Who will win the battle for the desktop? Microsoft Corp currently owns it with MS-DOS, and will do its darndest to hang on to it with Windows New Technology, IBM Corp reckons that it is so vital to its future to gain possession of it that it is betting the company on making a success of OS/2 2.0 – but it is far from certain that either of these two 32-bit contenders will be the ultimate victor. Unix – with the Unix System Laboratories Inc product code-named Destiny, currently the front-runner, is a very drak horse, and Novell Inc, whose success up to now has ridden on the back of that of MS-DOS, now appears to be an undemonstrative supporter of the Unix camp, with DR DOS as a place-holder that could turn out to be more important than most people currently imagine.
Mae West
Is history any guide? The 8-bit generation was won by CP/M with the Apple Computer Inc Apple II operating environment a very creditable second. The first 16-bit generation was won hands-down by MS-DOS with Macintosh System in second place – and winning a moral victory by setting the agenda for what should have been the 32-bit generation but in effect became the second 16-bit generation, which seems to have been won by Microsoft Windows, even though for the majority, MS-DOS is all most people want or use. The first thing that that litany reminds us of is that when it came to success, in the immortal words of Mae West (put into her mouth by scriptwriter Vincent Lawrence, goodness had nothing to do with it. CP/M was emphatically not the best desktop operating system available at the time, indeed it was not originally even written as a production operating system: like Unix, it started life as a development environment, and like Unix, it had intrinsic features that made it unsuitable as a production operating system. MS-DOS did remarkably little to improve on CP/M so was equally inadequate as a production environment. In each case, the Apple offering was demonstrably superior, even though the Macintosh System is by no means everyone’s cup of tea. The same strictures apply in spades to Unix, an environment that in its conception is quite unsuitable as a production operating system: its priorities were easy sharing of files and code and interaction between members of development teams and between teams working on different projects, where the priority for a business operating system is security and protection. Had IBM known back in the early 1980s what it knows now, it would have opened up its VM operating system, which also has its roots in development, but, coming from the button-down IBM world has many more security and management features than Unix started out with: too late now because the Open Software Foundation in particular is pulling out all the stops to get Unix under control, dump its unpopular policies, rein in its unruly socialistic element and render it electable as a production operating system for mainstream business data processing.
Populists
And does anybody now doubt that, as Computergram has been arguing since its foundation in 1984, the inadequacies of Unix as a business operating system are irrelevant: the weight of development money from such a vast army of different interests behind it mean that it will dominate what used to be known as the data processing department however much the purists suck their teeth in fastidious disdain. It’s called populism and we’re all populists now. And if Unix is destined to dominate mainstream data processing, doesn’t it make sense to run it at the desktop as well – why have two different – and ultimately competing – operating system worlds when the expressed ideal is to have everything working seamlessly together from desktop to the multiprocessor servers that threaten to supplant the monolithic mainframe? With Windows NT still a gleam in Bill Gates’ eye, OS/2 2.0 can lay claim to the goodness high ground, and it is difficult to see what more IBM could have done to make it electable – its erstwhile friends would say that it has cynically ditched all its principles – of closed systems, tight cen
tral control, the whole nanny knows best and you’d better believe it ethos that served the company so well for so long, and led to a string of successes at the ballot box in the 1960s and 1970s. But even back then in that golden age for the company, there were early forerunners of the devastating failures that have turned into a way of dismal life from the dawn of the 1980s. There was the Series/1 minicomputer, there was the 8100 distributed processor. The first was meant to unseat Digital Equipment Corp’s PDP-11 on the glow of the IBM name alone, but was sent naked into the market with no software and became the ultimate dogsbody product, a computing resource for any whim that needed one, a computer without portfolio that never found a central role. The 8100 is the more compelling indicator of what might happen to OS/2 2.0: IBM spent so much money developing both the hardware and a completely new operating environment and set of compilers and utilities for the thing that it had to recoup its investment, come hell or high water, before it could be killed off, despite the fact that within a couple of years, it became clear that what users wanted as a distributed processor was the machine that eventually appeared – fatally too late – as the 9370. IBM has invested so much in the development of, and so much more in the marketing of, OS/2 2.0 that it has to plough on with it even if in 18 months’ time it becomes clear that it is becoming at best a respectable also-ran in the desktop stakes. The problem there is that users that do adopt OS/2 2.0 wholeheartedly run the risk of finding themselves in the same fate of neglection that 8100 users suffered after the first glow of perceived success began to fade. There seems little reason why Windows NT should fare much better: there is no unarguable reason why the average MS-DOS user that has Windows 3.0 but doesn’t actually use it much, should decide to upgrade to Windows NT any time soon. One of the issues that has inexplicably failed to get even a whisper of an airing in the campaign wars of words is whether a graphical interface is the preferred environment for a large proportion of desktop users.
Don’t know
Those that like the Macintosh System love it: those that don’t like it detest the idea of that mass of graphical software between them and their applications, between them and the processor: like the nervous flyer, they hate the feeling that they are not in control, that if something goes wrong, there is absolutely nothing they can do to save themselves. They yearn for the feeling of security given by the alpha-numeric steering wheel and the clean, clear C: prompt. There is a very big market for graphical user interfaces but it is very far from being the whole market, and it seems highly likely that what the majority of today’s dogged MS-DOS users want is simply a few further refinements to the operating system they are familiar with. In the mid-1990s, object-oriented programming promises to bring the next irresistible revolution, but until that happens, no-one should be too surprised if the mass of desktop users splits into two camps, one that goes down the Unix-with-everything route, the other that decides as a matter of policy to remain in the don’t know came, judging MS-DOS with a touch of Windows here and there to be good enough for the next three or four years until the picture of the future becomes a bit clearer.