Missing the Target
Postol’s latest exercise in missile defense whistle-blowing began a continent away in Redondo Beach, CA, and then moved along its own convoluted course for five years before Postol took it on as his own personal cause. For Postol, the story of this intrigue would become the leverage with which he would try to force public discussion and oversight of national missile defense research in a world in which, as Democratic congressman Tom Allen of Maine says, “the devil is in the details, and the details are classified.”
In this case, Postol acted as not so much a whistle-blower as a whistle amplifier. The first report of something wrong came from Nira Schwartz, a naturalized U.S. citizen from Israel and an expert on computer image analysis and pattern recognition. In 1995, TRW hired Schwartz to work on the software for an exoatmospheric kill vehicle that would in turn be built by Boeing, all under the auspices of the Pentagon’s Ballistic Missile Defense Organization.
Schwartz’s task was to test and evaluate the algorithms that the kill vehicle would use to track an incoming warhead and to discriminate it from a potential swarm of decoys. Such target discrimination is the single most critical technology for the success of any antiballistic-missile defense plan. If target discrimination can be done in real-world circumstances-if a speeding bullet can truly track a speeding bullet, regardless of whatever countermeasures the enemy bullet deploys-then an antiballistic-missile shield might indeed protect the United States from attack. If target discrimination cannot be done under those circumstances, however, then any national missile defense system will fail.
As Schwartz later testified, she realized the TRW programs were incapable of discriminating warheads from decoys when they failed to do so repeatedly in her computer simulations. She further concluded that warheads produced no particular “signature” of heat or radiation or movement that would ever uniquely identify them.
Schwartz’s insistence that TRW and Boeing admit to the Ballistic Missile Defense Organization that the discriminator wouldn’t work did not endear her to her employers; in early 1996, she was fired. She then filed suit in Los Angeles district court alleging that TRW had defrauded the U.S. government and seeking to recover damages for the government. (Schwartz sued under the False Claims Act, a law that allows private citizens to blow the whistle on people or companies that are defrauding the government-and get a cut of the award money should they win.)
In such a case, the U.S. Justice Department can choose to join the suit if its investigators decide the suit is valid. The Justice Department asked the Defense Criminal Investigative Service to look into Schwartz’s allegations, and the service assigned Sam Reed to lead the investigation. Reed relied on Schwartz and other experts, including Roy Danchick, a senior TRW engineer who had also worked on the targeting software, to help him make sense of the complex scientific and technological issues. Reed concluded that Schwartz’s accusations had merit and that, as he later wrote to the Ballistic Missile Defense Organization, the TRW discrimination program “does not, cannot and will not work.”
While Reed’s investigation was in progress, the TRW software was put to its first $100 million flight test. In June 1997, the Boeing exoatmospheric kill vehicle was fired into space from a Pacific atoll to intercept a test missile that had been launched 20 minutes earlier from California. The kill vehicle wasn’t supposed to actually intercept the missile-just fly by it. The mission was to test the ability of TRW’s discrimination program to identify a mock warhead amidst a handful of decoys. Both the Pentagon and TRW would claim that the test was a success. But when Reed had Schwartz, Danchick and his other experts examine the data, they concluded otherwise. As Danchick would testify, the TRW analysis was “impermissibly manipulating” the data to get the right answer (see “Why Missile Defense Won’t Work”). As he described it to Technology Review, the researchers had “fudged, dry-labbed, manipulated and censored the data to get the result they wanted.”