Martin Banks, Personal Computer World 04/87 - checked

April 1987

There is nothing like being doubly sure and well-protected: it must be true, because I read that once in a book. It is something I have often tried to keep in mind, sometimes successfully.

I was doing just that the other week. I'm off on this little trippette to the US of A and, as I am scheduled to be landing at Boston, I have been taking some time out to practice circuits and bumps at Boston's Logan Field airport with Flight Simulator.

Yes, I know an Olivetti M21 is not desperately like the flight desk of a 747, but there is nothing like being well-prepared, that's what I say. So round and round I went. After a bit of practice I got quite confident and, therefore, more daring. Needless to say, I got caught out and found myself lined up nicely to ditch in the sea. To my surprise, instead of going `splash', as per normal, the thing landed.

There is, I assume, a bug in my copy of Flight Simulator - actually, there are several, but then, what can one expect in a program that only costs some £30. That isn't meant to sound as snide as it seems, for there is every reason to believe that the bug-free program has never been written.

It is a sad fact that every useful program ever written has been issued to an unsuspecting public with all sorts of bugs in it. Given the nature of software this is inevitable, I suppose, for the human race is not terribly logical, especially when it is actually trying to be so.

Given this sad fact, what are the results. For example, it was my esteemed colleague, Guy Kewney, who pointed out some time ago that the US Strategic Defence Initiative, `Star Wars', was really quite frightening given that there would probably only be the one chance to try out the software in its working environment, and that past history in software did not bode well.

He quoted MS-DOS as a prime example. Even with hundreds of thousands of users feeding in their observations to Microsoft, it still comes up with the occasional bug, even now. The biggest and the best in this business cannot create software that is bug-free. Indeed, it has been said about IBM that it has turned the `bug' into a sales advantage. If enough users complain about a bug which proves difficult to cure, then it is said that IBM marketing labels it as a new `feature'. An increasing number of software companies now talk in terms of 'undocumented features' in their software: you can guess what they mean.

Yet what does the user get from all this. For the mainframe user with a staff of programmers waiting to maintain as well as create applications, bugs are something that are planned and accounted for. But the average PC user doesn't always have such resources. Certainly, there will be a coterie of users who are aficionados, who like getting the hands dirty by diving into the code of their latest application acquisition.

But for the majority, all that they want is the apparently simple objective of a program that works in the way they expect it to, every time they use it. We all have personal experiences of bugridden software, or know someone who has. Most of us journalists have sat in press conferences and sniggered gleefully as some over-hyped application program crashes ignominiously during its launch demonstration.

Most often, the cause is something simple, such as in one desk-top publishing package I heard about that has a small bug in its pixel handling routine. This causes the displayed horizontal image to gently and artistically turn vertical at the horizontal scroll command. Sometimes, however, the cause is more fundamentally stupid on the part of the authors.

I remember, for example, a story of an accounting package written by a software house specialising in scientific applications. They wrote it in the language they knew best, Fortran, just about the least suited language to accounting applications. The result was a package that produced the most amazing invoices, as it multiplied the quantity ordered by the line number, and then by the part number to give a value.

Whatever the cause of the bug, however, the end result to the user is normally the same - aggravation and inconvenience. It has been argued before that the PC software industry could and should do better in ensuring that its products reach the market in a satisfactory condition. While many companies do try, there are enough of the other sort to make users suspicious of all applications.

What is worse, many companies offer poor to non-existent levels of support to the user when a bug is discovered, even an old and well-charted one. You telephone to report the problem and, if the phone is ever answered, you can be met with enough tortuous ducking, weaving and buck-passing to rival the most complex set of nested sub-routines.

Occasionally I hear of software companies that have offered users highly praised support and have been able to trace and cure bugs both quickly and efficiently. There seems to be a common theme in these operators; their product are in specialist, vertical markets, and they are expensive.

Now I know this is one of my favourite hobby horses, and that I am about to get on it again, BUT... you do get what you pay for and, given that bug-free software is a practical impossibility for now, paying for support by spending more on the purchase price is maybe an important step. While the economies of scale that PC sales volumes bring do affect product prices, you still can't get something for nothing.

Actually, in this case you do get something, though often it is just a can of worms.

end