Friday, September 07, 2007

STAR WARS TO OUR WARS: The Achilles Heel

Reagan had assembled a large committee of eminent scientists to advise him on Star Wars. One of them very publicly resigned. He was the man I wanted to meet.

I first saw him on some late-night tv news show, discovered he was coming to London to deliver a paper at a software engineering conference at Imperial College and take part in a discussion on Friday 30th August on 'The Technical feasibility of software for strategic defense initiatives.' I interviewed him after his presentation and wrote the following piece for Time Out magazine, which appeared on Sept 18th.

'The unlikeliest and sternest critic of Reagan's Star Wars programme is Professor David L Pamas, a short, irascible, bespectacled, ginger-bearded professor of computer science, wearing a navy blazer and brown boat shoes with fraying laces.

In London for the Eighth International Conference on Software Engineering, Professor Parnas does not object to the Strategic Defense Initiative (SDI) on moral grounds. He has worked on de­fence-related projects at the US Naval Research Laboratory in Washington for some 20 years and is not a liberal. His objection is that it won't work.

Software engineering is a field in which he's considered a giant by his peers and explains the importance one should attach to the fact that he recent­ly resigned from the SDI computer panel, the first scientist on the project to do so. His resignation letter read, in part, 'I do not believe that further work by the panel will be useful and I cannot in good conscience, accept further pay­ment for useless effort

In an exclusive interview and in his barnstorming speech to the Conferen­ce, crackling with intelligence and hum­our, Professor Parnas outlined his argu­ments. He knows from experience that in a complex engineering project the computer software is always the most unreliable part. He says: 'You never know when you've got the last import­ant bug out, so what is our basis for confidence in SDI when all defence sy­stems fail first time? Realistic testing would require a series of nuclear wars.'

Parnas thinks analogies to the Apollo missions and the Space Shuttle programs are misconceived. The software for SDI is much more difficult, he says, though not perhaps the longest pro­gram ever produced as some have sug­gested. 'On Apollo we could predict the behaviour of the moon and there were no decoy moons to confuse us. With the Shuttle there were manual fall­backs and no surprise launches. With SDI we have no reliable information on targets.'

Nor does Parnas accept the compari­son of the Manhattan Project which, unlike SDI, was based on theoretical principles that had already been made. He believes that if Reagan had pro­posed Faster Than Light Flight — a scientific impossibility — that half the defence scientists and corporations in the US would have signed up for con­sultancy fees ($10,000 a day is the going rate on SDI) and lavish research grants. When accused of negative thinking he says: 'Pessimists have been wrong before. So have optimists.'

Parnas first heard talk of an SDI-type system at the Los Angeles aerospace factories in the '60s. He remembers the first US ABM system called Start (ne­gotiated away at the SALT talks) and the US Navy's Advanced Avionics Digi­tal Computer, a misconceived project that took three years to kill off. He has also witnessed first-hand the problems with ADA, the Department of Defense project to produce a common computer language for all its forces.

Parnas says: 'Personally, I would feel safer if they spent the money on tanks.' Pretty soon the defence establish­ment is going to have to start listening.

No comments: