Index

The New York Times March 7, 2000

Ex-Employee Says Contractor Faked Results of Missile Tests

By WILLIAM J. BROAD

A former senior engineer at TRW, a top military contractor, has charged the company with faking tests and evaluations of a key component for the proposed $27 billion antimissile system and then firing her when she protested.

The engineer, Dr. Nira Schwartz, was on the company's antimissile team in 1995 and 1996 helping design computer programs meant to enable interceptors to distinguish between incoming warheads and decoys. In test after test, the interceptors failed, she has alleged, but her superiors insisted that the technology performed adequately, refused her appeals to tell industrial partners and federal patrons of its shortcomings, and then fired her.

Dr. Schwartz has made her charges in interviews and in newly unsealed documents filed with a federal district court in Los Angeles, where nearly four years ago she sued TRW. She seeks to recover for the government more than a half-billion dollars, some part of which a judge could award her as compensation.

In interviews and court filings, TRW has vigorously denied the charges. But citing the pending litigation, it has refused to address many details of the accusations.

Right or wrong, the investigations surrounding Dr. Schwartz's claims are shedding light on a contentious and secretive program. The Clinton administration has said it will decide this summer whether to proceed with an antimissile system meant to defend against warheads from so-called rogue states. Russia and other countries warn that such a decision could undermine a treaty limiting antimissile defenses that has served as a pillar of arms control.

In 1998 the Pentagon rejected the TRW interceptor as the leading antimissile candidate in favor of a rival design by Raytheon. However, it is still a backup and could win the lead role since the Raytheon design has stumbled in recent flights.

Dr. Schwartz's charges have split the Defense Department. Some top officials defend TRW as innocent, and the Justice Department has so far declined to join her lawsuit. But a three-year inquiry by the Pentagon's Defense Criminal Investigative Service, which ended last August with no action, cited in its final report "numerous technical discrepancies" that "appear to warrant further review."

Moreover, former TRW employees back Dr. Schwartz. In an affidavit filed in connection with her suit, Roy Danchick, a retired senior engineer at TRW, said he had firsthand knowledge of TRW's "impermissibly manipulating" a study of the antimissile technology and "censoring the test data" so it appeared more successful than it was.

Dr. Schwartz's allegations center on TRW's certifying to the government that interceptors using its computer programs would succeed more than 95 percent of the time in picking out enemy warheads, even if they were hidden in a confusing blur of decoys in space. In fact, Dr. Schwartz said in court documents, the interceptors could do so only 5 to 15 percent of the time.

Her charges are coming to light now because many secret court filings have been unsealed at her request and she is seeking public support for her case.

Dr. Schwartz, 53, a naturalized United States citizen from Israel, has a Ph.D. in engineering from Tel Aviv University and holds 18 United States patents, including ones involving computerized image analysis and pattern recognition.

TRW hired her to help develop computerized algorithms that are meant to enable an interceptor to differentiate between real and fake warheads by matching memories of threatening images against a rush of incoming sensor data.

Dr. Schwartz said in an interview that in time she had concluded that all the current discrimination technologies were too feeble to work and that at some level the Pentagon and its contractors were in collusion.

"It's not a defense of the United States," she said, eyes flashing. "It's a conspiracy to allow them to milk the government. They are creating for themselves a job for life."

According to court records, Dr. Schwartz was hired as a senior engineer by TRW on Sept. 5, 1995, joining the company's space group in Redondo Beach, Calif. TRW was allied with Rockwell (later bought by Boeing amid defense consolidations) in a competition to build a "kill vehicle," which would zoom into space atop rockets and smash enemy warheads to pieces.

The work for which Dr. Schwartz was hired, enabling the kill vehicle to spot enemy targets, represented the soul of the machine.

"She is almost uniquely qualified to strengthen the interplay" among TRW specialists pursuing image analysis and object recognition, her boss wrote personnel officials after she was hired.

Among her jobs was to help assess a kill-vehicle program called the Kalman Feature Extractor. From incoming sensor data, the extractor was to find critical characteristics of scanned objects, teasing out familiar "signatures" that could separate decoys from warheads.

TRW executives believed the extractor offered a competitive edge over the rival industrial team (then Hughes, later bought by Raytheon), and they told the government it was a potential breakthrough.

Familiar with the extractor from previous research, including that for her patents of 1983 and 1989, Dr. Schwartz proceeded to test it against nearly 200 types of enemy decoys and warheads in computer simulations, using secret intelligence data.

