Earlier this year in a high-stakes summit meeting, the leader of North Korea, Kim Jong-un, and U.S. president Donald Trump failed to agree on a way to end North Korea’s missile and nuclear weapons program. Within days of the summit, satellite imagery showed that Kim was rebuilding some of North Korea’s rocket facilities.

In 2017 Kim’s government had tested its first long-range missiles and soon followed up with a test of what appeared to be a powerful hydrogen bomb. After those operations, Trump told Americans not to worry. “We build the greatest military equipment in the world,” he said on the Fox News television channel. “We have missiles that can knock out a missile in the air 97 percent of the time. If you send two of them, they are going to get knocked down.”

The president was expressing extraordinary faith in the Ground-based Midcourse Defense (GMD) system, currently the country’s sole guard against intercontinental ballistic missiles that carry nuclear warheads. But his faith was woefully misplaced, and facts belie his claim. U.S. military testing data show there is no basis to expect GMD interceptors to work more than about 50 percent of the time. Using multiple interceptors against each target could in some cases improve these odds but will not fundamentally change the situation. The chances of a nuclear weapon getting through in a real-world attack using multiple missiles is still dangerously high. Our analysis of all 19 tests done—the most recent one was in late March—as well as several government reviews of the program, shows an alarmingly high failure rate.

The problems with these interceptors can be traced back decades, to the beginnings of the program, because the Pentagon rushed to develop it and abandoned tried-and-true oversight and testing requirements that have been used for most major weapons. Plus, the system is vulnerable to defense-penetrating countermeasures that any country capable of building a long-range missile could take. Yet the Pentagon is moving to expand the troubled interceptor fleet despite already spending more than $40 billion to produce these unreliable results.

The arms race

The world has been living under the threat of nuclear-armed intercontinental ballistic missiles (ICBMs) since the late 1950s. Two nuclear-armed nations, the former Soviet Union and the U.S., were desperate to deter each other from using these weapons. Both built more missiles, armed with multiple warheads, to overwhelm any response the other country might be able to mount. This cycle of action and reaction resulted in the rapidly growing potential for mass destruction.

Eventually U.S. and Soviet leaders recognized the dangers of this tit for tat, and in 1972 they signed the first Strategic Arms Limitation Treaty (SALT I). They also agreed to the Anti-Ballistic Missile (ABM) Treaty, which strictly limited defenses against long-range missiles and served to break the cycle of defensive advances prompting the other side to improve its offense. These arms-control agreements and those that followed did the job. The breakneck growth of the U.S. and Soviet nuclear arsenals peaked in 1986 at more than 60,000 weapons, and three decades of arms reductions have brought that number down to fewer than 10,000 today.

While the stockpiles were still high, in 1983, President Ronald Reagan—driven by distrust of the Soviets and faith in new technologies—tried to revitalize missile defense and announced the Strategic Defense Initiative, or “Star Wars.” His vision was based on nonexistent and impractical innovations, such as space-based lasers powered by nuclear explosions, and after a while plans for the system were scrapped.

But research on strategic missile defense technology continued as new threats emerged. North Korea’s pursuit of nuclear weapons and ballistic missiles, which became apparent in the 1990s, provided a fresh argument for strategic defenses that took advantage of improvements in missile tracking and interceptor guidance. Still, limitations of the technology, skepticism about the magnitude of the threat, and concerns that deploying defenses would threaten the successful and ongoing arms-reduction process led the Clinton administration to respect the ABM Treaty limits.

Then, on September 11, 2001, everything changed. In the political environment following the attacks on the World Trade Center in New York City and the start of the “war on terror,” opponents of missile defense found it difficult to argue against any military programs. Congress could muster only limited debate on arms control. In late 2001, citing a potentially growing threat from rogue nations and terrorism, the George W. Bush administration announced that the U.S. would withdraw from the ABM Treaty. Then the government said it would rush ahead with a plan to build a missile defense system. The era of negotiated limits on missile defenses was over.

The incoming threat

Destroying an ICBM’s warhead is not easy. These missiles launch thousands of kilometers from their targets, accelerated by powerful engines in what is called their boost phase. Within minutes they reach speeds of about 25,000 kilometers per hour and then can release multiple warheads that arc through the vacuum of space. After about half an hour of this midcourse flight, they reenter the atmosphere and drop toward their targets in the terminal flight phase, which lasts only a few minutes.

Defense efforts have focused on the midcourse period because it lasts much longer than the boost phase, and by intercepting at long distances from the targets, the system can defend much larger areas than it could at the terminal stage. The earliest interceptors carried their own nuclear weapons to blow up the incoming warhead. But in the late 1970s development began on prototype interceptors that carried a nonexplosive “kill vehicle.” Onboard sensors were supposed to guide the vehicle into the incoming warhead during its midcourse phase. At a collision speed of 10 kilometers per second, the kinetic energy per mass is more than 10 times the energy released by a similar amount of high explosives, so such impacts could destroy warheads in a direct hit and avoid the use of a nuclear detonation for defense. This “hit to kill” method requires sophisticated technology. The kill vehicles must be guided to within centimeters of a precise target point on the incoming missile warhead.

