United States would likely deploy such missiles for nonnuclear strikes.

This asymmetry might be mutually acceptable because of differing perceptions about the effectiveness of U.S. missile defenses. Russia’s development of boost-glide weapons has largely been driven by the possibility that U.S. missile defenses could undermine its ICBM and SLBM forces over time.11 Indeed, Russia’s intercontinental glider, Avangard, which Moscow claims has already been deployed, is probably exclusively nuclear-armed. Russia, therefore, might agree to forsake conventionally armed intercontinental gliders if the agreement also limited U.S. capabilities and mitigated the risk of nuclear war. And Beijing might follow suit, as it shares Moscow’s concerns about U.S. missile defenses. However, because China is less able than Russia to target the United States with conventional weapons, its incentives to place nonnuclear warheads on intercontinental gliders could be stronger. U.S. decisionmakers, by contrast, tacitly recognize that the United States is already vulnerable to a nuclear attack by China or Russia and that, as a matter of fact (if not of choice), the pursuit of invulnerability would be infeasible.12 Thus, they should be open to an agreement that limited the development of Chinese and Russian conventional capabilities, even if it left advances in their nuclear capabilities unchecked.

All that said, China and Russia are probably significantly more interested than Washington in eventually fielding ambiguous intercontinental missiles, particularly hypersonic boost-glide weapons.13 Therefore, they might view a U.S. proposal to refrain from acquiring such missiles as an attempt to secure a unilateral advantage for the United States. One way to address this problem would be to include the prohibition in a package with another measure of more interest to China and Russia. Beijing and Moscow could usefully consider what this measure might be.

Be Transparent About Capabilities

Nuclear-armed states that have acquired ambiguous weapons to reduce costs—rather than to enhance deterrence—may be amenable to implementing transparency measures that could modestly mitigate warhead ambiguity. Incorrect or uncertain design assessments are one potential cause of incorrect or uncertain characterization. Currently, China and Russia are only sometimes transparent about what type or types of warheads their weapon systems are designed to carry. The United States is generally more transparent. But the information Washington provides is usually intended to meet domestic needs (such as budget justifications for Congress), and Moscow’s and Beijing’s level of confidence in it is unclear. For example, Russian officials and experts have expressed concerns that future U.S. boost-glide weapons will be dual-use in spite of U.S. assertions that they will be exclusively nonnuclear.14 To address this problem, the United States should propose to China and Russia that they declare, publicly or privately, each type of missile and aircraft they deploy as nuclear-armed, conventionally armed, or dual-use. Each participant should provide updates when weapons are modified and new types are acquired. Declarations would be most effective if accompanied by intergovernmental discussions to build confidence in their veracity. But, even alone, they should still have value.

More ambitiously, the United States should propose to China and Russia that they privately discuss any observable differences in design or deployment patterns between their nuclear- and conventionally armed ambiguous weapons. Even weapons that are physically identical may be deployed or operated in somewhat different ways. For example, Beijing and Moscow could indicate to Washington whether their nuclear-armed mobile missiles are typically accompanied by different support vehicles than their nonnuclear missiles. Meanwhile Washington could explain whether, say, its nuclear- and conventionally armed aircraft typically have different operational profiles. This proposal would be more feasible if China and Russia were more confident in the survivability of their nuclear forces. Even if they did not acquire ambiguous weapons to enhance force survivability, they may worry that highlighting how their nuclear and nonnuclear delivery systems could be distinguished would make it easier for the United States to target the former. Therefore, discussions on observable differences may need to occur in the context of arms control designed to reduce perceived threats to nuclear forces.

Improve Operational Planning

The most promising area for risk reduction is probably military planning and crisis and conflict management. Most fundamentally, the U.S. Department of Defense and relevant combatant commands should plan for crises and conflicts on the assumption that each participant might mischaracterize or be unable to characterize the other’s ambiguous weapons. Starting from this premise, to reduce the risks of inadvertent escalation, defense planners should consider how to adjust the United States’ contingency plans and its approach to managing crises and conflicts. Identifying potential adjustments will require extensive analysis, particularly from personnel with access to relevant classified information. In undertaking this task, planners should consider the following four general principles.

