Martin Banks, Personal Computer World 01/90 - checked

January 1990

Talk is cheap, they say. It can also be effective, confusing, infuriating, obfuscatory and accurate - occasionally all at the same time.

It is also the way forward for a long term, enduring and deeply meaningful relationship with your personal computer. Forget all that nonsense about graphics and icons and mouses and such. It is all a waste of time, effort and precious computing resources.

If you want to really get to know your computer, work with it effectively, cajole it into finding that one little item in your database that you know is there but the program stoically refuses to find, then talk to it. Swear at it, even. After all, in such circumstances you probably do anyway.

I surely cannot be alone in sitting, deep in mordant frustration, yelling 'work you recalcitrant little ....... (expletive of your choice already deleted)' at my machine when it won't do what I want it to do. The reason it won't, normally, is because I have once again excelled myself in complete burkery and done something stupid.

Perhaps it would also be good if the machine could tell me, in a deeply soothing voice, that, perhaps, if I felt able to enter a colon instead of a semi-colon, just there, after the disk drive name, we'd all get along a good deal better.

Speech between computers and us has long been the dream of sci-fi writers from Asimov and Arthur C Clarke onwards. Every so often, the subject raises its head again, though rarely does it stay long above the parapet. But at a recent conference in Boston, the head of the Media Laboratory at the Massachusetts Institute of Technology, one Nicholas Negroponte, gave it a good airing.

He it was who also suggested that all the efforts being put into graphical user interfaces was a waste of time and effort. His contention seems really quite simple, and does address a personal doubt that I have held about such machines.

In essence, he feels that the whole idea of a 'desk top metaphor' doesn't work, and it doesn't work because it cannot, in practice, emulate a real desktop and the work that goes on there. The problem, according to Negroponte, is simple: a desktop metaphor cannot handle large quantities of data, especially large quantities of separate lumps of large quantities of data. Yet that, if you look at any desktop where someone is doing some real work, is what life is really like.

The only time any desk I have been intimate with looks as neat as a windowed display is just after starting work there for the very first time (or just before leaving rapidly for the very last time). Otherwise, mine, and every other desk I have seen (except for a few titular-head company chairpersons') is piled high with, well, everything and anything. That is the way people really work.

What is more, they are surprisingly good at knowing that a vital piece of paper is 'here somewhere'. But finding it is a much more complex task than simply sifting through piles. It means unconscious acts like keeping the other data to which the document relates within peripheral vision so that it is still 'in mind'. Try that on a 'desktop metaphor' system. My personal doubt about desk top metaphors is that, in practice, by the time you have found the data item, you have forgotten why you needed it, so the process starts over again.

The alternative, Negroponte suggests, is speech. For example, solving the problem just outlined would be a simple verbal instruction to 'go find X'. Once found it could be displayed on a screen or, perhaps, read to you, especially if it's bedtime.

It has been tried before, of course, and has failed miserably. Texas Instruments did it many years ago on a PC. They included the ability to drive 1-2-3 with voice commands, plus the trick of recording speech digitally so that it could be attached as a file to an Email document. It was not desperately effective, however, and never caught on.

But the technology has improved. It is possible that the power which will be available in the successors to such processors as the Intel 486 and the Motorola 68040 could also be exploited to drive workable speech interface and interaction front-ends. They could also provide all the other bits as well, such as transcription of the spoken word.

It even poses an interesting possibility. Given that most of you would, I feel sure, agree that the best, most meaningful conversations you have had with someone else have been when the lights were out, full speech interaction does raise the possibility of not requiring a display at all, or at least not very often.

Take the telephone, for example. Using it we can communicate effectively with just about anyone, just about anywhere. With the coming of Personal Communications Networks (PCN) in the 90s (where people rather than the handset will have a telephone number) we will be able to do it even more. We can do it effectively without needing to see anyone or anything - yes, sometimes both can be advantageous, but such occasions are in the minority. So, why not interact with the computer in the same way?

That could get rid of a fundamental problem with the beasts. Displays and keyboards mean that a truly personal computer still has to be big to be effective. Uncle Sir Clive's Z88 is still about the minimum useable package in my view, despite the coming of the Poqet and Atari Portfolio. And even if the latter are a better size, most of the space is taken up with keyboard and display technology.

Might it not be better to use that for a larger memory, faster processor(s) and the like? All that would be needed for I/O would be a tuppeny-ha'penny microphone and a tuppeny ha'penny ear piece.

And for communications with the rest of the world? Well, why not build the whole thing into one of the new PCN personal telephones together with an ISDN communications capability. Then you could talk to your computer, your friends, your friends' computers, your computer's friends (a cuddly and understanding database engine being very laid back in California) and your bank manager, who would undoubtedly go ape about the line charges.

Only occasionally, perhaps back at the office, would you need to see a display. In fact, according to Negroponte, it might be better for the computer to see you rather than you it. If it could see you, it would know whether you have simply stopped work for a while but are still there keeping it company, or have departed to Paris for the weekend and forgotten to switch it off.

Alternatively, it could learn that it has displeased you, not by the tone of your voice, but by the internationally understood extended digit waived in front of it.

end