Let's say you're buying the most complicated computer system ever devised. It's still in the early stages. The payments are costing a fortune. The software's riddled with bugs. Some software hasn't been written yet. Several scientists doubt the thing will ever work properly. Finally, just this week, you couldn't even get it to switch on.
Now let's say you're the program manager of the Pentagon's missile-defense agency.
But I repeat myself.
In the wee hours of Wednesday morning, the MDA attempted the first full test of the missile-defense system in two years—or at least the first full test of the few components built so far. A mock warhead was launched from Kodiak, Alaska. It was to be intercepted—smashed and destroyed—in outer space by an antimissile missile launched from the Marshall Islands in the Pacific.
But 23 seconds before the interceptor missile was scheduled to blast off, the system shut down, due to what the MDA is calling "an unknown anomaly." As of Thursday afternoon, Pentagon officials were saying—both publicly and off the record—that they still don't know what happened.
MDA officials have been trained to appear upbeat, no matter how catastrophic the news. This time, though, the effort was particularly heroic. "I definitely wouldn't categorize it as a setback of any kind," chief spokesman Richard Lehner said of the test. In fact, he added, it was "a very good training exercise." The program's engineers learned "quite a bit." No doubt, the Army is learning quite a bit from the insurgency in Mosul, too.
A question for the overseers in Congress: Is it time now for a serious look at this program? Missile defense consumed $10.7 billion of this year's military budget—far more than any other weapon system. About $80 billion has been spent on it since Ronald Reagan stepped up research and development for the mission 20 years ago. Another $80 billion is scheduled to be spent before the decade is out. It may be time to ask: Why?
The last time the MDA tried this test, in December 2002, the rocket got off the ground, but the interceptor—also known as the "kill vehicle"—didn't separate from the rocket booster. Eight other tests have taken place in the past few years. In five of them, the kill vehicle did hit the mock warhead, but they were, by the MDA's acknowledgement, carefully controlled tests; for instance, the warhead's radar signature was "enhanced" to make it more visible. They were tests of the basic concept—can a missile hit another missile? (yes, it can)—not tests of the actual technology.
Contrary to the MDA's straight-faced assurances, Wednesday's no-go was a serious setback. This was going to be the test that proved the program was making progress. It was the first time the actual rocket booster, which a full-blown system might someday use, was to be tested. More than that, it was the first time these elements of the program had been tested since the Pentagon started to deploy antimissile missiles. Yes, at President Bush's insistence, the first six operational missile-defense weapons were deployed at Fort Greely, Alaska, last fall—even though testing is far from complete. Many scientists and the Pentagon's own test director had criticized the deployment as premature. By long-established practice, weapons are supposed to be put in the field only after testing has proved their worth. Wednesday's test was supposed to dampen this criticism; it was going to show the program was on track, despite its unorthodox procedures.
Instead, it was a flop. It was a flop on a low-tech level—how many times has the United States launched a rocket? And nobody yet knows what went wrong.
There is no need to repeat here the dozens of reasons for skepticism that an antiballistic-missile system has much chance of shooting down a single enemy warhead. (For some of those reasons, click here.) If it can shoot down one warhead (a lucky roll of the dice), the bad guys can simply launch a second warhead—and there hasn't yet been even a rigged test involving multiple targets. Everything about the system is way too complicated—the software; the command-control network; the integration of early warning radars, target-acquisition sensors, and weapons-launch centers. Yes, landing on the moon was complicated, too (to use an example cited by many advocates), but that was child's play by comparison. For one thing, the moon landing was a one-sided enterprise. As the spacecraft approached the lunar surface, the moon didn't suddenly shift direction or turn into a mirage. By contrast, an enemy can easily load a missile with decoys, which can lure an interceptor to the wrong target. Also, the trip to the moon took days; if something went wrong, corrections could be made. The trip to an enemy warhead darting across the heavens at 15 times the speed of sound must be completed in a half-hour or less, everything must be automated (there's no time for human intervention), and nothing can go wrong at all.
But Wednesday's test tells us that we are a long, long way from having to discuss the system and its problems at this level of detail. We can't even count on the rocket getting out of its launch silo, much less the millions of minute operations that must follow. President Bush fielded a half-dozen antimissile missiles and called them "operational." But they're a ruse. The Pentagon's test director, Thomas Christie (a veteran missile engineer and lifelong civil servant who, alas, is retiring next month) has testified repeatedly that the program is not yet ready for deployment, not yet ready to be called "operational."
The Los Angeles Times today recalls Secretary of Defense Donald Rumsfeld appearing on CBS's Face the Nation shortly after the Bush administration took office. Rumsfeld, a longtime enthusiast for missile defense, was asked about critics who charged that the program couldn't work. "We have no intention of deploying something that doesn't work," the secretary replied, "but what the definition of 'work' is, is terrifically important."
A minimal definition might be: You can turn the thing on.