An Open Letter to the MIT Faculty

When Everything is Secret, There is No Truth

Theodore A. Postol

In November 2002, Professor Edward Crawley recommended a full investigation of my allegations of fraud in Lincoln Laboratory's evaluation of a critical National Missile Defense (NMD) test that was aimed at determining whether the NMD would ever be able to tell the difference between warheads and simple decoys. For two years, the MIT administration took no action on the recommendation despite its implications for academic integrity and national security. Then, as a parting gesture, President Vest declared in December 2004, that the investigation could not proceed because it depended on classified information that the Missile Defense Agency (MDA) would not let MIT use - even though the information was already held by Lincoln Laboratory.

In addition, MIT says it cannot provide the names of the individuals it claims to have "identified" for its "ongoing" investigation, nor can MIT explain why its ongoing investigation has no mechanism to review and assess information already in-hand and available. MIT can also not explain why the investigation process is now secret and opaque, nor can it demonstrate that the process is being implemented with impartial and independent investigators, or even that there is an investigative process.

The basic facts are either known, or available, to MIT investigators and could be used to investigate this matter using open, transparent, and peer-reviewed methods, rather than those that have been adopted.

In his play, "All My Sons," Arthur Miller depicts Joe Keller, a World War II manufacturer of aircraft engines who knowingly ships defective engines to combat units. As a result, 21 planes and their pilots are lost, and one of Keller's sons, a serving combat pilot, kills himself when he learns of his father's deception. Joe Keller's morality placed his short-term financial interests above his duty to country.

MIT's unwillingness to investigate its own role in concealing fundamental flaws in the National Missile Defense system raises the same moral issues; what is MIT's obligation to the country when it knows that the Institute may have lied about a defense that is supposed to protect the nation?

In what follows, I will address the question whether there is sufficient evidence that could be used by MIT to determine that Lincoln Laboratory misled federal agents during an investigation of possible fraud in the National Missile Defense system that is now being deployed by the Bush Administration.

The document that initially caused me to conclude that federal investigators were misled by Lincoln Laboratory management and staff was produced by Lincoln during the summer of 1998 to support a federal investigation of possible fraud by a defense contractor, known as TRW. Lincoln was supporting the investigators in its statutory capacity as a Federally Funded Research and Development Center (FFRDC). The title of the report Lincoln produced for the investigators is an "Independent Review of TRW Discrimination Techniques."

The statement of work (SOW) that set out the tasks for the study required that the study "be done in cooperation with the . Department of Defense Inspector General" as well as with various defense contractors and agencies associated with the Ballistic Missile Defense Organization (now known as the Missile Defense Agency). The basic tasks for the Lincoln Laboratory-led study was to "review the . [ability of computer algorithms to discriminate between warheads and decoys using] . the data from [the] IFT-1A [experiment] and [to] evaluate(s) the [accuracy of claims about performance] .reported by TRW to the government .."

Back to top

At issue was the Baseline Algorithm (or BLA). The algorithm was designed to examine signals collected by an infrared sensor that had been launched into space from the Lincoln Laboratory-run missile test site on Kwajalein atoll in the South Pacific. The experiment was to test the ability of infrared sensors to observe mock warheads that are accompanied by balloons and rigid objects that could serve as decoys to fool the infrared homing interceptors that are being used in what is now the Bush National Missile Defense system. If this experiment, and the closely related IFT-2 experiment, could not demonstrate that the "Kill Vehicle" could discriminate between the mock warhead and these simple decoys, the NMD system would have little or no chance of working in real combat.

The Baseline Algorithm was supposed to work in a way that is roughly analogous to computer programs that are designed to "recognize" text on printed pages.

Warheads and Decoys
(click on image to enlarge)

An obvious requirement for recognizing the text is a good prior knowledge of the geometry and variations in geometry of the letters and symbols that the computer program is trying to "recognize" (or more accurately, match). In the case of the Baseline Algorithm, the distant objects to be "recognized" appear as dots when observed at hundreds of kilometers range. The objects are recognized by first making careful physics-based calculations to determine the infrared signals from the objects when they are in space. These calculations are then used to create "templates" that predict the "color," brightness, and twinkling of the signal from each object as seen by the sensor. In order to make a match, the sensor must be able to accurately measure the color and brightness of each object. The predictions of how each object looks to the sensor must be accurate and closely match what the sensor actually sees.

In March of 2002, the General Accountability Office (GAO) issued two investigation reports that found that sensor in the IFT-1A had lost calibration due to the failure of its supporting cooling system.

Without calibration, red dots might look green, green might appear yellow, and the colors of the other dots would also be distorted. Without information on how to correct the color and brightness distortions, there was no way to match what was seen to what was expected. In turn, it was not possible to demonstrate that the Baseline Algorithm could select the right objects. In spite of this situation, the GAO reported that Lincoln Laboratory had told the federal agents that the Baseline Algorithm had worked well.

Prior to the GAO report, I had written to MIT's then Chair of the Corporation, Alex D'Arbeloff, warning him that the sensor in the IFT-1A had lost calibration and that Lincoln's claims to the federal agents could not possibly have been true. Provost Robert Brown acknowledged my warning in a letter to me dated February 11, 2002. When the GAO reports confirming my warning were published weeks later, I provided MIT with carefully annotated versions of the reports. As a result, MIT was informed of the problem and the issues with Lincoln Laboratory even before the GAO issued its reports confirming my warnings.

