William John [anthropy@inwave.com] wrote:
>*** WHY *** would any outages occur? How can a simple date cause such
>problems?
Here's a simple example: a few months ago Swedish engineers tested their nuclear reactors for potential date problems. They discovered that any date in the year 1999 caused a shutdown, because the programmers had used a '99' date as some kind of end-of-data marker; this kind of thing is common practice. I can hunt out the URL for the news story if you want it, but it doesn't say much more than that.
Another: some equipment requires maintenance on a regular basis (e.g. yearly), and when the date rolls back to 1900 it can get confused and shut down for liability reasons. I've read of at least one piece of medical hardware which has shown this problem, but don't have a cite handy.
On a more theoretical level, suppose you have two systems communicating with each other, and the messages include a date. One is compliant, the other isn't. The non-compliant system sends its data with a date of Jan 1st 1900, but the compliant system knows it's 2000. Hence this is clearly invalid and it throws the data away. If that was important data, suddenly your computerized control system fails.
There are many similar possibilities; no-one yet knows which will actually happen or how bad the effects will be.
Mark