Credit: Tavis Coburn (illustration); Penelope Breese Getty Images (Ronald Reagan); David Ake Getty Images (George H. W. Bush and Boris Yeltsin); Konstantin Zavrazhin Getty Images (George W. Bush and Vladimir Putin); U.S. Missile Defense Agency (interceptor); Getty Images (Kim Jong-un)

Bush’s plan was to get a system into the field quickly and then improve it. In September 2004 the administration stated that the system had achieved a “limited deployment option,” which meant it could be turned on and used if necessary. Only five interceptors were in place the day of that announcement.

Today the GMD comprises space-based sensors, terrestrial radars, 44 interceptors based in Alaska and California, and facilities and personnel to control operations. The Department of Defense’s current plan is to increase the number of interceptors to 64 by 2023 and possibly add more soon thereafter to reach a total of 100.

Holes in the defense

The push to deploy this system, however, has produced serious shortcomings, and the GMD has yet to demonstrate a useful military capability. The roots of the problems lie both in the shortcuts the government took to move the program forward and in the technical complexity of missile defense.

In 2002 Bush’s DOD exempted the program from the Pentagon’s traditional “fly before you buy” oversight rules, intended to make sure major defense systems and equipment work well before the nation has to depend on them. Under those rules, the GMD system would need to meet criteria for technical maturity and effectiveness and to undergo demanding operational tests to ensure that it worked as required under real-world conditions before being put in the field. But Bush’s exemption meant that prototype interceptors for research and development—by definition not intended for the real world—could be used in urgent situations. While faster, this approach permits the use of unreliable or poorly tested equipment.

In 2014—10 years after the limited deployment announcement—all GMD interceptors in the field were put there before the Pentagon had conducted a single successful intercept test of their design, according to reviews by the DOD and Congress. Ideally, a rigorous engineering process identifies problems early and allows them to be fixed before deployment. But with the GMD system, failed intercept tests revealed design flaws that required expensive retrofits of dozens of interceptors already in silos. Because the interceptors were being fielded as they were being tested, hardware and software components and designs vary from interceptor to interceptor, making it difficult to use the performance of one to predict that of another or to resolve problems across the entire fleet.

GMD interceptors have destroyed their targets in just more than half the 19 intercept tests conducted. The record is not improving with time. Six of the 11 tests since 2004 have failed to destroy their target. Of the most recent six tests, three have failed.

A number of these failures were attributed to lapses in quality control, according to the Missile Defense Agency, the Pentagon office that runs the program. The agency stated in 2007 that poor manufacturing and setup procedures by its contractors—which it attributed to the streamlining of the acquisition process and schedule pressures—had caused “test failures and slowed production.” A failed $236-million intercept test in January 2010 was attributed in part to a small device called a lockwire, which Raytheon, the contractor that builds the kill vehicles, did not install. A report by the DOD’s inspector general, following that mistake, found many other serious quality-management problems.

These quality-control troubles can slow progress by masking other flaws that tests are supposed to uncover. For example, the January 2010 test was repeated later that year and failed again, but that time the trouble was attributed to a design flaw: vibrations from the rocket motors the kill vehicle uses to change direction could cause errors in the guidance system. This design issue might have been identified earlier if the missing lockwire had not derailed the earlier test. Identifying the bigger problem and fixing the interceptors that had already been put in the field eventually cost nearly $2 billion.

Another disturbing aspect of the high failure rate is that it has occurred in highly simplified tests that do not resemble situations an interceptor would face against an actual enemy. No GMD test, for instance, has involved an incoming missile that used countermeasures such as realistic decoys. Incoming weapons can carry numerous decoys that appear very similar to warheads; the GMD must find the real warhead among the fakes. But tests have deliberately used decoys that appear very different from the actual mock warhead, making the interceptor’s job artificially easy.

“If we can’t discriminate what the real threatening objects are, it doesn’t matter how many [ground-based interceptors] we have. We won’t be able to hit what needs to be hit,” Michael Gilmore, then director of Operational Test and Evaluation for the Pentagon, told Congress in 2013.

The poor test record of the GMD system stands in stark contrast to repeated statements by U.S. military and political officials over the years that give an inaccurately optimistic appraisal of the system. For example, in congressional testimony in April 2016, Admiral Bill Gortney, then commander of the North American Aerospace Defense Command and U.S. Northern Command, said, “We are prepared to engage and protect Hawaii, Alaska and all the rest of the states with the existing system and have high confidence in its success.” In fact, the system has not demonstrated capability under real-world conditions. The most recent test, in March, was the first one that the Pentagon actually described as operational rather than developmental. The agency said the interceptors (it fired two) destroyed the target, but it has not released enough information about the test to permit an independent evaluation of the test conditions.

Ineffective deterrence

Even while acknowledging the GMD’s limits, some contend that any capability is better than none. This argument, however, has serious flaws.