Letters that MIT have in hand between the Defense Criminal Investigative Service and Lincoln Laboratory show that Lincoln repeatedly failed to comply with requests for information from the lead federal investigator. The particular information that was requested from Lincoln would have immediately and distressingly indicated that fraud had occurred at TRW.

For example, the lead federal agent believed that there were 60 seconds of data from the experiment. When he asked Lincoln about the first 30 seconds of data, he was not told that there was no data for this period because the sensor had malfunctioned. When he asked about the last 11 seconds of data, he was not told that the Baseline Algorithm had incorrectly identified a two-foot diameter inflated balloon for the warhead. When he asked about a 16-second time interval in the experiment that Lincoln claimed to be analyzing, he was not told that the templates for matching the expected signals to the measured signals had been altered to make it appear that the warhead was correctly selected. In other words, documentation that MIT has in hand shows that Lincoln not only did not reveal critical information to the federal agents, but made claims to the agents that could not possibly be true.

Back to top

Over the past four years I have spoken numerous times with the federal agent who led the investigation at Lincoln. He has repeatedly told me that he was not informed by Lincoln of the problems with the sensor, the loss in calibration, nor the dramatic increase in noise that obscured the signals in some cases and distorted the signals in others. I provided MIT with the name and contact information for the federal agent nearly three years ago; as yet no one from MIT has contacted him.

Independent investigators will need to examine MIT's own interim inquiry report, which was provided to me roughly 15 months after this matter was brought to the attention of MIT's then President Vest, and four months after the GAO had issued its two damaging reports.

Information from the MIT interim report and from discussions I had with the then Chairman of the Senate Armed Services Committee Senator Carl Levin, and with one of his aides, indicate that in the spring of 2001 a five-member team of MIT Lincoln Laboratory management and staff, including an assistant division head, went to Washington to brief Senate Armed Services Committee members and staff. The Lincoln briefing, which was unknown to me at the time, was aimed at debunking allegations I had made in a widely-publicized letter to the White House about scientific fraud in the integrated Flight Test 1A (IFT-1A).

Among those briefed by the Lincoln team were Senator Jack Reed, the then chairman of the Subcommittee on Strategic Forces, and staff members who worked for Senator Levin. The Lincoln Laboratory team told the audience that the allegations of fraud were bogus, and that the NMD would have no serious problems telling warheads from decoys. When the Lincoln team presented its analysis of why the allegations of fraud were bogus, they failed to inform the audience that the data from the IFT-1A experiment was corrupted by a failure of a cooling system, and that the experiment could not possibly have been used to prove their conclusion.

As Senator Levin explained to me when we later talked about the Lincoln briefing, "I have one MIT professor who tells me that the NMD system has problems, and a group of MIT professors from Lincoln who tell me the system will work fine."

An MIT investigation could surely determine, without access to secret information, why Lincoln personnel produced a misleading briefing for senior members of the Senate, who told them to do it, and how they were mistakenly taken for MIT faculty.

The MIT interim report asserts that every material fact central to whether or not Lincoln was involved in a cover-up of contractor fraud contradicted the facts reported by the GAO. Since the GAO had obtained almost all of the technical material for its investigation at Lincoln Laboratory, the GAO followed its normal procedure of providing Lincoln with its factual findings so that any inaccuracies could be corrected. Before I was allowed to see the MIT interim report, the MIT fact finder provided it to the same managers and staff at Lincoln who had reviewed the GAO report for accuracy. Yet the facts in the MIT interim report contradicted those in the GAO report. In addition, the interim report was also reviewed by MIT legal counsel before it was provided to me. I provided MIT with an extensive analysis of the interim report, pointing out specifically where the report appeared to indicate fraud in the management of MIT's internal investigation, but MIT has yet to review and respond to the materials I provided. These materials are not secret and need to be reviewed in an independent investigation of this matter.

Further post-flight documentation available to MIT provides detailed temperature records of the malfunctioning cooling system during the IFT-1A test.

Other post-flight data shows that a few tens of seconds after the failed one-minute attempt to collect data on the objects in space, the sensor was turned to look at the brightest infrared star in the Northern Hemisphere, Arcturus, a red giant with a very strong and well-known spectrum. This star sighting would provide data on a known object that could then be used to improve the precision (or calibration) of measurements made on the mock warhead and decoys.

The post-flight data from the star sighting shows that the observation of the star was swamped with noise. In addition, infrared data from the mock warhead shows that it too was swamped with very high levels of noise. The large amounts of noise in the data from the mock warhead, and from Arcturus, would have immediately indicated to researchers that the experiment had failed, since the precise data needed to derive results from the experiment was not obtained. The failure of the experiment would also have been immediately evident to the researchers monitoring the flight of the sensor, because the temperature of the sensor transmitted from the space experiment showed that there had been a failure in the cooling system. All this information is available to MIT, but has yet to be examined or evaluated.

So MIT's assertion that it cannot access information necessary to investigate this matter is simply not supported by even a cursory review of the extensive body of information that is already available to investigators. This matter can be resolved without access to secret information. To argue otherwise impairs the credibility of the Institute.

MIT's new president, Susan Hockfield, has the authority to reverse this ill-considered policy. She should do so in the interest of our nation's security and the integrity and reputation of academic research at MIT.

Editor's Note: We have asked the MIT administration to respond to Prof. Postol's article above and the one he wrote in the previous issue of the Faculty Newsletter. We hope to have their response for the next issue.

	Back to top
	Send your comments