At a newly constructed launch site on a tree-shorn plain in central Alaska, a large crane crawls from silo to silo, gently lowering missiles into their holes. The sleek white rockets, each about five stories tall, are designed to soar into space and intercept warheads headed toward the United States.
With five installed so far and one more due by mid-October, Defense Secretary Donald H. Rumsfeld is preparing to activate the site sometime this autumn. President Bush already has begun to claim fulfillment of a 2000 presidential campaign pledge -- and longtime Republican Party goal -- to build a nationwide missile defense.
But what the administration had hoped would be a triumphant achievement is clouded by doubts, even within the Pentagon, about whether a system that is on its way to costing more than $100 billion will work. Several key components have fallen years behind schedule and will not be available until later. Flight tests, plagued by delays, have yet to advance beyond elementary, highly scripted events.
The paucity of realistic test data has caused the Pentagon's chief weapons evaluator to conclude that he cannot offer a confident judgment about the system's viability. He estimated its likely effectiveness to be as low as 20 percent.
"A system is being deployed that doesn't have any credible capability," said retired Gen. Eugene Habiger, who headed the U.S. Strategic Command in the mid-1990s. "I cannot recall any military system being deployed in such a manner."
Senior officials at the Pentagon and the White House insist the system will provide protection, although they use terms such as "rudimentary" and "limited" to describe its initial capabilities. Some missile defense, they say, is better than none, and what is deployed this year will be improved over time.
"Did we have perfection with our first airplane, our first rifle, our first ship?" Rumsfeld said in an interview last month. "I mean, they'd still be testing at Kitty Hawk, for God's sake, if you wanted perfection."
This notion of building first and improving later lies at the heart of the administration's approach, which defense officials have dubbed "evolutionary acquisition" or "spiral development." Bush has scaled back President Ronald Reagan's vision of a vast anti-missile network and pursued a less ambitious system. At the outset, the system will be aimed only at countering a small number of missiles that would be fired by North Korea, which is 6,000 miles from the West Coast of the United States.
But Bush also has funded an expanded array of missile defense projects, including land- and sea-launched interceptors, an airborne laser, and space-based weapons. So far, he has spent $31 billion on missile defense research and development, and his plans call for an additional $9 billion to $10 billion a year for the next five years. Beyond that, the administration has provided no final price tag. In 2005, the cost of missile defense will consume nearly 14 percent of the Pentagon's entire research-and-development budget.
While more money has gone into missile defense under Bush than into any other military R&D project, the Pentagon has exempted the missile defense program from the traditional oversight rules meant to ensure that new weapons serve the needs of military commanders.
Administration officials say the procedural shortcuts and the increased spending have yielded record gains in record time. The urgency, they say, is justified by a growing U.S. vulnerability to attack from hostile states pursuing long-range missiles -- most notably North Korea and Iran.
Critics warn that such haste has made waste -- and is unnecessary. The urgency, they suspect, has been more a reflection of politics than concerns about the missile programs of North Korea and Iran, which still face significant technical hurdles. The deployment is being timed, they contend, to help Bush's reelection campaign.
They also caution that fielding a U.S. anti-missile system before it has undergone realistic testing risks inducing a false sense of security and locking the United States into flawed technology.
"The design gets frozen in order to build something, so development is stopped," said Philip E. Coyle III, the Pentagon's chief weapons evaluator during the Clinton administration. "You can't be building a house and changing the floor plan at the same time."
Out With the Old
Normally, when a weapons system is conceived, the Pentagon sets specific requirements that must be approved by a committee of senior military officers. The project is then assessed periodically by the Defense Acquisition Board, a group of high-ranking defense officials from various offices.
This accountability apparatus has been shunted aside in the case of missile defense. No requirements document was drawn up, and the traditional reviews and assessments have been bypassed. Instead, the Missile Defense Agency (MDA), which is responsible for developing the system, has been allowed to devise its own goals, test schedules and program reviews.
When Rumsfeld authorized this extraordinary autonomy in January 2002, he said that technological challenges and urgent national security concerns justified it. As a former executive in the pharmaceutical industry, Rumsfeld by his own account was influenced by the vigorous trial-and-error competition that often precedes the creation of new drugs.
Other historical models also inspired Pentagon authorities. One was the National Reconnaissance Office, established in great secrecy in the 1960s to develop and operate spy satellites. The other was Israel's decision, in 2000, to declare its Arrow anti-missile system operational after just one successful intercept test.
"Since we have urgent needs, we sometimes cut corners in developing systems, meaning we field them before we've developed everything," said Arieh Herzog, director of the Israeli Missile Defense Organization. He said he held "many talks" about Israel's approach with Lt. Gen. Ronald T. Kadish, MDA director at the time.
Opponents in Congress and elsewhere say this approach has been taken too far in the case of the U.S. system. They warn that the lack of established baselines for the missile defense program has made it difficult to hold the Pentagon accountable for performance and cost.
