A weathered gray building housing an obsolete radar looms over this tiny Pacific island as a reminder of how difficult it is to field an effective defense against missile attacks.
The Cold War relic was built as part of the Safeguard program deployed to protect U.S. nuclear missiles in North Dakota from a possible Soviet strike. Developed at a cost of $23 billion in today's dollars, Safeguard was operational for four months. It was shut down in February 1976 after a variety of holes were discovered in its would-be shield.
Twenty-eight years later, an unprecedented lack of testing has raised concerns about similar holes in a new U.S. defense against long-range missiles that is expected to go on alert before the end of the year.
Test launches of the system's interceptor missiles occasionally blast off from a man-made hilltop several hundred yards from the Safeguard radar. There haven't been any, however, since a failed attempt in December 2002. The next test isn't scheduled until late November at the earliest.
In fact, the United States is poised to activate one of the most complex weapons systems ever built after only eight attempts to intercept a missile, three of which were failures. In comparison, the mothballed Safeguard defense was declared operational after 70 intercept attempts, which included 58 successes.
Critics in Congress, the Government Accountability Office, the Pentagon and academia say it's folly to presume the interceptors already lowered into silos at Fort Greely, Alaska, will perform as advertised without more flight testing. They complain that the Bush administration hasn't independently or realistically tested the new system for fear of potential political embarrassment and the possibility of exposing vulnerabilities like those that killed Safeguard.
Even supporters acknowledge that the anti-missile system needs to demonstrate that it works -- and soon.
"What I'm afraid of is what happened to Safeguard," said Jim Hill, site manager for the program here at a U.S. Army facility on Kwajalein Atoll. "There is more pressure on this next test than any in recent memory."
Hill's concern is warranted. For the new anti-missile system to work, it must take guidance data from distant radar and orbiting satellites to slam a 55-inch kill vehicle traveling at 16,000|mph into a small enemy warhead -- essentially hitting a bullet with a bullet.
The kill vehicle's sophisticated sensors must be able to pick out the warhead amid rocket debris and possible decoys high above Earth's atmosphere. Everything, including the booster and the system's 5.5|million lines of computer code, must work perfectly for the defense to succeed.
Huge questions remain.
The kill vehicle and its new booster never have been mated and test-launched. A second planned booster won't be available before 2006. Several of the kill vehicle's components are new, including the upgraded software it will use to identify targets. None of these new features has been flight-tested. Neither has the latest version of the system's command-and-control software, which is still under development and has been checked out only in laboratory simulations.
Previous flight tests, which used a different booster and earlier versions of the kill vehicle, were largely rehearsed and unrealistic. They also were conducted without a critical X-band radar that is supposed to provide targeting data to the kill vehicle. Most of the system's testing has been done instead by giant supercomputers that run simulations to analyze software and hardware components.
In addition, no interceptor ever has been test-fired from Fort Greely. It's unlikely any will be because of concerns about the environment and launching over populated areas.
"This is like deploying a new military aircraft that doesn't have wings or a tail or a landing gear and without any testing to see if it will work," said Philip Coyle, director of the Pentagon's office of weapons testing and evaluation from 1994 to 2001. "It's completely unprecedented."
Now, some members of Congress are questioning whether the Missile Defense Agency has violated a federal law requiring the Pentagon to operationally test a weapons system before it enters full production, a process called "fly before you buy." On Oct. 9 lawmakers passed a defense-authorization bill that for the first time requires realistic testing of the system by Oct. 1, 2005.
Air Force Lt. Gen. Trey Obering, director of the Missile Defense Agency, concedes the systems are not being developed in the traditional way but insists the testing is thorough nevertheless.
"We are going through a very rigorous test program that includes a variety of ground tests, of integrated system tests and of flight tests," Obering said. He acknowledged, however, that "flight tests are what get all of the attention."
Lots of warnings
Criticism of the Missile Defense Agency's testing program is nothing new. Many test recommendations have been ignored since the Clinton administration.
