|
|
|
| Current Issue |
|
|
| Resources |
|
|
|
Why I Won't Work on Software for SDI
| Article
# : |
11046 |
|
|
Section : |
MODERN THOUGHT
|
| Issue
Date : |
6 / 1986 |
907 Words |
| Author
: |
David Lorge Parnas
David Lorge Parnas is Lansdowne Professor Department of
Computer Science at the University of Victoria in Canada. |
I am a computer scientist who specializes in software engineering. I have published more than ninety papers on computer system design, quite a few of them widely cited and reprinted. I have had twenty-five years of software-construction experience, including fourteen years of experience as a consultant for the Department of Defense and its contractors. My work on large-scale programming, hierarchical organizations, simulation, and cooperating sequential processes has influenced many Department of Defense efforts.
I was appointed by the Strategic Defense Initiative Organization (SDIO) to serve on a panel advising them on computing research in support of battle management. I resigned because I came to believe that no system of the sort being considered by SDIO can ever be trusted.
SDI depends on computers. Computers must detect missile firings, determine the source of the attack, and compute the attacking trajectories. Computers must discriminate between threatening warheads and objects designed to confuse our defense. Computers must aim and fire the weapons. Software controls the computers. SDI, therefore, depends on the software.
Software for SDI is far more difficult to create than any software system we have ever attempted for the following reasons.
1.SDI software must be based on assumptions about target and decoy characteristics; those characteristics are controlled by the attacker. We cannot rely upon our information about them. The dependence on any program on those assumptions is a rich source of effective countermeasures. Espionage could render the whole system worthless without our knowledge. It could show an attacker how to exploit the inevitable differences between the computer model on which the program is based and the real world.
2.Overloading the system will always be a potent countermeasure because any computer system will have limited capacity and even crude decoys would consume computer capacity. An overloaded system must either ignore some of the objects it should track or fail completely.
3.Satellites will be in fixed orbits that will not allow the same satellite both to track a missile from its launch and destroy it. Responsibility for tracking a missile will transfer from one satellite to another. Also, because of noise created by the battle and enemy interference, a satellite will require data from other
...
Read Full Article
Look for this article in Ask.com
|
|
