Table of Contents

Mischaracterization or uncertainty could be an intentional or unintentional consequence of warhead ambiguity. If a state wanted an adversary to mischaracterize certain ambiguous weapons, it could try to create misinformation by, for example, issuing misleading communications through either public statements or private messages intended to be intercepted. Alternatively, it could engage in deceptive operational practices. For example, a state could accompany conventionally armed delivery vehicles with communication or security equipment typically associated with nuclear-armed systems. Such misinformation techniques would not be guaranteed to succeed, of course. But if the adversary found it difficult to distinguish between nuclear and nonnuclear weapons in any case, adding misinformation would make characterization even more difficult. For this reason, this chapter focuses on the plausibility of mischaracterization or uncertainty occurring unintentionally.

If analysts are unable to draw firm conclusions, national and military leaders may assume, for various reasons, that ambiguous weapons are nuclear-armed— a process that sometimes yields false positives.

Incorrect or uncertain conclusions about how a weapon is armed can arise unintentionally in at least two ways. First, while collecting and analyzing intelligence, analysts may misinterpret the inevitably imperfect information that is available to them—a process that is probably equally likely to generate false positives and false negatives. Second, if analysts are unable to draw firm conclusions, national and military leaders may assume, for various reasons, that ambiguous weapons are nuclear-armed—a process that sometimes yields false positives.

Intelligence Collection and Analysis

Design assessments. Analysts can mischaracterize or be unable to characterize a delivery system if they make use of incorrect or uncertain assessments about its design—in particular, about whether nuclear warheads, conventional warheads, or both have been produced for it. The biggest contemporary challenge is probably determining whether a delivery system clearly designed to accommodate a nonnuclear warhead also has a nuclear role. Recent public U.S. assessments about some foreign delivery systems, such as Russia’s SSC-5 cruise missile, indicate uncertainty on exactly this point.1 In a similar vein, a series of official U.S. assessments from the past decade present contradictory assessments about whether various Chinese air-, ground-, and sea-launched cruise missiles are dual-use or equipped only with conventional warheads.2

Uncertainty—and even errors—in design assessments should not be surprising. In practice, indicators that a conventional delivery system is also available for nuclear operations are unlikely to be easily detectable. States have not tested delivery systems with live nuclear warheads for decades. Moreover, in peacetime, it appears that Chinese, Russian, and U.S. dual-use delivery systems are never loaded with nuclear warheads, which are often stored at entirely separate locations. In theory, a state could discover that an opponent had developed a nuclear warhead for some particular delivery system or that specialized troops had practiced loading weapons of that kind with mock nuclear warheads. In practice, obtaining such evidence would require the kind of deep insight into sensitive (and presumably well-protected) activities that may be available only rarely.

Superficially similar delivery systems. Even if design assessments for an adversary’s various delivery systems are accurate, analysts may be unable to distinguish between different weapon types after deployment. The U.S. government, for example, has publicly assessed whether each Chinese DF-21 variant is nuclear or conventional. Especially in a crisis or conflict, however, the United States might mistake those superficially similar weapons by misidentifying, say, a conventionally armed DF-21 as its nuclear-armed variant.

In fact, distinguishing between superficially similar delivery systems has sometimes proved to be difficult even outside of a crisis or conflict. The sale of Soviet MiG-23 aircraft to Cuba in the late 1970s exemplifies the challenges. In 1978, U.S. surveillance flights over Cuba failed to establish whether the aircraft in question were MiG-23Fs (known to the Soviet Union as MiG-23BNs) or MiG-23Ds (also known as MiG-27s).3 The former were widely exported and used exclusively for conventional operations; the latter were deployed, at that time, only by Soviet forces and had a nuclear role. Key differences between these models, including certain engine parts and components for carrying weapons, would not have been visible from aerial photography.4

Even though this incident occurred in peacetime, the resulting uncertainty sparked high-level concern within the U.S. government. A Special Coordination Committee, involving secretary of defense Harold Brown and national security adviser Zbigniew Brzezinski, was created to formulate a response. On November 14, 1978, this committee concluded that, although the introduction of either version would contravene a 1962 Soviet commitment not to deploy “offensive” weapons in Cuba, “[t]he [MiG-23D] represents an additional issue” because of “its nuclear capability and previous deployment only with Soviet forces.”5 Two days later, secretary of state Cyrus Vance delivered a formal demarche on the subject to Soviet ambassador Anatoly Dobrynin. Vance went beyond his formal talking points, which expressed equal opposition to both variants, and, in an apparent reference to the possibility that MiG-23Ds had been transferred, expressed particular “concern about the possible presence of ground attack aircraft which we regarded [sic] as offensive.”6

The ambiguity was not clarified until February 1979. Likely acting on Soviet instructions, Cuba published, in the official magazine of its armed forces, high-quality photographs of the aircraft, which the Central Intelligence Agency (CIA) determined were “clearly” MiG-23Fs.7 Assistance from the state responsible for a deployment may not always be so forthcoming in the future—especially during a crisis or conflict, when ambiguity might need to be resolved in hours or days, rather than months, to avert escalation.