"We're in this hugely expensive race to build something, but we don't know how much it'll cost in the end or what it'll do," said Sen. Jack Reed (D-R.I.), a member of the Armed Services Committee.
An audit by the Government Accountability Office, released in April, cited an absence of reliable, complete baseline estimates of system performance and cost. Without this information, the GAO said, policymakers in the Pentagon and Congress "do not have a full understanding" of the system's overall cost and actual capabilities. The audit concluded that the system being fielded this year remains "largely unproven."
Pentagon officials say the program remains subject to extensive internal supervision, even with the departure from traditional procedures. Michael W. Wynne, the Pentagon's acting head of acquisitions, told a Senate committee in March that he meets weekly with the MDA's director. In contrast with other programs he oversees from a distance, Wynne described his contacts with top MDA officials -- and with Rumsfeld -- as "more direct and generally carried out in face-to-face discussions."
Wynne said other senior Pentagon officials also have had a say in shaping and scrutinizing the program. He pointed to the Missile Defense Support Group, which consists of mid-level representatives from Rumseld's office, the Joint Staff and each of the military services. The group has met 47 times since its creation in March 2002.
"No program in the department receives more scrutiny -- either in level or frequency -- than the Missile Defense Program," Wynne testified to the Senate Armed Services Committee.
But interviews with support group members revealed they have played only an advisory role. Several said the group often learned of some important decisions after the fact.
"We're not a critical-decision review group," said Glenn F. Lamartin, a senior Pentagon acquisition official who chairs the group. "We're a support group. We provide advice. Our engagement is different than if we were operating under the old system of review and oversight."
Lately, some senior military commanders have signaled an interest in shifting back toward some sort of formal requirement process. The Strategic Command, which will oversee operation of the missile defense system, has proposed a "warfighter involvement program" to give commanders a greater voice in the system's development.
Another important source of internal review is supposed to be Thomas P. Christie, the Pentagon's chief weapons evaluator. But he is in an awkward position.
By law, his Operational Test and Evaluation office is mandated by Congress to judge the readiness of major weapons systems before they are deployed, which it does by comparing the results of "operational" tests with the requirements for the system. In the case of missile defense, however, no formal requirements exist, and the test data so far come from early "developmental" flights, not more realistic operational ones.
Christie's estimate that the system may be only 20 percent effective contrasts with a prediction from the MDA of more than 80 percent effectiveness. The difference reflects disagreement over which test data to include in computing the estimates.
Christie wants to count all flight results, including earlier test failures. The MDA argues that causes of those failures have been fixed, so the data can be discarded. Its estimates are based largely on computer simulations and testing of individual components.
The MDA's director, Air Force Lt. Gen. Henry A. "Trey" Obering III, said in an interview earlier this month that both sides were trying to settle on a common set of data. Two other senior defense officials said this week that an agreement had been reached on "selection criteria" for the data and that the gap between the disparate estimates had begun to narrow.
The numbers, which are classified, carry considerable importance because a future U.S. president would rely on them in a crisis.
"He will want to know what his options are and will turn to his commanders and ask how sure they are that the system will work," said a senior Pentagon official involved in the assessment.
After Bush took office, Pentagon officials outlined a plan of stepped-up flight intercept tests. The plan held, more or less, through the end of 2002 and several successful intercepts. But unexpected difficulty in producing a new booster rocket has stalled intercept tests since December 2002.
The booster's job is to carry a "kill vehicle," a 120-pound package of sensors, computers and thrusters. Once in space, the kill vehicle separates from the booster and closes in on an enemy warhead, destroying it in a high-speed collision.
Earlier flight tests used a surrogate booster that flew at only half the speed of the booster that is being produced for the system. MDA officials said they have not wanted to try more intercepts until that new booster can be incorporated into the tests.
By spring of this year, the new booster was ready, but the discovery of a faulty circuit board in the kill vehicle prompted Pentagon officials to order a lengthy bottom-up review of all components. In mid-August, the missile interceptor was again set to go when technicians found a glitch in the booster's flight computer. Replacing the computer created another delay.
Earlier this month, Obering, the MDA's director, announced a further postponement after discovering modifications that had been made to the interceptor without thorough ground testing.
This leaves the administration proceeding with deployment after only eight intercept tests -- the most recent conducted 21 months ago. Five tests resulted in hits, but all used the same limited test range in the Pacific and employed surrogates for tracking radars as well as for the booster.
A key X-band radar -- a towering structure being built to float at sea on two motorized pontoons the size of Trident submarines -- will not be ready for another year at least. Also still in development is a satellite network to replace a three-decade-old constellation of early-warning satellites. Both the X-band and the new satellites are critical in assisting the kill vehicle to distinguish the warhead from decoys and debris.
Physicists and defense experts, including some affiliated with the Union of Concerned Scientists, continue to dispute the MDA's claims that the system will be able to identify a warhead in a field of decoys -- a process known as "discrimination." Tests so far, using only relatively simple targets, have done little to resolve the issue.
