Martin Banks, Personal Computer World 02/88 - checked

February 1988

How often, I wonder, have you sat eyeball-to-phosphor with the display of your computer and said to it: `damn it, why do you keep asking me the same stupid questions? You know what I want to do!'

The computer, of course, doesn't hear. More to the point it wouldn't care even if it could, for a computer is a Jobsworth. You know the type...`its more'n my jobs worth to go against what the program tells me.' It then sits there, smug within its all-pervading sense of self-righteousness and seems to actually enjoy the fact that you are now going, ever so gently, stark, staring, raving mad.

You hate the damned box. Every time you turn it on and load up your favourite program you have to go through a set of keystrokes that establish the options you want. You always want the same options (well nearly always), yet every time the stupid box keeps asking you to pick a number from the menu, asks you seventeen times whether you're happy with this choice, then moves to the next menu and starts the process all over.

Yes, I know all about computers being morons that just do what they are told, and that programs have to follow a logical sequence that gets them to do it all nice and tidy. But, and if this is a disablist remark then I apologise in advance, even the dumbest human moron can be taught to do repetitive tasks.

With computers, we seem to have forgotten that this might be a good idea. Or perhaps we have just never thought of it, which is a shame, because I for one am well aware that there would be a big market for such a product....me.

I am happy to accept that with the early days of personal computing, there were two synergistic strands most prominent. One was the relatively puny power of the hardware and the first nervous questings of operating systems. The second was the downright keenness of the original users to dabble. These were people who predated the modern hacker in aspiration as well as time. To get their box working at all was often a substantial success.

Today we think of personal computers in an entirely different light. When there are machines coming on the market, especially at the business end of the spectrum, that are truly tantamount to a minicomputer on your desk, then surely we must look for new ways of helping them to work to our advantage.

Yet are we? There has been much in the press of late, and even I have written some of it, which speculates on which operating system might win out in the great race for the new 386-based PCs. Will it be OS/2?, could it be Unix?, might Concurrent DOS be the dark horse? will Windows 386 take the day? Who cares?

I suddenly wonder whether many of the users really do, for within certain criteria, I suspect they may now be looking for slightly different capabilities than the software companies are thinking of providing. One of those capabilities is, I suspect, the ability to learn.

I suspect it primarily because it is what I would like to see, but I also suspect there is a huge potential market out there, waiting to be picked up. While the ability to perform the most amazing feats of disk memory management at a single keystroke may be useful, and multi-tasking existing applications a very good idea, the ability to `learn' what the user normally does and make it happen seems such an obvious facility. With the power available in these new PCs, it is surely a feasible proposition.

Surely it is not too difficult a task to build into an operating system a function that will monitor the commands that you use and, after a while, build up a pattern of activity that allows the program to predict with increasing accuracy what you might want to. I am not suggesting that it should take things that one stage further and do it, for there will always be the exception when you want some different function performed, but...

Here is an example. I usually run my PC with the memory divided in two. Word processing runs in one partition and the other has no active program. If I want to find out how many words I have written so far (828 to be exact) I SAVE, then switch to the other partition. There I load up a small fudge-factor program that includes a word counter. This asks for a filename to check, and then it is kind enough to give me the information I desire.

This is OK, but is has its drawbacks. In a long document, it can mean repeating that process many times, just to see how things are progressing. And every time, the full keystroke sequence has to be completed. What if the operating system, on detecting the initial pattern of SAVE and switch-partition keystrokes, offered me the program and then the same filename as defaults; wouldn't that be convenient?

I feel sure that most other users have a similar tale to tell about their own applications, be it in work or leisure use of the box. The scope, therefore, is not in putting such learning facilities in individual applications packages, but in writing them into the fundamental capabilities of the operating system. There must be a huge number of users out there who would just love to have such a facility.

There is, I suppose, no hope for us old hands working our old PCs. DOS is too long in the tooth now to learn some new tricks. But there must be hope that the new operating systems coming along will take the chance to add something really useful, something that makes day-to-day computing a great deal easier.

After all, wouldn't it be better to get the users running one task well before we fuss too much over running lots of them at the same time?

end