Dual-use delivery systems. Systems designed to be dual-use, such as the SSC-8, create another layer of complexity and a particular risk of mischaracterization or uncertainty. Conclusive characterization generally requires identifying whether nuclear warheads are physically attached to delivery systems. Even the presence of nuclear warheads available for loading at short notice is not conclusive unless nonnuclear warheads are known not to be available.

Warheads, however, are difficult to detect. They are small; can be transported relatively secretly (at least when compared to many delivery systems); and are usually concealed within delivery systems, storage bunkers, or transportation systems. These characteristics complicate intelligence collection, particularly overhead imagery. Consequently, definitive evidence of the presence of nuclear warheads—from communications intercepts or human intelligence, perhaps—may not always be available. In this case, troops who have received specialized training or equipment and facilities typically associated with nuclear warheads can be useful heuristics but do not prove that nuclear warheads are present because they are also consistent with plans to deploy warheads at short notice.

These challenges are illustrated by contemporary efforts to characterize Russian forces in Kaliningrad. Various dual-use Russian delivery systems, including aircraft, short-range ballistic missiles, and coastal defense cruise missiles, are based in the enclave. From the early 2000s onward, senior officials in NATO member states have openly claimed, and unnamed intelligence officials have periodically told journalists, that Russia has deployed nuclear weapons there.8 Yet it is often unclear whether these statements are references to dual-use delivery systems or actual nuclear warheads. Satellite imagery tells a similarly ambiguous tale. One military facility in Kaliningrad appears suitable for storing nuclear warheads, given its security features, such as partially buried bunkers surrounded by triple fencing.9 Moreover, Russia has undertaken extensive recent renovation efforts. Yet, as the U.S. analyst Hans Kristensen notes, the existence of these upgraded facilities “does not provide conclusive answers” to questions about the presence of nuclear warheads.10 Just because Russia has the capability to store warheads in Kaliningrad does not mean that it is doing so or has decided to do so. Instead, Russia may want the ability to rapidly deploy warheads to Kaliningrad in a crisis or conflict.

NATO governments, of course, may have additional sources of information to help resolve this issue. Yet, even if they are certain that nuclear warheads are not present in Kaliningrad today, ambiguity could easily reemerge in a crisis or conflict. In this case, the fog of war, coupled with Russian efforts to conceal the movement of its forces, could prevent NATO from clearly determining whether nuclear warheads had been moved to the enclave.

Extrapolation. A particular challenge can arise if both nuclear and nonnuclear versions of the same ambiguous delivery system are deployed simultaneously. In such cases, if analysts identify only one version, they may incorrectly conclude that all the weapons are armed in the same way. Such extrapolation can result in false positives or false negatives. While the underlying logic may seem obviously flawed, extrapolation is often inevitable because intelligence information is often incomplete.

Theoretically, errors created by extrapolation could be identified as more information became available. In practice, however, unless analysts realized that their conclusions might be wrong and requested more information, intelligence collection on the ambiguous missiles might be curtailed. Moreover, even if contradictory evidence did emerge, at least two psychological biases might prevent analysts from revising their conclusions.

First, people have a well-documented tendency, known as confirmation bias, to interpret new information as being consistent (or at least not inconsistent) with the expectations created by preexisting beliefs.11 In the intelligence world, analysts can ignore new information because their preconceptions prevent them from understanding its significance. In 1962, for example, U.S. analysts were not expecting the Soviet Union to ship missiles to Cuba, and, as a result, they ignored reports about two unusual cargo ships bound for the island. These ships had “exceptionally large hatches” but also were riding high in the water, implying that they carried “cargo of low weight and high volume,” such as ballistic missiles.12 Similarly, after identifying one variant of an ambiguous weapon, analysts might ignore new information that suggested the presence of a second variant.

If analysts identify only one version of an ambitious delivery system, they may incorrectly conclude that all the weapons are armed in the same way.

Second, in making estimates, people tend to anchor—that is, they give undue weight to the first piece of information encountered and thus make insufficient revisions as new information emerges.13 Anchoring can help explain, for instance, the United States’ persistent and incorrect belief, in the late 1950s and early 1960s, that the Soviet Union had much greater ICBM production capabilities than the United States. For various reasons, initial estimates greatly exaggerated Soviet capabilities. Because subsequent estimates were formulated as revisions to previous ones, the U.S. intelligence community—Air Force intelligence staff, in particular—continued to assert the existence of the so-called missile gap long after the totality of the evidence suggested otherwise.14 Similarly, even if analysts recognize new evidence that calls into question their initial hypothesis about the type of armament on some ambiguous delivery vehicle, they may give that evidence insufficient weight.