First, and most importantly, the U.S. secretary of defense should require relevant decisionmakers to consider any potential escalation risks resulting from warhead ambiguity when deciding whether to authorize strikes with or against ambiguous delivery systems. To this end, those decisionmakers should be presented with a formal assessment of such risks. A team that includes civilian strategists and intelligence personnel with expertise on the adversary should prepare the assessment (which mirrors those already conducted to evaluate the legality of various kinds of operations). Critically, if the potential operation involves attacks against ambiguous assets, the intelligence analysts responsible for assessing how those weapons are armed should include a clear statement indicating their level of confidence in their findings.

Second, U.S. military planners and decisionmakers should be aware of the trade-offs associated with the use of ambiguous delivery systems for signaling operations. Ideally, the United States would refrain entirely from signaling with ambiguous assets. But its actual nuclear force structure makes a blanket recommendation imprudent. Dual-use aircraft are almost certainly the United States’ preferred delivery system for signaling operations. SSBNs, which are reserved for nuclear operations, are the most plausible alternative.15 Using them for signaling would reduce the risk of misinterpretation because of warhead ambiguity.16 However, the survivability of the SSBNs might be compromised. Moreover, China and Russia might worry about submarines’ being positioned close to their coasts to launch attacks on their nuclear forces with little warning. For this reason, a signal sent with SSBNs would still carry escalation risks, albeit different ones from a signal sent with dual-use aircraft. In practice, therefore, it is not possible to provide U.S. officials with stronger advice than to be aware of and weigh the risks associated with each signaling option.17

Third, if the United States uses ambiguous delivery systems for nuclear signaling, it should take steps to mitigate the risks associated with warhead ambiguity. For example, it could hold some ambiguous aircraft back from a conventional war and reserve them exclusively for nuclear operations, should such operations become necessary. Alternatively or additionally, it could accompany signaling operations involving ambiguous weapons with verbal threats that referenced the units involved and explicitly indicated that they were nuclear-armed. This approach could be described as messaging and might reduce the need for tacit understanding between the United States and its adversaries about how to interpret signals. Neither of these options would be foolproof, but each should help reduce the risk of inadvertent escalation.

Fourth, the United States should offer verbal assurances to reduce the likelihood of false positives resulting from operations involving conventionally armed ambiguous weapons. In a crisis or conflict, just as Washington might issue threats to clarify the meaning of nuclear signals, it could offer appropriate assurances that ambiguous weapons were conventionally armed. Given the need for operational secrecy, the U.S. military might be reluctant to describe the weapons involved in an ongoing operation. But even very general statements could prove useful. For example, if no U.S. aircraft were loaded with nuclear weapons, the United States could say so and periodically repeat its assurance, unless and until such loading began.

It would be understandable if busy military and defense officials balked at this long, complex to-do list. What should motivate them is the likelihood that they will have to manage the effects of warhead ambiguity if a conflict or crisis occurs. In past crises, and even in peacetime, ambiguous weapons have frequently been mischaracterized or uncertainly characterized. Now, given militaries’ increasing reliance on ambiguous weapons, the difficulty of characterizing an adversary’s weapons, particularly in the fog of war, is growing.

Even among the officials who agree that the danger is real, one view may be that no immediate action is required because it would be possible, in a crisis or conflict, to manage the risks on the fly. By the time one state had mischaracterized its adversary’s weapons, however, the prospects for managing any resultant escalation would have diminished. After all, the adversary would likely not even know about the mistake, precluding the possibility of its even trying to correct the misapprehension. The right time to start thinking about how to mitigate the risks of pre-launch warhead ambiguity is, therefore, long before the onset of a crisis or conflict—it is today.


1 Others may also be trying to use warhead ambiguity to their advantage, potentially for somewhat different reasons. North Korea has a large force of regional ballistic missiles, many of which may be dual-use (see appendix). However, because of limitations in fissile material, probably only a small fraction of them could be armed with nuclear warheads. Unless the United States could quickly and reliably identify which of these missiles were nuclear-armed, this arrangement would likely make it more difficult for Washington to eliminate Pyongyang’s nuclear arsenal in a crisis or conflict—but whether this kind of shell game is a happy accident, from the North Korean perspective, or the result of a deliberate strategy is unclear.

2 Other pathways for acquiring dual-use delivery systems are possible. For example, a state could acquire a delivery system for one purpose and then later make it dual-use.

