25 September 2020

The Hunt for Mobile Missiles: Nuclear Weapons, AI, and the New Arms Race

Paul Bracken

This report examines the increasing ability of major powers to destroy moving targets, in particular, land-based mobile missiles. Yet, at the same time, it analyzes something much broader and more fundamental. Technology has changed the use of force in peace and war. These changes stem from the growing importance of advanced technologies like AI, cyber, drones, cloud computing, data analytics, and hypersonic missiles.[1] These are increasingly becoming foundational technologies for new mission areas and strategies. One of these in particular is the focus of this report: locating and destroying mobile targets. The hunt for mobile missiles, seen in this broader way, is an exemplar of advanced technologies used in national security.

An exemplar is an ideal model – an outstanding example of something which shows its feasibility. Other exemplars of advanced technology include the Manhattan Project, Sputnik, and the AI win over champions in the game of Go.[2] An exemplar is important because it shows that something can be done, even if it is only on a small scale or in a limited way. Exemplars are significant because they change how people decide what is feasible. Further, they point to its potential for the future, and its wider application.

There undoubtedly are other exemplars for advanced technologies in defense. Together, these will change how the use of force in peace and war is conceived. This is the reason for the subtitle of the report: Nuclear Weapons, AI, and the New Arms Race. This subtitle is meant to capture certain key ideas related to the focus of the report: the hunt for mobile missiles is spilling over into the nuclear arena. It provides a growing ability for the United States and others to attack the nuclear deterrent forces of other states — even though they are mobile.

The principal findings of this report cover a wide range of national security topics. Advanced technologies will alter extant military power arrangements, and for this reason will have important political impact. Changes in international order, in turn, will shape national security. At the same time, the choice of national policies on how much to invest in technology innovation, where it’s focused, and restraint on these activities will have an impact on a new arms race. This is the reason reference is made to a “new arms race” in the subtitle. It may not look like historical arms races – denoted by increases in the number of tanks, aircraft, or atomic warheads – because much of this new arms race will be hidden in algorithms, data centers, and computers. But it will still be an arms race.

Advanced technologies are reshaping national security and international order. Changes in one area changes the other. The causality doesn’t go in only one direction, (e.g., technology driving change in the international system). Rather, both national security and international order coevolve in a dynamic way.[3] Advanced technology is an important element defining this coevolution. Technological innovation is shaping international order, just as it has disrupted the industrial order. 

The key findings of our research are organized into three classes. First, there is a strategic and system level, which covers findings dealing with technology and the changing international order. This level includes major power competition and its nuclear context. This point – a nuclear context – arises because nine countries now have nuclear weapons, including most of the major powers.

Second, the report offers methodological suggestions for raising the level of debate about advanced technologies in national security. This was not anticipated at the outset of the project. But the significance of AI, cyber, drones, hypersonic missiles, and data analytics are so large that new, non-standard frameworks are needed. These new methods were developed from research about how businesses dealt with their technological challenges — that is, how businesses analyzed investments in AI, cloud computing, and data analytics and how these technologies were aligned with corporate strategy. These new frameworks – value chains, touchpoints, information chains — are taught in business schools, but are virtually unknown in academic strategic studies or in professional military education. 

A third class of findings treats operational issues. These relate to execution, organization, and tactics. They are distinctive because they are “bottom up,” so to speak, more than top down. They reflect the fact that countries use technology in different ways. A one size fits all approach won’t work for understanding military uses of the technology. They also reflect the fact that there are so many new technologies at the present time that top down direction is extremely difficult. No one is able to manage all of the innovations that are possible with the new technologies.

Strategic and System Level Findings

Absent a broader, more sober view of the hunt for mobile missiles, one that goes beyond narrow measures of performance, the world is going to see more dangerous nuclear crises, and arm races that go beyond what is necessary for prudent national security. This arises because mobile missiles have unique crisis management implications, most of which have not been studied before. Arms races result because building up nuclear forces is one way to offset improvements in reconnaissance tracking of these systems.