In the run-up to the Cuban Missile Crisis, the CIA misidentified nuclear-armed cruise missiles as conventionally armed ones. Shortly before the crisis, the Soviet Union delivered two different types of missiles to Cuba for coastal defense and for targeting the U.S. naval base at Guantánamo Bay: conventionally armed SSC-2B missiles (or Sopka in Soviet terminology) and nuclear-armed SSC-2A missiles (Frontonvye Krylatye Rakety-1), along with about eighty nuclear warheads. Although SSC-2A and SSC-2B missiles had some differences in their guidance systems, launchers, and support equipment, they were based on the same weapon, the AS-1 air-to-surface missile, and had nearly identical airframes.15 Using photographs obtained from U-2 surveillance aircraft, U.S. intelligence analysts correctly identified operational conventionally armed SSC-2B missiles at four sites. The nuclear-armed SSC-2A missiles, however, were still crated at this time, preventing the collection of detailed intelligence about them and leading analysts to extrapolate and misidentify them as yet more SSC-2Bs.16 The historians Ernest May and Philip Zelikow have concluded that, as a result, the United States “did not know and never seriously imagined that the coastal defense cruise missiles were deployed with nuclear warheads.”17

Crisis and wartime complications: U.S. and Russian forces. Intelligence-collection capabilities have improved enormously since the Cuban Missile Crisis and by almost as much since the Soviet Union sold MiG-23s to Cuba in the late 1970s. These improvements, which include more capable remote-sensing technology and the advent of espionage in cyberspace, could help mitigate the effects of warhead ambiguity. In assessing modern intelligence-collection techniques, however, the real question is not how they perform in peacetime but how they would perform during the confusion of a crisis or, especially, a contemporary conventional conflict. In such contingencies, nuclear and nonnuclear weapons would be more entangled and the consequences of mischaracterization would be more serious.

All dual-use U.S. aircraft are available for both nuclear and nonnuclear operations; they are not grouped, geographically or organizationally, according to function. Likewise, Russia appears to group few, if any, of its ambiguous aircraft and ground-launched missiles functionally.18 But while the entanglement of these delivery systems in any context increases the likelihood of an adversary’s mischaracterizing them, in peacetime, that likelihood is very low. Since 1968, the United States has followed a policy of not flying live nuclear warheads on combat aircraft outside of a crisis or conflict.19 Russia is reported to have a similar policy in place.20 Moreover, as recently as 2015, the Russian government stated that all of its “non-strategic nuclear weapons” were “non-deployed” and located in “centralized storage depots.”21 These policies help reduce the likelihood of a nonnuclear delivery system being mistaken for a nuclear one, even if they were adopted to help prevent nuclear accidents and enhance physical security. Indeed, these policies—especially the prohibition on loading combat aircraft with nuclear weapons—are highly credible precisely because they have such sensible motivations.22

Russia and the United States might attack each other’s ISR assets as a way of undermining the other’s ability to fight a conventional war. Such attacks could make the characterization of ambiguous delivery systems even more difficult.

In a crisis or conflict, however, mischaracterization due to entanglement would become more likely. Nuclear warheads might be loaded onto ambiguous delivery vehicles. In theory, the observing state could determine how individual aircraft or missiles were armed by identifying whether they had been loaded from a storage facility for nuclear warheads or one for nonnuclear munitions. In practice, amid the fog of war, making such a determination could be very difficult, if not impossible. Moreover, Russia and the United States might attack each other’s ISR assets as a way of undermining the other’s ability to fight a conventional war.23 Such attacks could make the characterization of ambiguous delivery systems even more difficult.

Further exacerbating the challenges, nuclear operations would likely be conducted alongside nonnuclear ones. As a result, delivery systems carrying nuclear warheads might be operating nearby identical, or nearly identical, delivery systems that were conventionally armed. While there could be some indicators that a delivery system was nuclear-armed (such as the presence of escort vehicles), the most reliable way to determine its warhead type would be to track each delivery system continuously, from the time it was first loaded with a warhead. But, in practice, this task could prove daunting. Some dual-use U.S. and Russian aircraft are stealthy, and any capability to detect such aircraft at close range would likely be ineffective at greater distances.24 Moreover, to ensure its mobile missiles are survivable, Russia would likely try to interfere with the United States’ ability to monitor them. Such efforts—for example, the use of camouflage or inflatable decoys—would only need to be partially effective to prevent the United States from tracking each Russian missile continuously.25

Crisis and wartime complications: Chinese forces. China’s nuclear and nonnuclear forces are less entangled than their U.S. and Russian equivalents on a day-to-day basis. To date, the PLA Rocket Force and its predecessor, the Second Artillery Corps, appear to have operated geographically and organizationally distinct launch brigades for nuclear and conventional missiles.26 (Because China is believed to store nuclear warheads and missiles separately in peacetime, missiles that are assigned a nuclear role—that is, “nuclear missiles”—are generally not nuclear-armed prior to a crisis or conflict.) Therefore, if a particular launch brigade undertakes an exercise in peacetime, it should be clear whether its missiles are conventional or nuclear.