Pentagon officials say the system has been subjected to an extensive array of ground tests and computer simulations. The failures on intercept attempts, they note, resulted from problems with the quality of individual parts, not from basic design flaws.
"If you really look at where we've encountered problems, they have not been things that required technological breakthrough. They have been in attention to detail and quality," said Maj. Gen. John Holly, the MDA's manager for what is formally known as the Ground-based Midcourse Defense system.
But both Christie and the Defense Science Board, a Pentagon advisory board, have cited limits to the computer models. And the persistent quality control problems have led a number of scientists, defense specialists and Democratic lawmakers to argue that the system's sheer complexity makes it highly vulnerable to the malfunction of a single part.
'We Needed a Date'
The Pentagon has aimed at having the Alaska site ready by Sept. 30. Kadish, the former MDA director, said that date was chosen by his agency in early 2002 for "internal management purposes," not by political appointees with an eye toward the Nov. 2 presidential election.
"We needed a date for people to work to," he told the strategic forces subcommittee of the House Armed Services Committee in March.
At the time the date was picked, the Alaska site was being conceived primarily as a "test bed" that would allow for more realistic flight testing and cold-weather ground operations. It could be used in an emergency to thwart a real attack, defense officials figured, but that would not be its main purpose.
In the summer of 2002, the plan began to change. Pentagon officials proposed turning the site into a fully operational anti-missile facility and deploying more interceptors there while still using it as a test area. Bush approved the plan a few months later, ordering deployment in 2004.
Despite the slippage in the test program, the administration has held fast to that deadline. In recent weeks, Bush and Rumsfeld have reiterated their intention to activate the system by year's end.
This will not be the first time Pentagon officials have pushed a new weapon system into service while it is still in its experimental phase. The Joint Surveillance Target Attack Radar System (JSTARS), a ground surveillance aircraft, was rushed into action in the 1991 Persian Gulf War, and Predator and other unmanned reconnaissance aircraft were hurriedly deployed in the Balkans and Afghanistan.
But JSTARS and the drone aircraft were far simpler to develop than missile defense. Coyle, the former Pentagon weapons evaluator, compared the idea of deploying while testing the missile defense system to building a picket fence, one picket at a time, over several years. Until the whole thing is complete, such a fence is not much use, he said.
Coyle and others also worry that placing the system on alert will distract from further testing and development. They point to the experience in the 1990s of a shorter-range anti-missile system known as Theater High-Altitude Area Defense, or THAAD. The Army required the developer of that system to produce 40 prototype missiles, reliable and rugged enough to use in a combat emergency. The pressure to deploy something early led to compromises in design and testing, which hampered the program and contributed to years of delay.
Pentagon officials are still wrestling with what balance to strike between keeping the system functioning and suspending operations to run tests.
"You could say that if you put it on alert, it might be a distraction to the development and evolution of the system," Rumsfeld said. "But you could say just the opposite -- that by putting it on alert, you force up a whole series of issues that you need to think through, work through."
Rumsfeld has made it clear that in the absence of an international crisis involving a heightened threat of missile attack, he would favor giving priority to continued testing. Nevertheless, the administration does not intend to wait for more proof of performance before expanding the system.
In addition to 16 interceptors already ordered for the Alaska site at Fort Greely -- plus four for an alternate California site at Vandenberg Air Force Base -- the 2005 budget provides money for 10 more interceptors in Alaska. Talks also are underway with several countries about establishing an interceptor site in Europe.
'The Sky Did Not Fall'
In the early months of the Bush administration, congressional Democrats plotted to block White House plans to expand work on missile defense. After the World Trade Center and the Pentagon were struck on Sept. 11, 2001, they tried to argue that the attacks showed that the United States had more to fear from low-tech terrorism than high-tech missiles.
But the attacks, by giving new emphasis to homeland defense, have played to the advantage of missile defense proponents. Amid a general surge in military spending, Bush has received nearly all of the money he has sought for missile defense.
Democratic lawmakers opposed to Bush's program concede the debate has shifted. It is no longer an ideological battle, centered on arms control concerns, over whether to deploy at all. Now, they say, it is a more practical argument over how much to build and how fast.
"The debate is now about whether or not we continue to press ahead at the full speed we're going, with record amounts of money being spent, despite the fact that there's been no realistic testing," said Sen. Carl M. Levin (Mich.), the ranking Democrat on the Armed Services Committee.
Rumsfeld, addressing an audience of government officials and contractors last month, said the upcoming deployment is "somewhat of a disappointment for those who were convinced it would fail." Noting cordial discussions he had held only days earlier with Russian officials about missile defense, he chided arms-control advocates who had forecast that the U.S. initiative would upset relations with Moscow.
"The sky-is-falling group was wrong," he said. "The sky did not fall. It's still up there."