The United States, China, and Russia are sharply increasing their investment in the hunt for mobile missiles. The United States is doing this to find missiles in North Korea, Iran and other places. China’s main thrust is to track maritime targets like ships and aircraft. Russia’s emphasis is on disruption of NATO’s ability to project military power into Poland, the Baltics, and Ukraine. The plans for these programs are highly secret in all countries.

Although the importance of technology is sometimes exaggerated and too much attention is placed on it, the reality is often the opposite: the significance of technology is systematically understated. The speed and scale of technological change in the security environment has been underestimated in the United States for over a decade. Examples include the North Korean nuclear missile program, its hydrogen bomb, the shrinkage of warheads to fit on mobile missiles; and China’s manifold military technology advances. All of these developed faster than most groups in the United States anticipated. Until 2015, the enormous vulnerability of 5G technologies, electric power systems, satellites, and telecommunications had been discounted or ignored by most elements of the U.S. security establishment, including intelligence agencies, think tanks, and universities.

Major powers (U.S., China, Russia) are using the hunt for mobile missiles as an “exemplar” for integrating AI, cyber, data analytics, and other technologies into their kill chains. The hunt for mobile missiles serves as an innovation platform. The “hunt” mission, therefore, has a military purpose — i.e., to destroy enemy missiles. It also has broader strategic purposes. It is a platform to organize additional advanced digital technologies around a clear mission. 

There are two drivers behind the hunt for mobile missiles. One is operational: to locate and destroy mobile missiles that could cause harm. The other is not to fall behind a rival in bringing digital technologies like AI into defense. This second dynamic is driving the arms race that is now taking shape among major powers. No great power wishes to fall behind in using advanced technologies for defense.

The mobile missiles of highest urgency are, clearly, those with nuclear warheads. Since the 1990s, mobile nuclear missiles have become the preferred nuclear weapon for nearly all countries, the United States excepted. The reason for this shift to mobile missiles is that fixed-site targets are vulnerable to precision conventional or nuclear attack, as has been demonstrated repeatedly in recent wars.

High touch” reconnaissance strategies will become a focus of technology strategy. High touch is defined as frequent, stealthy, tailored contact between reconnaissance systems and a target. It is a recurrent, surreptitious monitoring — the target is “touched” repeatedly by many types of sensors (drone video, cyber hacks, satellites, insider reports, etc.) that are continuous and unobtrusive. This high touch world of today contrasts with the Cold War, which had “low touch reconnaissance.” Satellite passes occurred intermittently, aerial reconnaissance was cumbersome, and insider agent reports were infrequent.

The combination of multiple “touchpoints” of different sensors – drone video, cell phone tracking, security cameras, hacked computers, etc. – will lead to greatly improved tracking of mobile targets. AI will be necessary to “aim” this complex reconnaissance system, extract target information from it, and link it to hypersonic missiles and other weapons in an overall value chain. Any one reconnaissance technology by itself (e.g., satellites or drones) will not deliver the required breakthrough.

An AI arms race among the major powers could upset the global nuclear balance for this reason. One way this could happen is from improvements in tracking the mobile missiles of another major power (e.g., the United States on China). Combined with massive cyber attack, follow-on nuclear and conventional strikes, a plausible first-strike threat is returning in a way not seen since the 1980s. This theoretical possibility could drive an AI-nuclear arms race. In the 1980s, the United States combined several “new” technologies (MIRVs, improved missile accuracy, SDI, ASW) to create a theoretical first-strike capability that created paranoia, and dangerous nuclear operating practices in Moscow.

An AI-nuclear arms race does not have to lead to actual war to impact international order in a significant way. Analogy with the 1980s shows this. It could produce heightened insecurity, loosening of the nuclear trigger, paranoia, hypervigilance, and nuclear groupthink. Another analogy with the 1980s stems from the synergy of several technologies, not just one.