In a crisis or conflict, however, China might disperse its conventionally armed and nuclear-armed weapons simultaneously, complicating the task of characterizing them—a concern that the U.S. Department of Defense has recently highlighted.27 Characterization would be most challenging if the deployment areas for conventionally armed and nuclear-armed variants of the same missile overlapped. In this regard, the key question is how far missiles might travel from their garrisons while deployed. Open sources do not provide a definitive answer. That said, under the 1991 Strategic Arms Reduction Treaty, Soviet mobile ICBM deployment areas were limited to an area of 125,000 square kilometers (49,000 square miles), which is equivalent to a circle with a radius of 200 kilometers (130 miles).28 Presumably, Moscow would not have agreed to this limitation if its plans had called for mobile ICBMs to travel significantly greater distances from their garrisons. If, in a conflict or crisis, Chinese missiles migrated over roughly the same distance, then conventionally and nuclear-armed variants of the same missiles might intermingle if based less than 400 kilometers (250 miles) apart.

According to open-source information, only one pair of brigades—the 611 and 612 brigades, located almost 200 kilometers apart—may meet this criterion today. While reliable information on these units is scant, it is possible (but far from certain) that one operates nuclear DF-21s and the other conventional DF-21s.29 Other pairs of brigades that host nuclear and conventional variants of the same missiles appear to be many hundreds, and more often thousands, of kilometers apart. While this situation could change if China restructures its missile bases, the degree of entanglement within China’s missile force would probably be quite limited in a contemporary crisis or conflict—assuming, that is, that Chinese mobile missiles remained within 200 kilometers of their garrisons.30

There are reasons to question this assumption though. Chinese missiles travel much farther to conduct exercises.31 Moreover, The Science of Second Artillery Campaigns, a leaked classified textbook from 2004, states that conventional missiles could be transported between theaters in a conflict to where they were needed.32 Such “cross-theater maneuvering” could significantly increase the degree of entanglement between China’s nuclear and conventional missile forces. Even if China planned to avoid the intermingling of nuclear and conventional missiles in this scenario, the United States would have to be aware of this planning for it to decrease the likelihood of uncertainty or mischaracterization.

Regardless of the distances over which Chinese missiles might migrate, the United States could try to track them continuously, starting from their garrisons, as a way of characterizing them. Other evidence might also be available; some exercises, for example, suggest that missiles and nuclear warheads might be mated in the field (though some analysts believe that mating would likely occur before deployment, most likely in underground facilities for protection).33 If, therefore, a DF-26 missile were to meet up with a vehicle for transporting nuclear warheads, there would be strong grounds for concluding that the missile was nuclear-armed. More speculatively, there may be some differences in the support vehicles for nuclear and conventional launch units.34

Looking forward, the characterization challenge will grow if the Rocket Force further entangles its nuclear and nonnuclear missiles.

Yet, in a crisis or conflict, this evidence could be elusive. The PLA Rocket Force would likely try to obscure its operations to enhance the survivability of its missiles. At the very least, it would probably try to conceal its weapons. Indeed, the techniques discussed by the leaked PLA textbook include the use of camouflage and moving missiles at night, under cloud cover, or during the “blind intervals” of enemy ISR assets (an apparent reference to satellites’ periodic coverage).35 The Rocket Force might also hide nuclear missiles in tunnels or shelters at various times.36

Moreover, China’s efforts to protect its missiles might not be limited to such passive measures. The Science of Second Artillery Campaigns also discusses various disinformation techniques, including the “feint maneuver” (in which a small force is used to divert attention from a larger one).37 Even more significantly, the PLA appears to plan to—and would likely experience strong pressures to—attack U.S. ISR assets in a conflict, most likely to undermine U.S. conventional operations.38 Regardless of Chinese intentions, such attacks would increase the likelihood of the United States’ mischaracterizing, or being unable to characterize, ambiguous Chinese missiles.

Looking forward, the characterization challenge will grow if the Rocket Force further entangles its nuclear and nonnuclear missiles. In the near term, the most likely driver of change is the dual-use DF-26 missile. This missile could be integrated into the Rocket Force in two ways, and it is not yet apparent which approach China is adopting. One option would be to retain the Rocket Force’s existing structure and create separate conventional and nuclear DF-26 brigades.39 However, this approach would not leverage the weapon’s “change the warhead, not the missile” feature.40 Consequently, it seems possible that China will posture individual DF-26 brigades for both nuclear and conventional operations, making it more difficult to characterize DF-26 missiles. This step could be a harbinger of more widespread restructuring. There may be flexibility and significant cost savings associated with dual-use missile brigades, and if the Rocket Force sets up one and establishes a precedent, internal supporters of separation may find it more difficult to defend their position, potentially resulting in the widespread adoption of dual-use brigades.

