China’s growing nuclear arsenal has prompted a debate — in think tanks and research institutions and, more quietly, in the bowels of the Pentagon and U.S. Strategic Command —about how the United States should adapt its nuclear posture to the emergence of a second “nuclear peer.” The central issue is the arcane subject of nuclear targeting: the question of what facilities the United States should, in the detached language of nuclear strategists, “hold at risk,” or in plain English, threaten to nuke.
America’s current targeting policy is often described as “counterforce.” This term is widely and correctly understood to mean that the United States targets its adversaries’ nuclear forces and nuclear command-and-control capabilities, and widely and incorrectly understood to imply that it does not target other kinds of assets. As a result, the conventional wisdom is that the United States should build-up its nuclear forces so it can target China’s and Russia’s nuclear forces simultaneously. Earlier this year, for example, a study group convened by the Lawrence Livermore National Laboratory argued that “[a]n increase in the number of targetable Chinese nuclear weapons implies an increase in the number of U.S. nuclear weapons to target them.” Last year, a former senior U.S. official, Franklin Miller, went further and called for the United States “to threaten, separately and in combination, both Russia’s and China’s key assets,” including their “military forces.”
An alternative view, which two colleagues and I recently advanced in Foreign Affairs, is that the United States should abandon targeting nuclear forces, their command-and-control systems, and an adversary’s leadership. China’s and Russia’s nuclear forces are, and will remain, sufficiently survivable that preemptive strikes on them would not succeed in meaningfully limiting the damage that the United States would suffer in a nuclear war. As a result, targeting those forces does not enhance deterrence — but it does create serious risks and costs. In a conventional conflict, Chinese or Russian fears that the United States might launch large-scale attacks on their nuclear forces could induce them to use nuclear weapons first, before those weapons were destroyed. Such use could be limited and aimed at terrifying the United States into ceasing the conflict rapidly. Alternatively, the Russian military might launch large-scale preemptive attacks on U.S. nuclear forces (China currently lacks the capabilities to do so but may acquire them). Moreover, if the United States builds up its nuclear forces, China and Russia are likely to respond by building up theirs, which, by the logic of counterforce targeting, will then require a countervailing U.S. increase. The result will be a futile and difficult-to-stop three-way arms race that is likely to exacerbate tensions and increase the danger of war. In sum, enlarging its arsenal will likely make the United States less secure and less solvent.
Of course, calling for an end to counterforce targeting requires articulating an alternative. I advocate CMI targeting, which is an acronym I came up with to describe holding Conventional Military forces and war-supporting Industry at risk. Crucially, such categories are likely already included within existing U.S. war plans so should avoid the moral and legal complications of so-called countervalue targeting, that is, deliberately targeting population.
The debate over targeting — and by extension U.S. nuclear strategy toward China and Russia — is an important and complex one. Unfortunately, it is being warped by the existence of two persistent myths:
Myth 1: Counterforce targeting would avoid strikes on cities in all circumstances.
Myth 2: The only alternative to counterforce targeting is planning for large-scale strikes against cities.
I argue that these myths are, in fact, myths. To be sure, accepting my argument hardly settles the debate over targeting policy — it is entirely possible to favor counterforce targeting while recognizing that many counterforce targets are in cities and that population targeting is not the only alternative. However, the existence of these myths matters because they unfairly skew the debate by exaggerating the difference in civilian casualties that would likely be caused by counterforce attacks compared to the alternatives — an important factor in assessing the legality and morality of different targeting policies.
Myth 1: Counterforce targeting would avoid strikes on cities in all circumstances
There’s a widespread assumption that counterforce strikes would be directed exclusively at nuclear forces and their command-and-control capabilities and would thus spare cities. It’s a notion propounded everywhere from Wikipedia to the Department of Defense’s Nuclear Matters Handbook, which states that “counterforce targeting plans to destroy the military capabilities of an enemy force” and lists nuclear-weapon delivery systems, “command and control centers, and weapons of mass destruction storage facilities” as “typical counterforce targets.” In this vein, four former officials — Keith Payne, John Harvey, Franklin Miller, and Robert Soofer — recently stated baldly that U.S. targeting policy precludes attacking cities. (They were responding to a call for the partial rejection of counterforce targeting).
To be fair to Payne et al., they actually made two conflicting claims about whether cities are entirely off-limits. First, they stated that “the intentional destruction of an opponent’s cities and population would violate” moral values and the Law of Armed Conflict. This formulation would not preclude attacks on legitimate military targets that are located in cities. Immediately thereafter, however, they went on to make a stronger claim that is worth quoting at length:
Within DoD [the Department of Defense] … avoiding the targeting of civilian populations and objects has long been taken very seriously and given great credence (including by the authors of this article) — it is not window dressing or pretense. For example, in 1986, Gen. Bernard Rogers, then Commander in Chief of NATO forces, made precisely this point regarding U.S. nuclear targeting: “I place certain restraints on myself in regard to collateral damage. I will not fire a nuclear weapon into a city. I am concerned about those targets that are militarily significant, that we need to strike because it will have an impact on the battlefield, but which are close to cities. I will not strike those targets if a large percentage of civilians are going to be killed.” For over four decades, every Republican and Democratic administration has publicly confirmed the continuing significance of this stricture on U.S. nuclear target planning, including most recently the Obama, Trump and Biden Administrations. [Italics in original]
Most readers would reasonably conclude from this paragraph — especially the italics that the authors added to the sentence “I will not fire a nuclear weapon into a city”— that the United States would not attack military facilities in cities under any circumstances (and would only attack military facilities nearby if such attacks were not expected to not kill “a large percentage” of a city’s inhabitants).
Such a conclusion would be incorrect, however. The 37-year-old quotation from the Supreme Allied Commander Europe was not offering a general description of U.S. nuclear strategy, but of Rogers’ specific role in it. Rogers was describing how he might use the short-range nuclear weapons assigned to his command to stave off advancing Warsaw Pact forces. Accordingly, it was almost certainly West German cities that he planned to avoid attacking with nuclear weapons.
In fact, National Security Decision Directive 13 — the U.S. nuclear weapon employment policy that was operative when Rogers made his statement — unambiguously allowed for attacks on cities in the Soviet Union and its Warsaw Pact allies. This policy’s target categories included “[n]ational-level political and military leadership” and “[t]he industrial/economic base of the Soviet Union (and its allies).” To be sure, this directive gave the president the option to withhold attacks on targets in these categories to incentivize Moscow to terminate the conflict before an all-out nuclear exchange — but this option was needed precisely because many such targets were in cities and if the United States had destroyed them the Soviet Union would likely have attacked U.S. cities in response. Had efforts to manage escalation failed and a limited nuclear war escalated to all-out exchange, then it would have been the responsibility of the commander of Strategic Air Command to fire many nuclear weapons into many cities.
Not only were cities very much on U.S target lists in 1986, but it is incorrect to claim that “every” U.S. administration for over 40 years has publicly supported a policy of not launching nuclear weapons into cities. It is true that the Obama, Trump, and Biden administrations all stated that U.S. targeting policy complies with the Law of Armed Conflict (which is presumably what Payne et al. are alluding to), but the first time that the United States made this commitment was in 2013. Before then, it was more equivocal about, if not entirely opposed to, ensuring its nuclear strategy adhered to the Law of Armed Conflict. In 1995, for example, the United States claimed that “it has long taken the position that various principles of the international Law of Armed Conflict would apply to the use of nuclear weapons” — a statement that fell short of an unconditional commitment to apply the Law of Armed Conflict to nuclear targeting. During the Cold War, the United States essentially rejected the idea that the Law of Armed Conflict should be applied to nuclear weapons. Indeed, in 1985, the Joint Chiefs of Staff privately (and successfully) opposed U.S. ratification of the 1977 Additional Protocol I to the 1949 Geneva Conventions, the international agreement most relevant to nuclear targeting, partly because “the rules against indiscriminate methods of warfare and excessive collateral damage … might severely limit the utility of [nuclear and chemical] weapons.” “Indiscriminate” attacks that could cause “excessive collateral damage” — exemplified by National Security Decision Directive 13’s provision for attacks on the “industrial/economic base of the Soviet Union (and its allies)” — would be considered illegal by contemporary U.S. standards.
Today, consistent with Payne et al.’s first claim, the U.S. government does not categorically reject the option of using nuclear weapons to target legitimate military facilities in cities. The 2022 Nuclear Posture Review report and reports on U.S. nuclear employment strategy from 2013 and 2020 all state that “the United States will not intentionally target civilian populations” (my italics). According to the 2018 Nuclear Posture Review report, the United States would aim to “minimize civilian damage to the extent possible consistent with achieving objectives” — a clear hint that the United States might launch nuclear attacks that could cause large-scale collateral damage if doing so would significantly advance legitimate military goals. Finally, the Department of Defense Law of War Manual requires that nuclear weapons (like all weapons) “be directed against military objectives” and that commanders assess whether attacks with them would be proportionate (that is whether the military benefits would outweigh the harm to civilians). These provisions would not prevent nuclear attacks against military targets in cities if commanders deemed the potential benefits to be sufficiently large.
To their credit, Payne et al. are quite open about their belief that it is essential for the United States to hold at risk military objects in cities and other densely populated areas. For example, they write that “what the authoritarian leadership in Moscow values most highly includes military power, political control and possibly their own lives, and that these must be held at risk for effective deterrence” (italics in original). They imply that Chinese leaders have similar interests. Elsewhere, Miller advocates targeting Russia’s and China’s “leaders’ ability to command and control the state, their military forces, and the industrial potential to sustain war.” In a similar vein, the newly released report of the congressionally chartered Strategic Posture Commission calls for targeting “key elements of [China’s and Russia’s] leadership, the security structure maintaining the leadership in power, their nuclear and conventional forces, and their war supporting industry.” All these statements are broadly consistent with what little is actually known about current U.S. targeting policy. Specifically, the most recent American nuclear war plan to have been even (very) partially declassified dates to 2009 and includes four target categories: “military forces,” “WMD [weapons of mass destruction] infrastructure,” “military and national leadership,” and “war supporting infrastructure.”
Effects of “Counterforce” Strikes
During the opening salvos of a limited nuclear war, the United States would almost certainly refrain from attacking military targets in cities in an effort to give its adversary a reason to show similar restraint and thus try to manage escalation. However, the United States has the option to attack military targets in cities and, over time, restraint would likely break down. Ultimately, even under a counterforce strategy, a large-scale nuclear war would likely involve extensive strikes on cities. The purpose of such attacks might not be to intentionally kill civilians, but their effects would almost certainly be deaths numbering in the tens of millions — perhaps even the hundreds of millions in the case of China.
Even attacks on an adversary’s nuclear forces could induce mass casualties in cities — a point that is often missed by those who believe that nuclear forces are in “remote areas.” Some nuclear forces are indeed in remote areas — but attacks against them could still have appalling consequences. For example, a recent study estimated that counterforce strikes against China’s three new silo fields, which are all built in isolated desert locations, could lead to millions of deaths if the order to shelter in place was given and tens of millions if it wasn’t. Most of these casualties would be in Beijing, which is 700 kilometers away from the nearest silo field, but typically downwind from all three. It is not just Chinese who would suffer. There would be deaths further afield, including in states that are allied to the United States or would likely be bystanders to a U.S.-Chinese conflict. Specifically, thousands or tens of thousands (depending once again on whether protective measures were taken) would perish in North Korea, South Korea, Japan, and Mongolia.
Some nuclear forces are closer to cities. For example, within 200 kilometers of Moscow (population 13 million) are one air base for nuclear-armed bombers and one base for mobile intercontinental-range ballistic missiles. A silo field is located within 250 kilometers of Moscow and a second base for mobile intercontinental-range ballistic missiles within 350 kilometers. Moreover, these effectively ring Moscow, meaning that Muscovites would likely be subject to substantial fallout from nuclear attacks on these targets irrespective of the wind direction. Meanwhile Russia’s third largest city, Novosibirsk (population 1.6 million) has a mobile missile base within just 30 kilometers of the downtown area.
Moreover, if the United States holds at risk the facilities that Miller, Payne et al., and the Strategic Posture Commission believe it must, then it needs to target facilities that are located next to and, indeed, actually inside cities.