In 1998, a panel headed by Air Force Gen. Larry Welch issued a report on reducing risk in missile-defense programs after a series of flight-test failures. One finding was that the rush to field missile defenses could result in "less-than-minimal testing." In a follow-up report in 1999, the panel noted that a Pentagon directive to keep open the option of an "emergency" deployment by 2003 "could lead to emphasis on near-term deployment readiness at the expense of properly completing the development activities."
A devastating report by Coyle's office in 2000 -- made public despite the Pentagon's objections the next year -- detailed the scripted, unrealistic test program for the new long-range missile defense. Four years later, with the system on the verge of being declared operational, little has changed.
In every intercept test so far, the mock warhead has carried a homing beacon to guide the kill vehicle to the target, simulating data that are supposed to come from the unfinished X-band radar. Each test also has been staged from the same locations, with target missiles launched from Vandenberg Air Force Base and interceptors lifting off from a U.S. Army facility in the Marshall Islands. That geography restricts the intercepts to low speeds and relatively short distances.
Testers knew in advance the target's location, launch time, trajectory and projected impact point, as well as what the mock warhead and any decoys would look like.
"I can look pretty smart if you give me the answers to a test in advance," said Coyle, now a senior adviser for the Center for Defense Information. "If you want to find what I really know, start asking me questions to which I haven't been told the answers. Missile defense is no different."
Despite the scripted nature of the tests, the most recent intercept attempt failed when the kill vehicle failed to separate from its booster. Nonetheless, government officials developing the system contend it is sound.
"The basic technology and designs have been proven in this test program," Obering said. "The failures that we've experienced have been quality-control failures, not basic-design problems."
In April, the GAO weighed in with a report pointedly titled "Missile Defense: Actions Are Needed to Enhance Testing and Accountability." It recommended that the Pentagon's office of Operational Test and Evaluation subject each component of the missile-defense system to independent and realistic testing, and then forward the results to the secretary of defense and Congress. The Defense Department declined.
In fact, the Bush administration has resisted all efforts to have testing done by an independent authority outside the Missile Defense Agency.
Federal law requires that weapons systems undergo independent operational testing before entering "full-rate production." Missile-defense proponents have evaded this requirement by claiming some parts of the overall system are still in development. While the interceptors in Alaska are on the verge of being deployed, other components, such as sophisticated new missile-tracking satellites, are still being designed.
Critics, including U.S. Sen. Carl Levin, D-Mich., contend the pieces clearly are separate systems. He considers the rationale a smoke screen designed to circumvent "fly before you buy" laws.
In a June 9 letter to acting Pentagon acquisition chief Michael Wynne, Levin charged that the system in Alaska violates the law because the interceptors have entered "full-rate production" without being operationally tested.
Levin noted that on June 4, a member of his staff visited an Orbital Sciences Corp. booster plant outside Phoenix and a Raytheon plant in Tucson, Ariz., where the kill vehicles are built. The companies were working two or more shifts a day to meet the administration's goal of deploying 30 interceptors by 2007. An Orlando Sentinel reporter visiting the facilities less than two weeks later found the same production pace.
According to Levin, three kill vehicles had been produced at Raytheon the previous month, compared with two in the previous year.
"They proceeded to full-rate production before operational testing was planned or occurred, and that is a violation of the law," Levin said in an interview. "If you are producing at full-rate capacity before you know if something is going to work, you are risking losing huge amounts of money."
Wynne, however, replied that the interceptors are part of a larger system and not a single element.
Congressional supporters of missile defense, such as U.S. Rep. Curt Weldon, R-Pa., insist complaints about testing from Levin and others are part of an ongoing effort to derail the system.
"They have denied the ability for the Missile Defense Agency to test all during the '90s by putting constraint after constraint on the testing," Weldon said. "Then when we move forward with the system, what is their biggest argument? 'You haven't tested it.'"
