Over the past few months an interesting topic has appeared now and then on the Internet and in computer magazines. The subject appears to be a joke but, I assure you, it really is not. This great problem concerns the discussion that is going on about what will happen at midnight on 31 January 1999. How many computers in the world will be able to handle the 'date problem'? Quite a few mainframe computers were constructed when people were not thinking as far ahead as 30 years. Neither the software nor the hardware was meant to handle anything like a radical new format for the date. Some older machines are really going to have problems. There is, however, an amusing problem already occurring now. Some people are worried their PCs will also be unable to handle the problem so they have been testing them by changing the date. There is nothing wrong with wanting to see what will happen but there are a few things to be careful about. Some systems employ a kind of self-cleaning application that automatically deletes any file that is older than three months or half a year. If you change the date to 2000 when such a program is running in the background, all your files will be out of date and may be erased. This is not the disaster it sounds. There is no computer system I know of that really 'deletes' a file. It is almost always to recover erased files just as long as you have not saved any new ones. This was a lesson Colonel Oliver North learned the hard way. It does, however, make one think about all the possible problems that can occur when we ask our machines to do things for us automatically. I have always been sceptical about letting a machine do too much for me. It is a bit like working out a calculation in your head. You know all the steps you are taking and do not rely on any mechanical or electronic assistance. There are times, of course, when the calculations cannot be done in our heads and we must rely on the machines. It is most unlikely that a man would have walked on the moon without computers. Nevertheless, we can sometimes become too believing in what they do. We are asking them to do more and more and we become dependent on them. This is unlikely to change. It does not mean that we must become dumb operators, though. There are many things computers are not going to do for us. They are not going to know that a simple change in the way we reference a date may throw everything out of kilter. The two areas of computing that have failed the most have been artificial intelligence and machine translation. The Japanese have spent millions of dollars on machine translation programs and have so far produced nothing of value. The problem is that there are not always one-to-one relationships in language. In fact, we rely on this fact for most of our great literature and certainly much of our poetry. In science the enemy is ambiguity but in art it is a friend. I enjoy computers and I especially enjoy trying to program them to do strange and wonderful things. I am aware that certain things such as air traffic control will have to depend on them and that means that thousands of lives will depend on them. There are certain things the computer will always be much better at doing than any mere mortal. This does not alter the fact, however, that they can completely 'misunderstand' what it is we want when we instruct them to do something. Being sceptical is not the same thing as being a luddite.