3 Amos Tversky and Daniel Kahneman, “Availability: A Heuristic for Judging Frequency and Probability,” Cognitive Psychology 5, no. 2 (1973): 207–232. The use of the availability heuristic seems likely here because there is no obvious “initial value” that can be adjusted (unlike, say, the task of estimating the value of a stock in a year’s time, in which case its current value provides the obvious starting point).

4 Colin F. Camerer and Howard Kunreuther, “Decision Processes for Low Probability Events: Policy Implications,” Journal of Policy Analysis and Management 8, no. 4 (1989): 569.

5 To compound this problem, if decisionmakers are unaware of the risks associated with pre-launch warhead ambiguity—a danger that has attracted little attention, at least in open sources—they will be unable to compensate for the lack of historical instances of inadvertent escalation with imagined examples.

6 Paul Slovic, Baruch Fischhoff, Sarah Lichtenstein, Bernard Corrigan, and Barbara Combs, “Preference for Insuring Against Probable Small Losses: Insurance Implications,” Journal of Risk and Insurance 44, no. 2 (June 1977): 237–258. Similar behavior has also been observed among corporate managers. March and Shapira, “Managerial Perspectives on Risk and Risk Taking,” 1411.

7 For a general discussion of the challenges associated with implementing cost-benefit analyses, see Baruch Fischhoff, Paul Slovic, and Sarah Lichtenstein, “Weighing the Risks: Which Risks Are Acceptable?” in The Perception of Risk, 122–125.

8 S. O’Mahony, “Medicine and the McNamara Fallacy,” Journal of the Royal College of Physicians of Edinburgh 47, no. 3 (September 2017): 281–287; Kenneth Cukier and Viktor Mayer-Schönberger, “The Dictatorship of Data,” MIT Technology Review (May 31, 2013),; and Paul Slovic, Howard Kunreuther, and Gilbert F. White, “Decision Processes, Rationality and Adjustment to Natural Hazards” in The Perception of Risk, 20. The origin of this term is unclear, but the earliest exposition of the concept appears to be “Adam Smith” (pseudonym), “The Last Days of Cowboy Capitalism,” Atlantic Monthly (September 1972): 54.

9 Alan M. Jacobs, “Policy Making for the Long Term in Advanced Democracies,” Annual Review of Political Science 19 (2016): 439.

10 Ideally, a requirement to consider escalation risks should apply to all acquisition decisions, but it is particularly important in this case.

11 Putin, “Presidential Address to the Federal Assembly.”

12 U.S. Department of Defense, Missile Defense Review, 2019, 31,

13 Unlike the United States, both China and Russia field shorter-range dual-use missiles. Moreover, there is evidence of Chinese and particularly Russian interest in conventional intercontinental boost-glide systems. James M. Acton, “Russia and Strategic Conventional Weapons: Concerns and Responses,” Nonproliferation Review 22, no. 2 (2015): 147–148; and Mark Stokes, China’s Evolving Conventional Strategic Strike Capability: The Anti-Ship Ballistic Missile Challenge to U.S. Maritime Operations in the Western Pacific and Beyond (Project 2049 Institute, September 14, 2009), 10, Given that both states are concerned about the potential impact of U.S. ballistic missile defenses on their nuclear deterrents, they have clear incentives to develop dual-use intercontinental boost-glide systems eventually.

14 For example, Anatoly Antonov, “Russia Forced to Develop Global Prompt Strike Weapons,” Security Index 19, no. 3 (2013): 6–7. See also, Acton, “Russia and Strategic Conventional Weapons,” 146.

15 SSBNs have occasionally been used for signaling in peacetime by conducting publicized port calls. Because the ICBM force is kept on alert, there is much less flexibility in how it is postured, making it significantly less useful for signaling.

16 The risk would not be zero, however, as SSBNs might be mistaken for SSGNs, which are SSBNs that have been modified to carry nonnuclear cruise missiles instead of nuclear-armed ballistic missiles.

17 Chinese and Russian decisionmakers face a similar dilemma. Their states’ only missiles that are reserved exclusively for nuclear operations have intercontinental ranges (though, Russian capabilities, in particular, are very murky). As a result, in issuing nuclear signals, Beijing and Moscow would most likely choose between using ambiguous shorter-range missiles (or perhaps ambiguous aircraft) and mobile ICBMs that carry only nuclear warheads. Once again, a trade-off between different types of risk would arise. Signals sent with mobile ICBMs would carry less risk of being misinterpreted but, because of the potential reach of ICBMs, could be more aggressive than desired.