The first line of defense in America's next anti-missile system fails or succeeds in a window of 90 seconds.
That's all the time there is, designers estimate, for a satellite to detect the flash of an enemy launch, determine that it is real and send off a counter-missile from the ground.
It all happens too fast to include a human in the loop.
"Time is of the essence," said Craig van Schilfgaarde, the Northrop Grumman Corp. engineer in charge of the project.
Known as "boost phase" interception, it is designed to be the first "layer" of defense, firing rockets at enemy missiles just after launch, when they are most vulnerable.
The military has deployed parts of the two other layers in the missile defense system - one targeting missiles as they cruise through space in mid-flight and the other aimed at descending warheads when they are just above their targets.
The three layers are the cornerstone of President Bush's plan to defend the United States against rogue nations, such as North Korea and Iran, that are gradually developing the ability to produce weapons with global reach.
But the system has faced serious problems.
The midcourse missile failed a test Dec. 15 when it shut down before leaving its silo at the Ronald Reagan Test Site at Kwajalein Atoll in the central Pacific Ocean. It was the second failure in a major test in two years.
On Dec. 17, the Pentagon announced that it was dropping plans to activate the existing pieces of the missile defense system this year because it had not completed full "shakedown" testing.
The boost phase reaches into an even more complex realm of design, in part because of the speed with which it must identify and destroy an enemy missile.
The payoff could be big. Terry Little, executive director of the government's Missile Defense Agency, said the boost-phase interceptors could destroy 80 percent to 90 percent of enemy ICBMs, leaving the other layers to take care of the rest.
But a recent Congressional Budget Office technical report suggested that the boost-phase system, scheduled for deployment in 2011, would press the far edge of what is physically possible in an anti-missile system.
Philip Coyle, who led the Pentagon's testing office during President Bill Clinton's administration, said the design of the boost-phase system is buckling under its own complexity.
"The [congressional] analysis confirmed that boost-phase missile defense isn't practicable," Coyle said.
Inspired by Reagan
Today's missile defense programs were inspired by President Ronald Reagan's promise to end "nuclear blackmail" with his Strategic Defense Initiative, a plan to shield the nation against an all-out nuclear attack using satellite-fired interceptors.
Dubbed "star wars" by opponents in Congress, Reagan's program fell victim to technical dead-ends, cost overruns and concerns that it would violate the 1972 Anti-Ballistic Missile Treaty, which banned nationwide missile-defense systems.
Missile defense languished until 2002, when Bush withdrew from the treaty, which he considered a Cold War-era anachronism.
Instead of trying to defend against all-out nuclear attack by a major power, today's plan targets the less advanced arsenals of emerging nuclear states.
The entire system is budgeted at about $50 billion over the next five years and is likely to cost several times that sum to build, deploy and maintain.
In July, the Missile Defense Agency began deploying the midcourse interceptors in Alaska. A second battery is scheduled for deployment next year at California's Vandenberg Air Force Base.
Mobile Patriot anti-missile systems, a key part of the descent layer have been deployed.
A year ago, Northrop won a $4.5 billion contract to develop the boost-phase interceptors. Congress has approved $348 million for the current fiscal year.
Boost defense "would never be able to handle every situation that anybody could conceive of," said Little, of the Missile Defense Agency. "But we could handle enough that we could look at ourselves as an 80 percent or 90 percent solution."
The allure of striking enemy missiles in the boost phase is that they are easily identified by their plumes just after launch and, because they are ascending, cannot use their full bag of tricks to dodge and deceive.
To destroy a missile in the boost phase requires an unprecedented coordination of space-based sensors, signal-analysis computers, interceptor agility and enough sheer thrust to lift a 10-ton object to about 20 times the speed of sound in less than a minute.
Each interceptor consists of a two-stage booster, followed by a liquid-fuel rocket that steers the kill vehicle on the last leg of its journey to the target. It would travel at about 13,400 mph.
After infrared sensors on satellites detect the enemy launch, interceptors would be directed to the target by terrestrial command stations that constantly update the target's flight path. Onboard sensors would take over at close range.