Interpretation by Decisionmakers

Given the challenges of intelligence collection and analysis, assessments about how ambiguous weapons are armed can be inconclusive. Yet, especially in a crisis or conflict, military and national leaders may not fully account for uncertainty for two reasons. First, intelligence analysts may not communicate the full extent of their uncertainty to decisionmakers for fear that it would be unwelcome or reflect badly on their work. Second, even if their uncertainty is properly conveyed, decisionmakers may nonetheless assume that ambiguous delivery vehicles are nuclear-armed. This tendency, to which both military considerations and psychological factors contribute, does not always lead to the wrong conclusion, but when it does, it produces false positives.

In theory, decisionmakers do not need to make assumptions; they could accept the uncertainty and plan on that basis (by, for example, trying to identify a course of action that would produce an acceptable outcome whether or not the weapons in question were nuclear-armed). To help manage the uncertainty, decisionmakers could even try to estimate the probability that the weapons were nuclear-armed. Yet there seem to be few precedents for such behavior (as U.S. experience during the Cuban Missile Crisis, described below, exemplifies) and both military and psychological reasons not to expect it.

In trying to manage a crisis or conflict, decisionmakers inevitably make assumptions, including about their opponents’ capabilities and intentions. One driver is the acute need to reduce, to a manageable level, the complexity that is associated with any contingency. A second is people’s tendency to seek certainty, even when the evidence is inconclusive. For example, politicians have criticized scientists for giving “wishy-washy, iffy” probabilistic answers to questions about, say, the health effects of sweeteners or pollution instead of providing an unequivocal yes or no.41

In making assumptions, decisionmakers are likely to—and, indeed, should—consider the risks of being wrong. When warhead ambiguity arises, the costs of a false negative could be high and much greater than those of a false positive. Of course, decisionmakers should recognize that nuclear weapons are generally less likely to be used than conventional weapons, but in a major conflict, when conventional deterrence has already failed, the danger that nuclear deterrence could also fail may seem very real. As a result, it is reasonable, from a military perspective, for decisionmakers to assume that ambiguous weapons are nuclear-armed.

Less rationally, the negativity bias—a psychological effect that leads to “negative events [being] more potent with respect to their objective magnitude than . . . positive events”—may also play a role.42 One consequence is that possible negative outcomes exert a disproportionate influence on human behavior (that is, the influence is larger than predicted by a rational cost-benefit analysis). This effect, for example, helps explain why states tend to inflate threats when assessing their severity.43 And in this vein, when ambiguous weapons are deployed, the possibility of their being nuclear-armed may loom even larger in decisionmakers’ minds than it should.

Assumptions during the Cuban Missile Crisis. U.S. decisionmakers on ExComm, the Executive Committee of the National Security Council set up to manage the crisis, confronted multiple deployments of ambiguous Soviet weapons. At no time before or during the crisis did the U.S. intelligence community detect any Soviet nuclear warheads in Cuba. Because warheads are easy to conceal, however, the U.S. intelligence community had little confidence that, if warheads were present, it would detect them. Therefore, logically, neither the intelligence community nor the ExComm members took the absence of evidence to be evidence of absence. Yet the decisionmakers went further than simply recognizing that nuclear warheads may have been present; they quickly assumed that warheads were, in fact, available for SS-4 and SS-5 missiles, IL-28 bombers, and MiG-21 fighters. Decisionmakers were fully aware of the lack of evidence supporting their assumptions, especially in the case of aircraft, suggesting that they were deliberately planning for a worst-case scenario.

ExComm members’ assumptions about the missiles in Cuba were well-grounded. The CIA had previously assessed that these weapons were deployed exclusively with nuclear warheads.44 Echoing his agency’s logic, director of central intelligence John McCone explained to members of Congress, on October 22, 1962, that because SS-4 and SS-5 missiles “are relatively ineffective weapons without nuclear warheads, we think it prudent to assume that nuclear weapons are now or shortly will be available in Cuba.”45 The historian David Coleman notes that “policymakers accepted, internalized, and applied this principle” to other systems, even though the logical basis for doing so was significantly weaker in the case of the aircraft.46

Decisionmakers were fully aware of the lack of evidence supporting their assumptions, especially in the case of the aircraft, suggesting that they were deliberately planning for a worst-case scenario.

Prior to the crisis, the U.S. intelligence community had concluded that IL-28 aircraft had the “capability of delivering nuclear weapons,” implying that they could also be used with conventional weapons.47 This assessment could have led decisionmakers to be cautious about assuming that the IL-28s deployed to Cuba were nuclear-armed, but they showed no hesitation to make plans on that basis. At various points during the first week of the crisis, secretary of defense Robert McNamara and chairman of the Joint Chiefs of Staff Maxwell Taylor, as well as Air Force chief of staff Curtis LeMay (who was not on ExComm), opposed U.S. military strikes that were focused solely on Soviet missiles.48 Their argument, which was more explicit in some cases than others, was that because aircraft could have similar effects against the United States as the missiles, there was little point taking out the latter but not the former.

The case for the MiG-21s being nuclear-armed was substantially weaker than for the IL-28s. Prior to the crisis, U.S. intelligence analysts were not certain that the MiG-21 had been assigned a nuclear role and assessed that, even if it had been, carrying nuclear weapons would severely limit its range.49 In fact, on September 4, 1962, the day before the United States first detected MiG-21s in Cuba, McGeorge Bundy, the national security adviser and future ExComm member, described the arming of this type of aircraft “with jerry-built nuclear weapons” as “not a likely configuration.”50

During the crisis, no clear evidence emerged about how the MiG-21s were armed, but some ExComm members nonetheless assumed that nuclear warheads were available. The issue arose on the first day of the crisis, October 16, once again in discussions about the targets for possible U.S. airstrikes. McNamara emphasized the threat the MiG-21s posed, arguing that they should be attacked in any military strike. He explicitly recognized the limitations of the available intelligence regarding their armaments but then added that “if there are nuclear warheads associated with the [missile] launchers,” which everyone believed there were or shortly would be, “you must assume there will be nuclear warheads associated with [the] aircraft.”51 Taylor made a similar argument later the same day.52 However, not all ExComm members shared McNamara and Taylor’s assessment. Most importantly, president John F. Kennedy argued that the MiG-21s were probably armed with “iron [conventional] bombs . . . because, obviously, why would the Soviets permit nuclear war to begin under that sort of half-assed way?”53 That said, given that the secretary of defense and the chairman of the Joint Chiefs of Staff believed the MiG-21s were nuclear-armed, it seems likely that military plans were developed on that basis.

What matters about these assessments is not whether they turned out to be correct (as with the SS-4s, SS-5s, and IL-28s) or incorrect (as McNamara and Taylor were with respect to the MiG-21s). Rather, what matters is that they were assumptions and that the decisionmakers who made them did so consciously, knowing that the available intelligence did not support a definitive conclusion about how the IL-28s and MiG-21s, in particular, were armed. A sense of what “prudent” crisis management required shaped the thinking of most ExComm members. Moreover, it is probably not coincidental that McNamara and Taylor, the two ExComm members most responsible for military planning, were the most vocal in urging their colleagues to think in this way. The relative importance of rational military calculations and the negativity bias in shaping these perceptions is, however, unclear.


1 Defense Intelligence Ballistic Missile Analysis Committee, Ballistic and Cruise Missile Threat 2017, 37. In this document, the SSC-5 is listed by its Russian designation, 3M55.

2 Kristensen and Norris, “Chinese Nuclear Forces, 2018,” 294. Moreover, no recent edition of the U.S. Department of Defense publication, Military and Security Developments Involving the People’s Republic of China, has suggested that any of these missiles are nuclear-armed.

3 Jeffrey T. Richelson, “Spying on the MiGs,” Air Force Magazine (February 23, 2011),

4 U.S. Interests Section Havana to Department of State, “MiG-23’s in Cuba,” Telegram 1052, February 20, 1979, 1979HAVANA1052, RDP06T01849R000100040019, CREST System, 1, Confusingly, sources differ on the correct “translations” between U.S. and Soviet designation; this report uses the designations from the contemporary U.S. documents it references.

5 This commitment had helped end the Cuban Missile Crisis. “Summary of Conclusions of a Special Coordination Committee Meeting,” November 14, 1978, in Foreign Relations of the United States 1977–1980, vol. XXIII, Mexico, Cuba, and The Caribbean (Washington, DC: Government Printing Office, 2016), document 38.

6 Arnold L. Horelick, “MIG-23s,” Memorandum for the Record, November 16, 1978, RDP06T01849R000100030028, CREST System, 1, For Vance’s official talking points, see Department of State to Embassy Moscow, “Demarche to Soviets on MiG-23’s in Cuba,” Telegram 290640, November 16, 1978, 1978STATE290640, RDP06T01849R000100030029, CREST System,

7 Director of Central Intelligence, “National Intelligence Daily,” March 7, 1979, RDP79T00975A031200060002, CREST System, 12, Further MiG-23s exports in the early 1980s reignited the controversy.

8 Hans M. Kristensen, Non-Strategic Nuclear Weapons, Special Report 3 (Federation of American Scientists, May 2012), 70–74,

9 Kristensen, Non-Strategic Nuclear Weapons, 74.

10 Hans M. Kristensen, “Russia Upgrades Nuclear Weapons Storage Site in Kaliningrad,” Strategic Security (blog), June 18, 2018,

11 Jervis, Perception and Misperception in International Politics, 117–202.

12 Roger Hilsman, To Move a Nation: The Politics of Foreign Policy in the Administration of John F. Kennedy (Garden City, NY: Doubleday & Company, 1967), 186–187.

13 Amos Tversky and Daniel Kahneman, “Judgment Under Uncertainty: Heuristics and Biases,” Science 185, no. 4157 (1974): 1128–1130.

14 Lawrence Freedman, U.S. Intelligence and the Soviet Strategic Threat (Boulder, CO: Westview Press, 1977), 79–80. Freedman labels this effect a “bureaucratic tendency,” but it is probably better explained as a psychological phenomenon given the weight of experimental evidence for anchoring.

15 Steven Zaloga, letter to the editor, Cold War International History Project Bulletin, nos. 12–13 (Fall/Winter 2001): 360.

16 Coleman, “The Missiles of November, December, January, February,” 18–19.

17 Ernest R. May and Philip D. Zelikow, eds., The Kennedy Tapes: Inside the White House During the Cuban Missile Crisis, concise edition (New York: W. W. Norton & Company: 2002), 425.

18 There is no single definitive proof of this claim; rather, many pieces of evidence, when taken together, paint a convincing picture. On Russian dual-use missiles, for example, see Sutyagin, Atomic Accounting, 55. Dual-use Tu-160 bombers are based at only one site, the 6950th Guards Air Base in Engels (where nuclear weapons are almost certainly stored), strongly suggesting that those aircraft are available for both nuclear and nonnuclear operations. Tu-95MS bombers housed at the same base have been involved in nonnuclear operations (including striking targets in Syria with conventional Kh-101 cruise missiles), suggesting that they are similarly available for nuclear and nonnuclear operations. Hans M. Kristensen and Matt Korda, “Nuclear Upgrades at Russian Bomber Base and Storage Site,” Strategic Security (blog), February 25, 2019,; Ministry of Defense of the Russian Federation, “Strategic Tu-95MS Bombers Destroyed the ISIS Militants’ Command Post and Storages in Syria With a Missile Attack,” press release, July 5, 2017,; and Pavel Podvig, “Strategic Aviation,” Russian Strategic Nuclear Forces (blog), June 20, 2017,

19 Hans M. Kristensen, “Flying Nuclear Bombs,” Strategic Security (blog), September 5, 2007,

20 Sutyagin, Atomic Accounting, 9.

21 Russian Federation, “Practical Steps of the Russian Federation Towards Nuclear Disarmament,” 2015 Review Conference of the Parties to the Treaty on the Non-Proliferation of Nuclear Weapons, New York, April 27–May 22, 2015, 20,

22 For the origins of the U.S. policy, see Office of the Historian, Headquarters Strategic Air Command, Alert Operations and The Strategic Air Command, 1957–1991 (Headquarters Strategic Air Command: Offutt Airforce Base, NE, December 7, 1991), 23–24,

23 Daniel R. Coats, “Worldwide Threat Assessment of the U.S. Intelligence Community,” statement for the record, Office of the Director of National Intelligence, February 13, 2018, 13,; Defense Intelligence Agency, Russia Military Power, 36; and Air-Sea Battle Office, “Air-Sea Battle: Service Collaboration to Address Anti-Access & Area Denial Challenges,” May 2013, 7,

24 In fact, because of the curvature of the earth, it may not even be possible to keep track of nonstealthy aircraft continuously.

25 On Russia’s use of inflatable decoys, albeit not for the purpose of protecting the specific weapons under discussion here, see Andrew E. Kramer, “A New Weapon in Russia’s Arsenal, and It’s Inflatable,” New York Times, October 12, 2016,

26 Fiona S. Cunningham and M. Taylor Fravel, “Assuring Assured Retaliation: China’s Nuclear Posture and U.S.-China Strategic Stability,” International Security, 40, no. 2 (Fall 2015): 42–44. On China’s practice of storing missiles and warheads separately, see Mark A. Stokes, China’s Nuclear Warhead Storage and Handling System (Project 2049 Institute, March 12, 2010), 6,

27 Office of the Secretary of Defense, Military and Security Developments Involving the People’s Republic of China 2019, 66. See also, Talmadge, “Would China Go Nuclear?,” 73–75.

28 1991 Strategic Arms Reduction Treaty (START), article VI.3.

29 Mark Stokes, PLA Rocket Force Leadership and Unit Reference (Project 2049 Institute, November 30, 2018), 4; Lewis, Paper Tigers, 116. The 611 and 612 brigades were, until recently, named the 807 and 811 brigades, respectively.

30 If the United States were unable to distinguish between nuclear- and conventionally armed variants of different missiles—such as conventional DF-15s and nuclear DF-21s—China’s nuclear and conventional missiles forces would be significantly more entangled.

31 Hans M. Kristensen, “Chinese DF-26 Missile Launchers Deploy to New Missile Training Area,” Strategic Security (blog), updated January 31, 2019,

32 Second Artillery Corps, People’s Liberation Army, The Science of Second Artillery Campaigns (Beijing: PLA Press, 2004), 159.

33 Li Bin, “Tracking Chinese Strategic Mobile Missiles,” Science & Global Security 15, no. 1 (2007): 10–11. For the view that mating would be most likely to happen before deployment, see Lewis, Paper Tigers, 119.

34 For example, one interpretation of the data in Office of the Secretary of Defense, Military and Security Developments Involving the People’s Republic of China 2019, 117, is that conventional units may typically be assigned more missiles with which to reload their transporter-erector launchers than nuclear units. If so, the association of multiple missile transport vehicles with a single transporter-erector launcher may suggest missiles that are conventionally armed. For a description of the support vehicles involved in missile operations, see Dennis J. Blasko, The Chinese Army Today: Tradition and Transformation for the 21st Century, 2nd ed. (London: Routledge, 2012), 107.

35 Second Artillery Corps, The Science of Second Artillery Campaigns, 376–377 and 379–380.

36 Joshua H. Pollack, “Boost-Glide Weapons and U.S.-China Strategic Stability,” Nonproliferation Review 22, no. 2 (2015): 159; Lewis, Paper Tigers, 115 and 118; and Li, “Tracking Chinese Strategic Mobile Missiles,” 10.

37 Second Artillery Corps, The Science of Second Artillery Campaigns, 388.

38 Office of the Secretary of Defense, Military and Security Developments Involving the People’s Republic of China 2019, 51; and Avery Goldstein, “First Things First: The Pressing Danger of Crisis Instability in U.S.-China Relations,” International Security 37, no. 4 (Spring 2013): 67–68.

39 Jordan Wilson, China’s Expanding Ability to Conduct Conventional Missile Strikes on Guam (Washington, DC: U.S.-China Economic and Security Review Commission, 2016), 8,

40 Erickson, “Academy of Military Science Researchers: ‘Why We Had to Develop the Dongfeng-26 Ballistic Missile.’”

41 Paul Slovic, Baruch Fischhoff, and Sarah Lichtenstein, “Rating the Risks” in The Perception of Risk, ed. Paul Slovic (Abingdon: Earthscan, 2000), 110–111. For similar behavior in business contexts, see James G. March and Zur Shapira, “Managerial Perspectives on Risk and Risk Taking,” Management Science 33, no. 11 (November 1987): 1407–1408.

42 Paul Rozin and Edward B. Royzman, “Negativity Bias, Negativity Dominance, and Contagion,” Personality and Social Psychology Review 5, no. 4 (2001): 298. See also, Dominic D. P. Johnson and Dominic Tierney, “Bad World: The Negativity Bias in International Politics,” International Security 43, no. 3 (Winter 2018/19): 102–109.

43 Johnson and Tierney, “Bad World,” 102–103. The negativity bias may have contributed, in the late 1950s, to the U.S. intelligence community’s exaggerated initial estimates of Soviet ICBM production capabilities. Freedman, U.S. Intelligence and the Soviet Strategic Threat, 79.

44 National Intelligence Estimate 11-5-61, Table IV.

45 May and Zelikow, eds., The Kennedy Tapes, 166.

46 Coleman, “The Missiles of November, December, January, February,” 41.

47 Director of Central Intelligence, Soviet Gross Capabilities for Attack on the US and Key Overseas Installations and Forces Through Mid-1959, National Intelligence Estimate 11-56, March 6, 1956, RDP79R01012A006500040001, CREST system,, 7.

48 May and Zelikow, eds., The Kennedy Tapes, 82, 117, 130, and 135.

49 Coleman, “The Missiles of November, December, January, February,” 16.

50 Timothy Naftali and Philip Zelikow, eds., The Presidential Recordings: John F. Kennedy: The Great Crises, Volume 2, September–October 21, 1962 (New York: W. W. Norton & Company, 2001), 21.

51 May and Zelikow, eds., The Kennedy Tapes, 41. For McNamara’s and Taylor’s comments suggesting that the MiG-21s could also be armed with conventional warheads, see Naftali and Zelikow, eds., The Presidential Recordings, Volume 2, 439–440.

52 May and Zelikow, eds., The Kennedy Tapes, 58. Taylor hints at this again on page 65. Brief statements by McNamara and Taylor later in the crisis suggest that they did not change their views. These statements went unchallenged—though it is unclear whether the other principals had come to agree with McNamara and Taylor or simply decided the issue was not worth arguing about. See May and Zelikow, eds., The Kennedy Tapes, 82 and 135.

53 May and Zelikow, eds., The Kennedy Tapes, 49. Later the same day, Kennedy made the same argument, eliciting approval from Bundy and Secretary of State Dean Rusk. Naftali and Zelikow, eds., The Presidential Recordings, Volume 2, 440.