"The moment I analyzed the signatures," she said in an interview, "I saw there was a problem." Most of the time, she said, the kill vehicle's extractor program failed to distinguish between warheads and decoys because their identifying signatures differed wildly depending on variables like spin, attitude, temperature, wobble, deployment angle and time of day and year.

"For every RV there was a decoy" that produced an identical signature, Dr. Schwartz said, referring to a warhead as a reentry vehicle, or RV.

The variations she used in the tests, she added, were not arbitrary but spelled out in the Pentagon's antimissile bible, the 1993 "Technical Requirements Document."

The results were roughly the same, Dr. Schwartz said, with the math tool failing about 90 percent of the time, when she tested TRW's less sophisticated method for separating wheat from chaff, known as the baseline algorithm.

In interviews and court documents, Dr. Schwartz said she told her boss and colleagues of her discoveries. "They said not to worry," she recalled. Flight tests of mock objects, she said she was told, would always be arranged so discrimination would be relatively easy.

When she pressed her superiors to tell industrial partners and the military of the shortcomings, especially of the extractor, the purported star of the show, they refused, according to court documents. Days later, in late February 1996, she was fired.

"If you will not notify the U.S. Government," she warned TRW in a furious letter after the firing, "then I will."

Robert D. Hughes, her former boss, expressed no regrets. In company documents, he told colleagues that Dr. Schwartz had been completely off target, quick with her diatribes yet misreading her own tests and TRW's process of product development. The extractor was not finished but evolving.

"Stating that there is a 'defect' that we should immediately report to the customer makes no sense," Mr. Hughes told his boss in a March 1, 1996, letter. "We continually improve and verify system performance."

Mr. Hughes added that the extractor "appears to perform properly under all conditions that have been specified to us" by the government and fails only when flight environments are "highly improbable" and "appear not to be within our requirements."

Dr. Schwartz contacted federal investigators, and on April 29, 1996, filed in federal district court in Los Angeles a lawsuit under the false claims act that sought to recover damages for the government. The suit was kept under seal. It alleged that TRW had "knowingly and falsely certified" discrimination technology that was "incapable of performing its intended purpose."

By June, the Defense Criminal Investigative Service had opened an inquiry out of its field office in Mission Viejo, Calif. The investigative service, an arm of the Pentagon's inspector general, is a federal law enforcement agency that can make arrests and subpoena documents.

The lead investigator, Samuel W. Reed Jr., interviewed Dr. Schwartz, military experts, TRW officials and a network of informers. The Justice Department monitored the case. Under the law, it could choose to join Dr. Schwartz's lawsuit, which would raise her chances of winning.

Meanwhile, a $100 million test was nearing; it would help determine whether Dr. Schwartz's charges were plausible. For the first time, a kill vehicle under development in the program would be shot into space to see if it could discriminate between a warhead and decoys.

Despite the long work on the extractor, TRW decided to abandon it and rely on the backup, the baseline algorithm. TRW told the military that the extractor was simply too big for the kill vehicle's computer, but court documents suggest that by then it was widely regarded as a failure.

On June 23, 1997, the mock enemy missile -- containing the cone-shaped target warhead, six balloons and three decoy warheads -- was launched from California and sped southwest. Twenty minutes later, the interceptor was fired from a Pacific atoll, its computer brain ready to try to discern the target warhead. The test was to be a simple flyby, with no effort at interception.

The Pentagon hailed the test publicly as an outstanding success, but privately, analysts were dismayed. Even with all the coaching, the computer brain of the kill vehicle had selected a decoy, Dr. Schwartz and other experts said in interviews.

Undaunted, TRW found good news, as Mr. Danchick, the former senior TRW engineer, later told Pentagon investigators, when data beamed back from the kill vehicle was later reworked. In a postflight analysis meant to mimic the real thing, TRW managers interrupted the analysis to select a more favorable stream of data for the baseline algorithm to digest, Mr. Danchick told the investigators. Then, finally, the program zeroed in on the warhead -- a point hammered home in a vivid company graphic.

This was what Mr. Danchick later called in the affidavit TRW's "impermissibly manipulating" test results. Officials at TRW and in the government will not say if the government back then learned of these acts, though the company in a recent statement said it "categorically denies that it misrepresented or improperly manipulated any flight test data."