Congress finally agreed this month to order more testing using criteria set by Coyle's former office. The law doesn't specify any penalties, however, if the testing isn't done.
"This is the first time Congress has been willing to put in language saying the program must be tested against an outside standard," a congressional staffer said. "But this is locking the barn door after the horse is out. I can't think of a single other system that was deployed and then tested to see what it could do."
'Bang for the buck'
Much of the recent testing for the new system in Alaska has been done in Huntsville, Ala., at an Army-sponsored facility called the Advanced Research Center and at other labs in a Boeing plant near the city airport. There are no launches or midair collisions, just the quiet hum of giant supercomputers fighting battles in cyberspace.
An exact replica of the control room at Fort Greely, Alaska -- from the six stations at the horseshoe-shaped console to the gray-green color of the carpet -- is behind one door at the Advanced Research Center. The system's operators train there by killing imaginary threats.
Elsewhere, engineers run simulations to check for bugs in the computer software that runs the defense and test pieces of the system's hardware. "We have to test it before it gets deployed for real," said Bill Lokken, Boeing's manager of test resources for the missile defense being deployed in Alaska. "This is the heart of the program to make sure we have bang for the buck."
But many experts say the simulations are no substitute for flight tests. Besides proving whether the system actually works, flight tests provide critical data needed to make sure the simulations are accurately "anchored" to reality. The GAO's April report questioned whether enough flight tests were occurring to ensure the simulations are realistic.
Capabilities vs. requirements
There is growing concern that the root of the problem is a new Pentagon policy for fielding missile-defense systems that was mandated by the Bush administration in 2002. Under this so-called "capabilities-based" approach, new systems are rushed into the field as soon as they appear to be useful. Then, through a process dubbed "spiral development," they are continually refined using new engineering and feedback from soldiers. No longer must a system meet a list of hard-and-fast requirements before it is deployed.
The goal is to give the military more flexibility in meeting ever-changing needs and get systems on the battlefield more quickly.
"We're not saying that we are walking away from requirements," Obering said. "Oftentimes what happens is the world changes from the time you wrote the requirements to the time you begin to satisfy them... Often you end up delivering obsolete capabilities."
Opponents of this approach, including some in the Pentagon, say the result is flawed systems with limited capabilities being deployed sometimes years before they are ready.
An example cited by both sides is the Patriot missile system. Conceived in the 1960s as an anti-aircraft weapon, its new, added capability to shoot down short-range missiles was still being tested on the eve of Operation Desert Storm in 1991. The Pentagon decided to deploy the system against Iraqi Scud missiles under the belief that anything was better than no defense at all.
"You take the capabilities that you can get today and then go back and add capabilities," said Dave Hartman, a former Patriot battalion commander who is now one of the program's managers at manufacturer Raytheon. "That is what we've done since the initial fielding of the system, which was an aircraft-threat-based requirement, to today being able to handle tactical ballistic missiles."
Proponents say the system saved lives. But 1992 government studies determined that Patriot destroyed less than 10|percent of the Scuds it engaged -- an assessment Raytheon disputes.
Skeptics who oppose rushing to deploy the interceptors in Alaska warn that Patriot's past performance shows why it's a huge mistake to rely on any defense before it is thoroughly tested.
Proponents counter that it's time to deploy the system -- whatever the capabilities are -- and then improve it.
"We've delayed a long time in missile defense because of the concept of researching forever and building never," said U.S. Sen. Jon Kyl, R-Ariz. "That's why I think the president's idea of getting started and learning as we proceed is a good program."
The fact that missile-defense programs operate under a different set of rules rankles many in Congress. The special treatment could doom some systems to the same fate as Safeguard when political winds shift, as they inevitably will.
"If we are going to buy a system, let's make sure it works," said Sen. Jack Reed, D-R.I. "The way we are going about this is terribly inefficient and wrong."
Michael Cabbage can be reachedo at email@example.com or 321-639-0522.