Letter: Blame for 2000 bug

Sir: As a mainframe software professional of more than 10 years' standing, I agree with your correspondents (Letters, 23, 25 September) who have pointed out that the year 2000 issue is overwhelmingly concerned with old mainframe applications software. Such software, unlike that which runs on personal computers, is often written in programming languages which can easily manipulate an individual part of a date - the day, month, or year - in isolation. So instead of being tucked away as a set of routines which every program must use, like a single unique manuscript in a reference library, date-based calculations might be anywhere within the software. It is seeking out and verifying or correcting each of this multitude of calculations that costs time and money.

This might not be such a big problem if applications software were produced only by software companies and used without customisation by the vast majority of businesses. Leaving aside whether that would have been desirable in other respects, it never happened, because the software industry was still in its infancy. Instead many organisations employed a "do-it-yourself" approach.

So today companies employ staff of their own to look after computer systems. This is not without advantages, since many companies have a constant need to modify their existing software. The result is yet more "original work", and so yet more date calculations.

Consequently, in many cases, if a company does experience year 2000 problems, it will have no one to sue but itself.

DUNCAN AITKEN

Frankfurt am Main, Germany