Russia’s National Defense Control Center is in downtown Moscow. The headquarters for the People’s Liberation Army Rocket Force, which is believed would play a role in relaying any orders to use nuclear weapons, is in northwest Beijing (population 22 million). While these facilities and other key urban command-and-control assets may have backups in more remote locations, it is difficult to believe that the United States does not target them (as well as all known backups).
War-supporting industry presumably includes facilities for producing energy, oil, metals, and munitions, and perhaps ports and other transport facilities. Many such facilities are in cities or in densely populated areas on their outskirts. To be sure, some of them are more isolated, but it would be impossible to destroy a state’s “industrial potential to sustain war” by targeting only isolated facilities. This objective would require the obliteration of cities across China or Russia.
Destroying the “the security structure maintaining [China’s or Russia’s] leadership in power” would certainly involve attacking domestic security agencies. It may also entail attacking national government departments, local government and party facilities, and conceivably even state media. The vast majority of relevant targets are in cities and towns. For example, the headquarters of the Federal Security Services (usually known as the FSB) are located in Lubyanka Square, in central Moscow, about 5 kilometers from the National Defense Control Center. If the goal were to deprive the Chinese or Russian regimes of their hold on power, attacks would probably not be limited to capitals but would extend to key regional centers of power. Some of these are very populous. St. Petersburg, for example, has a population of 6 million, while 19 or so Chinese cities have populations greater than 5 million. In the aftermath of attacks against just a fraction of these cities, tens of millions fewer Chinese or Russians would be left alive to enjoy their newfound freedoms.
Important conventional military facilities are spread around China and Russia — as with industrial facilities some are isolated, but many are located near cities for convenience’s sake. If the United States targets such facilities, it most likely focuses on those that could play a major role in any conflict — meaning bases in the highly populated regions of south-eastern China (close to Taiwan) and western Russia (near NATO states).
Finally, if the United States seeks to hold the lives of an adversary’s leaders at risk, it must be prepared to target them wherever they are — including perhaps in cities. Indeed, there appear to be various complexes in and around Moscow and at least one on the outskirts of Beijing that are designed to shelter national leaders during a nuclear war and allow them to direct the war effort. Because these facilities are buried, the United States would probably have to attack them with some of the highest-yield nuclear warheads in its arsenal and might use multiple such warheads against each facility — potentially leading to particularly large numbers of civilian casualties, depending on the targets’ exact locations and the prevailing wind direction.
Overall, if the United States follows the “counterforce” targeting policies advocated by Miller, Payne et al., and the Strategic Posture Commission, Russian civilian death tolls could end up being measured in the tens of millions. Given China’s greater population and population density, it seems plausible that more than 100 million Chinese might perish — though this would depend on the extent of the U.S. attacks against regional power centers, conventional military bases, and war-supporting industry. Of course, actual U.S. targeting policy may be more restrained — but, even so, it is difficult to imagine that, in the event of an all-out nuclear war, the United States would not directly cause tens of millions of civilian deaths in China or Russia, even before accounting for the potential indirect effects, including famine induced by nuclear winter.
Myth 2: The only alternative to counterforce targeting is planning for large-scale strikes against cities
Most proponents of counterforce targeting give the impression that, if opponents get their way, the only nuclear strike option available to the president will be a large-scale attack on cities. Payne et al., for example, presented nuclear targeting as a binary choice between “the intentional targeting of population” and “a counterforce-oriented deterrent with graduated options” (my italics). Similarly, Henry Sokolski claimed that my co-authors and I, in our Foreign Affairs article, argued that “that the U.S. should focus its nuclear targeting against ‘value’ or ‘infrastructure’ — aka cities.” While it is true that some opponents of counterforce support population targeting and oppose limited options, others do not. Most importantly for current purposes, I oppose population targeting and support providing a president with a wide range of options.
In my view, the United States should target two categories of facilities — conventional military forces and war-supporting industry — under a policy of Conventional Military forces and war-supporting Industry (CMI) targeting. Supporters of counterforce targeting generally advocate for these categories (our disagreement is about whether the United States should hold targets in additional categories at risk). While both categories have likely declined in importance in U.S. war plans since the Cold War, I believe the United States probably still includes them in current war plans. Certainly, as of 2009, “war-supporting industry” remained a dedicated target category, while conventional military forces were probably included within “military forces” (though it is possible that all targeted “military forces” were nuclear). Indeed, it might be possible to construct CMI targeting plans by starting with current war plans and simply deleting all targets in other categories.
CMI targeting would afford the United States a wide range of options. At one extreme, a strike against an isolated industrial facility or conventional military base with a low-yield warhead would cause few civilian casualties. At the other extreme, large-scale CMI strikes would include targets in cities. As with large-scale counterforce strikes, they would result in the complete destruction of China or Russia as functioning states — so that, even if Chinese or Russian leaders survived the nuclear war, the loss of their states would effectively deprive them of political power.
The level of civilian casualties resulting from CMI strikes would depend on the exact target list. If a U.S. president wanted to adopt CMI targeting but did not want to risk causing more collateral damage, he or she could — and, indeed, should — just order the military to construct new war plans so that the expected number of civilian casualties in a large-scale nuclear war did not increase relative to existing counterforce plans. In my assessment, the possibility that a nuclear war could escalate to this level of destruction would deter any action that could be deterred by the threat to use nuclear weapons.
That said, even if CMI targeting did result in the potential for more civilian casualties, the benefits of counterforce targeting are much less marked than Payne et al. suggest. They wrote that “[r]igorous studies over decades have concluded that the intentional targeting of cities and population would likely inflict much higher civilian casualty levels than would counterforce/city avoidance targeting options.” This statement is both true and misleading. It is clearly true that the nuclear strategy designed to maximize civilian casualties would lead to more civilian casualties than any other. It is misleading because it sets up an apples-to-oranges comparison. Studies that model counterforce strikes with city avoidance spare urban targets that Miller and Payne et al. say “must” be held at risk for deterrence to be effective. They do not, therefore, model the worst-case outcomes resulting from counterforce targeting without city avoidance — the relevant benchmark for comparing with the worst-case outcomes from other targeting policies.
Conclusions
Fairly comparing the collateral damage associated with different targeting policies has important implications for their legality and morality. I do not believe that Law of Armed Conflict provides sound guidance for nuclear targeting (I agree with the argument there is a tension between preventing nuclear war and abiding by the Law of Armed Conflict, and that the former is a more important goal than the latter). That said, I recognize that, for America’s uniformed and civilian officials, adherence to the Law of Armed Conflict is an obligation and not a choice. Moreover, I firmly believe that, given the impossibility of eliminating nuclear weapons for the foreseeable future, nuclear-armed states have a moral imperative to adopt nuclear strategies that minimize the expected amount of death and suffering globally — a task that requires accounting for both the likelihood and consequences of both nuclear war and conventional war.
The United States regards war-supporting industry and conventional military forces as legitimate targets — indeed, it probably already targets them with nuclear weapons. As a result, although I am not a lawyer, it seems to me that CMI targeting is not inherently any more or less legally defensible than counterforce targeting (although to comply with the Law of Armed Conflict, individual strikes would still need to be assessed according to the principles of distinction, proportionality, and precaution).
Meanwhile, CMI targeting is the more moral choice. In my judgment, its adoption would not reduce the deterrence effectiveness of U.S. nuclear forces, but would mitigate arms-racing pressures and, much more importantly, the risks that a conventional war would turn nuclear and that a limited nuclear war would escalate into an all-out one. By decreasing the likelihood of escalation, I believe that CMI targeting would reduce the number of deaths expected to arise from nuclear war, even if, in the event of an all-out nuclear war, it would lead to greater collateral damage than counterforce targeting (though, as I note above, the United States can decide how much collateral damage its strikes risk through its choice of targets).
To be sure, developing a targeting policy is no easy task and reasonable people can reach very different conclusions. I welcome a long-overdue debate about this issue. We should argue about the legality and morality of different targeting policies as well as how they affect deterrence effectiveness and escalation risks. These issues raise questions that are objectively difficult to assess answer and would benefit from robust discussion. But let’s not start that debate from false premises.
James M. Acton is co-director of the Nuclear Policy Program and holds the Jessica T. Mathews Chair at the Carnegie Endowment for International Peace.
Image: Wikipedia
