Catastrophic failure - Washington Post reported story on SDI software study

National Review, June 10, 1988

Catastrophic Failure

The Washington Post recently ran a front-page story on SDI with the headline CATASTROPHIC FAILURE LIKELY IF EVER USED. The Post was reporting faithfully on a study of SDI software prospects by the Congressional Office of Technology Assessment (OTA), which advises Congress on technical issues relating to public policy. In a White House briefing, General Abrahamson, director of SDIO, called the report's conclusions "nonsensical," according to the Post. Why did the normally affable general use such strong language?

Actually, the facts about SDI software are encouraging. Original estimates, four or five years ago, cited a need for up to 100 million lines of code. But now, after some serious study, General Abrahamson said in congressional testimony that only four to six million lines will be needed. Of that, less than two million lines suffice for the most important part of the defense - the boost phase, which shoots down ICBMs as they rise into the air, before they have deployed their deadly warheads. That is, relatively speaking, a snap.

It is easy to confirm that a few million lines of code are sufficient for the critical boost-phase defense. Software requirements for battle management of missile defense divide into a number of discrete packages, each dealing with a separate function. One package covers situation assessment (Is an attack underway? What kind of attack? Directed at which targets?); another calculates the tracks of enemy ICBMs; still another calculates the timing and direction for the release of the "smart bullet." Other software packages control the spin-up of the gyros which keep the stellite platform steady or change its direction, and still others do satellite housekeeping and station-keeping. Finally, one package operates the programs for self-defense.

How daunting is two million lines of code? A famous SDI experiment in 1986, called the Delta 180, required one million lines of code. The Delta 180 programs involved tracking rockets in space, maneuvering to intercept targets in space, and homing in on the targets to destroy them by collision. These are the same programs needed for boost-phase defense in SDI. Yet the million lines of code were written in six months and worked perfectly in their first trial. One begins to see why General Abrahamson was incensed by the OTA study.

The OTA was also worried about how the software could be tested short of war; but as the Marshall Institute Panel on Missile defense has pointed out, "software is the one thing that can be tested. When signals are fed into the front end of the program, they look exactly the same to it regardless of whether they have been produced by a Soviet missile leaving its silo or by a piece of equipment that generates signals imitating the real thing."

It is a pity that General Abrahamson's staff issued a politely laudatory release on the OTA study, instead of telling the public why he regarded the OTA report's conclusions as "nonsensical."