A doubling of China’s strategic nuclear forces over the next decade is forecast in a recent DIA estimate (May 2019).[4] This may reflect Beijing’s understanding of just how effective new AI-directed search technologies are, with a recognition that Beijing’s nuclear deterrent will become vulnerable as a result. China has been a leader in all of the advanced technologies discussed in this report. The issues in this project, then, are already having an impact on how the global nuclear balance develops. For another example, STRATCOM has been given overall authority to remodel the U.S. nuclear command, control, and communications system. Discussions with STRATCOM, CYBERCOM, and British nuclear planners indicate a very high level of concern about the vulnerabilities created by reconnaissance tracking and cyber attacks.

Some secondary nuclear powers (Pakistan, North Korea) are slow to appreciate the threat to their “small” nuclear forces posed by improved tracking of mobile missiles. They have also failed to understand how their actions in a crisis – like full dispersal of their mobile missiles – could tip a confrontation over the edge into nuclear war. At some point however, they will understand the danger they face from technologically advanced reconnaissance of the United States and China, and perhaps of South Korea.

None of the major or secondary powers, have seriously analyzed the long-term consequences of the hunt for mobile missiles beyond the operational level. The longer-term impact on nuclear stability, arms control, escalation, early warning, and accidental war have gone largely unexamined. The present focus is on “not falling behind.” A similar pattern characterized Cold War nuclear dynamics. In the 1950s, the United States did everything it could to build up its nuclear forces. But, by the mid-1960s, restraint, arms control, and détente became new U.S. goals. Arms competitions have ups and down. The goals change. Today it would be a good idea to emphasize the long-term hazards of the arms race simply to bring this issue forward by several years, rather than waiting for this “turn” to develop or be discovered on its own. This is the personal view of the author.

The arms control regime created during the Cold War can no longer guarantee strategic stability. Advanced technologies, such as cyber, AI, and hypersonic missiles, will alter the global nuclear balance from what it is today.

Methodological Issues

Progress in tracking mobile missiles is likely to be rapid because the underlying technologies are from commercial innovation. DoD innovation, so to speak, has “sped up,” and the locus of innovation in U.S. defense is now in small and medium-sized enterprises (SMEs) rather than in big defense companies and in-house government laboratories. 

There is a systematic bias in the “legacy” U.S. defense innovation system (the big defense companies, the intelligence community, DoD, in-house government laboratories) to understate the impact of technology on defense, especially of other countries. China’s rapid increase in military ability, North Korea’s nuclear missile program, were overlooked for years, in part, because of this behavioral bias.

Tracking mobile missiles is an example of “spin-on” innovation. The underlying technologies originated in the mass market (Apple, Facebook, Google, Uber, etc.). It is only in the last fifteen years that DoD financial backing has tried to systematically spin on this technology into the defense sector.

Some new aids to thought are badly needed to break out of what has become a stale academic treatment of deterrence and nuclear weapons. A useful set of tools for this comes from management, and from the companies who’ve spent billions of dollars on advanced technologies like AVs (autonomous vehicles), 5G, AI, machine learning, vehicle tracking, and data analytics. These approaches are taught in business schools to analyze technology and corporate strategy. They have not been previously used in strategic studies or political science.

It is very important to think in terms of “technology packages,” instead of individual technologies. This is how business develops technology. The packages integrate several technologies into a coherent system. Uber, for example, integrates three separate digital platforms into a single, seamless, integrated package: a communications system to connect customers with rides (the cell phone network), a map direction system (Google maps), and a payments system (PayPal, credit cards). The hunt for mobile missiles is likely to depend on such technology packages, too, rather than breakthroughs in any one technology, like super satellites that “see” everything.

There is often far too much focus on the specifics of a new technology, without sufficient attention given to its overall impact on a strategic posture or on its arms race consequences. Consideration of technology packages, rather than specific technologies, is one way to see these larger impacts.

Value chains, touchpoints, innovation platforms, and information chains (discussed later in the report) are four management frameworks used for analyzing technology packages in business. 

All countries have value chains, in the same way that all have “organizations.” It would be impossible to deploy a military of any complexity without them. North Korea’s or Israel’s mobile missiles, for example, can be analyzed in terms of touchpoints and value chains – regardless of whether they conceptualize them in these terms or not.

Technological competition between countries is best thought of as a contest between value chains, rather than between technologies per se. The United States is building a value chain to hunt North Korean missiles. North Korea, in turn, builds a value chain to hide mobile missiles. The rivalry is between these two value chains. Moreover, there is a coevolution between these two systems. Simply counting North Korea’s nuclear arsenal — or American missiles — misses the key dynamics shaping the evolving rivalry. This competition between value chains describes other technology intensive rivalries as well, such as the U.S.-China contest in the western Pacific, India vs. Pakistan, and Israel vs. Iran in missiles.

Operational Issues

In the Cold War and until quite recently, limitations of reconnaissance determined targeting, for both conventional and nuclear weapons. There was almost no way to kill moving targets short of massive barrage attacks.

Today, new reconnaissance technologies overcome many of these limits. New initiatives in cyber further offer ways to disrupt command and control.

The ability to track mobile missiles uses computer algorithms, vast data centers, cloud computing, and deep learning. The “work” of finding missiles is done by secretive organizations. A satellite picture of the physical plants tells one nothing about what is going on inside. The situation was different in the Cold War. With difficulty, satellites could count enemy missiles. This “counting” of the threat is much more difficult today.

There are ways to penetrate this secretive world, however. Insiders, turncoats, and cyber espionage can to an extent assess capabilities. This is one reason that the “insider threat” receives emphasis in this report.

Insider attack of mobile missiles and command and control by agents, turncoats, special forces, saboteurs, and IT department staffers raise the insider threat to an altogether new level. The potential of “insiders with a flash drive” to wreak damage has been highlighted by reports that the Stuxnet virus, which infected Iran’s centrifuge enrichment control system in the late 2000s, was implanted by an insider employee working for a foreign intelligence service. These and other attacks could cripple a mobile missile force (e.g., by disrupting its command and control or by interfering with locks on atomic weapons).

A full dispersal of mobile missiles from their peacetime locations is an extraordinary, dangerous threshold to cross. It has never taken place in any country. Full dispersal intensifies a crisis and leads to hypervigilance in the enemy and the region. It could provoke preemption, by a rival, or conceivably by major powers, like the United States or China.

But raising the danger of war is one reason for missile dispersal. It shows a willingness to risk war to bolster one’s bargaining position. It signals resolve. It could also signal irrationality — in a sense providing a rationality for irrationality. Missile dispersal is like the U.S. nuclear alerts of the Cold War. But the differences with the Cold War alerts are important to underscore. The United States did not know where Soviet missiles were located in 1962, especially the Soviet IRBMs in Europe. Alerts in the future may increase crisis instability, as a major power may be able to locate enemy missiles precisely, and attack them with conventional counterforce strikes. It is not hard to write nuclear escalation scenarios from this situation.

Dispersal of mobile missiles as a signaling tactic has not had nearly the attention it deserves. Its salience as a “nuclear head game” does not stand out in military plans or academic studies. A decision to disperse missiles would fundamentally transform a crisis, making a confrontation far more dangerous. It will produce a political shock effect for the nations involved, and for most other countries. It could, for example, lead to ripple alerts that spill over to other countries outside of the region. 