In any event, Mr. Reed, the Pentagon investigator, became convinced that the matter warranted close scrutiny, as he detailed in more than a dozen reports and letters later filed in the court case. He also noted that TRW was the antimissile program's overall system engineer, with responsibilities to "analyze and validate results" of all flight tests, and thus had a potential conflict of interest.

By this time, Mr. Reed's talks with military officials had produced a scientific analysis of Dr. Schwartz's charges. Nichols Research of Huntsville, Ala., a regular adviser to the antimissile program, found in December 1997 that TRW's discrimination work was "no Nobel Prize winner" but met contract requirements. It added that TRW had failed to answer many questions.

Mr. Reed then asked Pentagon officials for a deeper study, which was done in 1998 and delivered late that year. The five scientists for the study were drawn from a high-level antimissile advisory board with Pentagon ties dating back to the Reagan "Star Wars" era.

The new study team found the baseline algorithm "basically sound" but questioned some of its purported strengths. For instance, it showed that small changes in planned sensor data greatly lowered the odds of selecting the right target. Importantly, it blessed TRW's manipulating some data from the first flight test, saying a computer program was being developed that, if perfected, would do the same in space automatically.

As for the abandoned extractor, TRW's onetime star, the new team said dismissively that its "application to discrimination may be limited."

Though equivocal, that study was enough for the Justice Department. Three years after Dr. Schwartz's initial charge, it declined to join her false-claims litigation.

But it hardly ended the story. In a memorandum dated March 15, Mr. Reed asserted that the Army had falsely reported to the Justice Department that he was against federal intervention. "This statement has no factual basis," he seethed.

Mr. Reed formally closed his investigation of TRW on Aug. 31, 1999, citing numerous "irregularities" and "discrepancies." Shortly thereafter, colleagues said, he retired from the Pentagon's police force.

A federal official unconnected to the Pentagon but familiar with the case, who spoke on the condition of anonymity, faulted the military's two scientific studies of Dr. Schwartz's charges as biased. "They don't go biting the hand that feeds them," he said of the antimissile advisory groups.

He added that the studies had enough of the appearance of impartiality to dissuade the Justice Department from entering her case.

Meanwhile, Dr. Schwartz has added to her antimissile criticism. Her investigations had shown that the Pentagon planned to use up to nine decoys in interception flight tests. But when it actually performed the tests last October and this January, each flight had but one decoy balloon. In an interview, she attributed this simplification to the military's belated discovery of fraud and how hard the job actually was.

TRW, though spared a federal assault by the Justice Department's decision, fared poorly in the investigations. In May 1998, its industrial team lost a bid to become the antimissile system's "lead system integrator," a post worth $1.6 billion over three years. Simultaneously, the company also lost its system engineer status. In December 1998, its team lost the competition to build the kill vehicle. The prize was won by Raytheon, a relative newcomer to the field that developed its own discrimination software and is now struggling with problems uncovered in recent flight tests.

Last October, the Pentagon hailed the Raytheon test as an unqualified success. But it later acknowledged that the kill vehicle had initially drifted off course and picked out a decoy balloon rather than the warhead.

In a January test, the interceptor missed the target altogether.

Dr. Schwartz lives with her husband, an aerospace engineer, in a two-bedroom town house in a gated community in Torrance, Calif., and runs computers for a plastics company.

After representing herself in her case against TRW for years, Dr. Schwartz recently retained David W. Affeld, a Los Angeles lawyer with a bachelor's degree in physics. She has filed a wrongful termination action against TRW. And she still, despite the ups and downs, wants the government to join her false-claims suit.

Keith Englander, director of system engineering for the Pentagon's antimissile program, echoing TRW, said the two studies had discredited Dr. Schwartz. Any problems with the company's discrimination work, he added, were a normal part of the engineering design process.

He also defended his program against her wider criticisms, saying flight tests were never rigged and results never doctored.

The reduction on interception tests from numerous decoys to just one, Mr. Englander said, was because the government had cut the program's goal from trying to knock out advanced warheads from countries like Russia and China to more primitive ones from rogue states.

"Another objective," he added, "is to walk before we run. When you start, you don't go to the most stressing environment."

The Pentagon is sharply divided on such issues. Last month in a public report, Philip E. Coyle 3rd, the Defense Department's director of testing and evaluation, faulted the antimissile tests as insufficiently realistic "to support acquisition decisions."


Copyright 2000 The New York Times Company