To the young man's surprise there was an extra trigger for no obvious purpose. The pilot said it was the software reboot trigger. Yep, they actually reboot software mid-flight, during battle, and sometimes more than once.

So here was a $40 million machine with flaky software. I tried to get more detail and was told that it was not uncommon and did not constitute a primary risk or degrade the machine's flying ability. But I couldn't help wondering about other defence systems, civil aircraft, nuclear reactors and my car. Yes, my PC can lock up from time to time, but please, not my engine management system. I already have a car hi-fi with a mind of it's own and no reboot mechanism. The garage is trying its best to fix it, but the only way to reboot is to disconnect the battery. Soon I shall resort to cutting into the wiring harness to install a trigger on my gearshift.

Looking back to my childhood, I recall when most technology was unreliable; the radio, TV, record player, cameras, clocks, wrist watches and electricity supply - everything was flaky. Driving a car 50 miles was a major expedition, and switching the radio on was a bit chancy too.

All of those technologies are now phenomenally reliable, but it has taken over 100 years of commercial development, competitive markets and consumer pressure to achieve. Mechanical, electromechanical, electrical, electronic and production engineering have come a very long way, as have design and manufacturing.

In contrast, popular software has been around for only a couple of decades, and as an industry it lacks competition. Programs tend to be unique, non-interoperable and non-interchangeable, and there are no common operating platforms or systems. However, among some classes of software and system there are signs of a stability. Watches, calculators, pagers, mobile phones, washing machines, dishwashers and other small and contained systems have perhaps come to the end of an evolutionary road and are stable. Thankfully, engine management, TV and hi-fi systems are also in this category.

But large systems, like the PC present major difficulties. Loading new applications and upgrading software still makes you hold your breath. When will that change? Only when there is sufficient competition, uniformity, compatibility, and we complain enough, and have the option to vote with our feet.

Network computing might just be the opportunity we need to bring about the necessary change. A universal operating system and language with an effective infinity of vendors may be on the horizon. Then we might get a universal and basic word processor, graphics package, spreadsheet, mail, plug and play, and much more. All of it honed up and non-flaky, and no reboot triggers.

Peter Cochrane holds the Collier Chair for the Public Understanding of Science & Technology at the University of Bristol. His home page is:
http://cochrane.org.uk