Dispersal of missiles is one example of a larger set of unrecognized nuclear dynamics. These are thresholds which, if crossed, change a crisis to make it far more dangerous – yet whose significance goes unrecognized or is overlooked in peacetime plans and studies. There were several of these in the Cold War. In the future, examples include massive counterforce cyber attack of electrical power systems and telecommunications, blinding of satellites, and others. The “mating” of atomic warheads to dispersed missiles is another such threshold. It will have extraordinary political impacts, and is increasingly likely to be detected by the advanced collection technologies.

One key problem in the future is the use of missile dispersal for political signaling. This could be the source of complex or complicated “nuclear head games.” Partial dispersal might be limited, or might involve only a handful of missiles, or unarmed missiles. These tactics need careful study. Otherwise they will suddenly “appear” in some future crisis. No president or White House staff (NSC, CIA) can be expected to make sense of them in the time pressured conditions of a nuclear crisis.

In the Cold War, such nuclear head games developed into a high art form. Crises in Berlin, Cuba, and elsewhere saw movement of U.S. nuclear weapons to signal rivals, (e.g., B-52 airborne alerts, dispersal of NATO tactical nuclear weapons form storage igloos). President Richard Nixon had the famous “Madman nuclear alert” to signal Beijing and Moscow to stand down against his bombing of Hanoi. Dispersal of mobile missiles by Pakistan, Israel, North Korea, or Iran are like this Cold War practice. They might well be intended to be detected by the enemy, and allies, as signals. They are more likely than actual strikes, yet receive far less attention.

Dispersal of mobile missiles for political signaling shows something else that is very important: a country does not require a sophisticated technological force in order to have a very sophisticated political strategy for nuclear weapons. The political strategy may be designed for spoofing, and keeping the pot boiling, rather than for deterrence. This is particularly true for secondary, smaller powers (Israel, Iran, North Korea, Pakistan), who need to manipulate major powers to come to their assistance when they get in trouble. Moving mobile missiles around, “noisy alerts,” are an ideal way to do this.

The hunt for mobile missiles undermines the deterrents of the second-tier nuclear powers. New reconnaissance technologies make them vulnerable to conventional precision strike (conventional counterforce) from armed drones, hypersonic missiles, and stealth aircraft like the F-35. The mere perception that their relatively small nuclear forces are not going to be survivable could have far reaching consequences. 

Countermeasures to the hunt for mobile missiles could involve dangerous or highly undesirable developments. The most likely counter is to deploy more nuclear missiles. Other countermeasures include AI-generated “pictures” of different data sets (voice, images, text, intercepts) to create a deceptive picture of a missile force’s position and readiness. Generative Adversarial Networks (GANs) are an AI method to tweak the probabilities of detection and movement to accept a falsified picture of a situation.

Arms control efforts to “cap” second-tier nuclear states’ arsenals at levels of minimum deterrence will likely be an early victim of major power reconnaissance technology improvement.

Efforts to “fool” missile tracking systems will be a feature of the new arms race. Deception, AI-doctored images, decoy missiles, and virtual electronic missiles are a few of the many possibilities. GAN technologies will have especially important roles here.

Advanced technologies offer unprecedented ability to track individuals in key military units, including senior officers, their staffs, missile crews, atomic weapon protection guards, and political leaders. Their location and movement are an extraordinary way to gain intelligence and warning. Tracking people could provide insight into changes in alert levels and intentions.

Predictive analytics may be used to estimate the future state of a nuclear force. It could be used to predict the next location of a missile, and the location after next. Analytics could also distinguish routine from non-routine missile and warhead movements (e.g., by whether key senior commanders were present, or whether atomic warhead units were close or distant from the missiles).

The main way to locate mobile missiles comes from the synergy of combining technologies: phone hacks, drone video, cyber penetration, spies, communication intercepts, facial recognition, hacked security cameras, and new kinds of radar. The vast amount of data generated from these collectors requires AI, cloud computing, edge computing, and data analytics to process and display it. It also requires real-time information chains linked to quick-reaction alert (QRA) weapons like hypersonic missiles, armed drones, F-35s, and other systems.

No comments: