North Korea’s recent and dramatic tests of long-range missiles have created a sense of urgency and vulnerability in the United States, leading to renewed calls for expanding missile defenses. The administration and Congress have approved huge funding increases for existing systems, and call for developing new types of defenses—potentially including interceptors in space.

Is this the answer? How should one think about missile defense: as a protective shield or a dangerous illusion?

Missile defenses have as long a history as missiles do, and in the late 1960s American and Soviet scientists came to believe that a defense against long-range missiles would never be effective because the other country would simply build more weapons to defeat it, leading to a dangerous arms race. The 1972 Anti-Ballistic Missile (ABM) Treaty, which placed strict limits on U.S. and Soviet/Russian strategic missile defenses, reflected that understanding.

President Reagan’s 1983 “Star Wars” speech challenged that idea by calling for the United States to develop a large defensive system that included orbiting interceptors. Recognized by most experts as unworkable, this expansive system was pared down over the next decade and finally shelved, although work continued on interceptor technology during the Clinton administration.

Then, in 2002, President George W. Bush abandoned the logic of the ABM Treaty, by withdrawing from it and announcing that the United States would field the first interceptors of a new Ground-based Midcourse Defense (GMD) in less than two years. To do so, the administration exempted its development from the strict “fly-before-you-buy” rules that govern all other large Pentagon projects—a step that has had dire and long-lasting consequences.

GMD remains the sole system designed to counter intercontinental ballistic missiles. Its 44 silo-based interceptors in Alaska and California are designed to be guided by space, ground and sea-based sensors to collide with an incoming warhead and destroy it with the force of impact.

Reflecting the difficulty of the task, and the haste and lack of rigor of its development, the GMD system today has a very poor test record, even though these tests were “scripted for success” according to former Pentagon head testing official Phil Coyle.

The problems are well documented. Only about half of the 18 intercept tests since 1999 successfully destroyed their targets, and the test record has not improved with time: only two of the last five tests were successful—and GMD has still has not been tested under operationally realistic conditions. Thus, there is no evidence that the $40 billion GMD system provides a reliable defense, even against a country like North Korea.

More fundamentally, even if the reliability is improved, GMD’s prospects for providing an effective defense in the future are poor because it will face countermeasures that any country that has developed a long-range missile and a nuclear warhead could readily use to confuse or overwhelm the system.

Despite these problems, however, the administration and Congress plan to expand the system; the current budget includes funding to build 20 additional interceptors.

Given North Korea’s pursuit of a nuclear-armed long-range missile, it seems reasonable to ask whether something isn’t better than nothing. That sounds plausible, but does not hold up upon closer examination. The unconstrained pursuit of missile defenses can, perhaps counterintuitively, create even greater risks.

For example, a belief that missile defense works better than it does can lead political and military leaders to adopt a more aggressive foreign policy and take more risks. U.S. officials regularly describe the system as much more capable than it has been demonstrated to be. Even President Trump stated on television last October that “We have missiles that can knock out a missile in the air 97 percent of the time.” Yet the testing data show there is no basis to expect interceptors to work more than 40 to 50 percent of the time even under the most generous and easiest conditions.

Using multiple interceptors against each target can improve these odds, but it does not fundamentally change the situation; the chance of a nuclear weapon getting through would still be dangerously high. Consider an attack with five missiles. Using four interceptors against each target, each with a kill probability of 50 percent, the odds that one warhead gets through are 28 percent—or higher, if the failure modes are not independent of each other (for example, if the guidance systems of all the interceptors are faulty in the same way).

Overestimating defense effectiveness could increase policymaker support for a pre-emptive attack against North Korea, which might then fire missiles in retaliation. It would then become clear that the system could not stop those missiles.

Missile defenses can also increase nuclear risks by blocking arms control and providing incentives for Russia and China to build more and different kinds of weapons; preventing this dynamic was a core reason for the ABM Treaty’s limits. Russia and China worry the United States may come to believe it could launch a first strike without fear of retaliation because it could shoot down any surviving missiles. This fear is exacerbated by U.S. development of conventional “counterforce” weapons that can attack Chinese and Russian nuclear weapon systems.

These concerns are not theoretical. Russia has repeatedly stated that any future arms control agreements must include limits on missile defenses, and says expansion of U.S. defenses could lead it to withdraw from the New START treaty. And on March 1, Russian President Vladimir Putin announced plans to field several new nuclear systems that could avoid U.S. missile defenses, including a nuclear-powered nuclear-armed cruise missiles and underwater drones.

China has begun to build more long-range missiles, develop hypersonic weapons and deploy multiple warheads on its missiles, and has also discussed putting its missiles on high alert. At worst, U.S. defenses are driving developments that result in more threats and risks; at best they are providing justifications for them. The irony is that they do not provide an effective defense in any case.

Unfortunately, things are on a path to get worse. The United States is developing a ship-based interceptor that in theory could intercept strategic missiles and plans to field hundreds of them in the coming years. An influential minority in Congress has been calling for space-based missile defenses, with plans for a “space test bed” that would put dedicated weapons in orbit for the first time. Chinese and Russian military planners will not ignore these developments.

As long as nuclear-armed countries continue to believe their security relies on the ability to retaliate with nuclear weapons, missile defenses will interfere with efforts to reduce—and eventually eliminate—these weapons. Given the inherent problems with building reliable and effective missile defenses, these defenses are more a dangerous illusion than a realistic solution.