The 2019 Missile Defense Review asserts that a missile defense system such as the GMD helps to deter a missile attack by increasing an adversary’s uncertainty. The attacker might doubt its ability to destroy enough U.S. forces to avoid a retaliatory strike, for instance. But such doubts are already in place: U.S. retaliation is assured by nuclear forces safely hidden on submarines at sea. And this logic certainly does not apply to an adversary whose intent is not to target U.S. retaliatory capability in the first place, such as North Korea or even China. Those nations’ missile arsenals are too small and inaccurate to mount an effective strike against U.S. nuclear forces. Instead they would target cities or other large, unprotected sites. Therefore, missile defense is unlikely to offer anything that adds to the deterrence currently provided by U.S. forces.

The Missile Defense Review also argues that the U.S. needs a defensive shield so an adversary’s missile threats cannot force this nation away from taking military actions in its own interests or on behalf of an ally. But to make U.S. decision makers confident enough to ignore adversarial threats, a system such as the GMD needs to demonstrate high effectiveness, and it has not.

At its core, missile defense is meant to defeat a nuclear attack if deterrence were to fail. While this is where the “some is better than none” argument is the most persuasive, for any realistic scenario, missile defense will likely do very little. Even if the system’s ability to deal with real-world complexities such as countermeasures were greatly improved, a nuclear attack will still present enormous risks. For example, if the system achieved an improbably high 95 percent effectiveness against one missile, in an attack by just five missiles there is still a one-in-four chance of at least one nuclear warhead penetrating the defense. The likelihood of a city being destroyed would be higher than correctly predicting the roll of a die. The effectiveness against a real attack is likely to be much lower.

What missile defenses may actually do is get in the way of reducing the nuclear threat faced by the U.S. or even increase it. As long as nations such as Russia and China continue to rely on strategic arsenals for deterrence, pursuing defenses that appear to threaten that deterrent—or that lay the groundwork for a system that may threaten it in the future—will at best hinder nuclear reductions. At worst these efforts will lead to the growth of more offensive weapons designed to overwhelm the defense and reduce stability by increasing the incentive to launch missiles first in a crisis.

There is growing evidence that global powers are already returning to this type of brinkmanship, which the ABM Treaty sought to quell. As part of its ratification process for the 2011 New START arms-control treaty with the U.S., Russia stipulated that further cuts to its arsenal would require limitations on strategic defenses. More recently, President Vladimir Putin announced that Russia is developing several new strategic nuclear delivery systems that are designed specifically to defeat or evade U.S. missile defense systems. These include a nuclear-capable hypersonic weapon that could fly undetected by current sensors and a drone submarine that could carry nuclear weapons designed to destroy U.S. coastal cities.

China, for its part, recently added multiple nuclear warheads to its large ballistic missiles—a change the DOD, in its report to Congress on China’s military power, attributed in part to concerns about advances in U.S. strategic defenses.

Beyond the potential for missile defense to increase nuclear threats the U.S. faces is the real possibility that a false sense of security will distort U.S. decision-making. Misunderstanding the system’s capability and believing that missile defense is highly effective or even somewhat effective could lead U.S. leaders to take more risks in foreign policy. An unfounded faith in missile defense reduces incentives to pursue political solutions to national security problems and to improve nuclear arms control. Nuclear-armed missiles are a political problem that technology cannot solve.

Heightened risk

Although the modest size of the current GMD system somewhat limits its destabilizing potential, missile defense proponents are pushing to expand other U.S. missile defense capabilities. The navy’s Aegis ship-based missile defense system was developed to defend against short- and medium-range missiles in particular regions, but Congress has called for testing a new Aegis interceptor against an intercontinental-range missile, thus demonstrating its potential for strategic missile defense. Current plans call for deploying several hundred of these new interceptors on ships over the next two decades to establish a large, mobile, strategic missile defense capability that could be used around the world. Such a system is certain to cause concerns in Russia and China and is the kind of system the ABM Treaty was intended to stop.

The current defense budget also requests money to begin developing a space-based missile defense system designed to intercept long-range missiles right after launch, during their boost phase and before they can deploy countermeasures. These space-based systems would be enormously expensive yet vulnerable to attack and therefore ineffective—and highly destabilizing.

As with the decision to proceed with the GMD program, this missile defense expansion is taking place with very limited discussion and not enough assessment of the benefits and costs. The price is high. The total GMD cost is projected to reach at least $67 billion if the Pentagon fields 64 interceptors, according to a 2018 U.S. Government Accountability Office report. An “austere” space-based interceptor capability would require 650 satellites and cost upward of $300 billion, says a 2012 National Research Council report. Real resources are being spent on the illusion of a defense.

But there is another, even more important cost: our national security. Current U.S. missile defense plans are being driven largely by technology, politics and fear. As in the past, this is happening with insufficient understanding and consideration of the limited protection these systems can realistically provide. Missile defenses will not allow us to escape our vulnerability to nuclear weapons. Instead large-scale deployments will create barriers to taking real steps toward reducing nuclear risks—by blocking further cuts in nuclear arsenals and potentially spurring new deployments. This process of moving blindly and quickly ahead threatens to lead to a world filled with greater threats, not lesser ones.