Anything you can do I can do better: A turf war is looming in the computer industry as key players attempt to emulate each other's software programmes. Andrew Hawker reports

Click to follow
The Independent Online
For pounds 5, it is possible to buy a kit to convert your IBM PC into the equivalent of a Sinclair Spectrum or Commodore 64. The benefits of this may not be immediately obvious: after all, if you buy yourself a Porsche you do not usually try to convert it into a go-kart. However, the ability to carry on running old favourite versions of Pac-man and Newsboy evidently has some market appeal.

Kits such as these are known in the trade as emulators. Emulators have a chequered past, and have hitherto played little part in the mainstream of the business. Today, however, emulation seems set for a big revival. This has little or nothing to do with children's computer games and everything to do with some major turf wars looming between the key players in the industry.

Emulator wars first broke out more than 30 years ago. In 1963, IBM was coasting along nicely on the success of its 1401 commercial processors, when rival Honeywell announced a range of machines equipped with a 'Liberator'. The Liberator enabled IBM users to switch to the new Honeywell range without having to modify any of their programs. From then on, the trick was to ensure that new users only wrote Honeywell-

style programs, and so became locked in as Honeywell customers.

IBM responded with a crash programme to bring forward its next range of machines, the 360s. Almost as an afterthought, and after a lot of internal wrangling, the smallest 360 machine was also given a 1401 emulator. The message for the industry was clear: winning customers was not just a matter of offering price and performance. It was also vital to make it easy for them to switch their allegiance to a new system.

IBM was later to be hit with some heavy competition from compatible machines, which mimicked the original as closely as possible. The best example was the IBM PC, replicated in millions until clones outnumbered the genuine article. These machines did not usually attempt to switch-sell the user on to an alternative system, although some suppliers did try to market multi-standard PCs. For example, in 1984 a company called Micro Craft launched a machine capable of running IBM, Apple, Tandy and Unix software. It proved unable to perform particularly well under any of its various split personalities and sank without trace within a few months.

Today, everyone is eyeing up a market that, in the space of two or three years, has become totally dominated by Microsoft's Windows operating system. The challenge facing rival suppliers is to lure all these users away on to the alternatives developed by IBM, Apple, and the 'Open Systems', Unix-based providers. And so, all of a sudden, everyone is back in the emulator business. IBM has added a Windows environment to its OS/2 operating system ('Your Windows could use a little fresh air,' trill the ads, also announcing deeply discounted prices). Apple has announced the PowerPC processor, capable of running both traditional Macintosh software and Windows ('Nobody expected a Macintosh that would run Windows'). Unix providers have developed Windows emulators with esoteric names such as Softwindows and WABI (the Windows Application Binary Interface). Meanwhile, Microsoft is sending out reassuring noises to the effect that all new systems will continue to run existing Windows applications. To do this, Microsoft, too, will have to emulate its previous software.

All this sounds like good news for the consumer - a 'win-win' mixture of competition and continuity. But experience suggests that users should be wary about stepping into the emulator world. There are overheads involved in setting up an emulated environment, for example, and these can have a big impact on system performance. This means that moving to a machine with a faster chip or a bigger memory can fail to deliver any noticeable improvements in the processing throughput or may give erratic response times. Most emulators on offer have come under fire on this count, including one used by Microsoft itself to steer customers from the original version of Windows to its newer NT version.

There is also the problem that no software product is ever perfect, and so it has to be subjected to constant corrections and improvements. Some of these may be quite trivial but may still cause problems in an emulator. Earlier this year, for example, Microsoft and IBM got into a spat over relatively small changes that Microsoft introduced to fix minor errors or 'bugs' in their code. Microsoft claimed these were needed to improve the reliability of the software.

There is perhaps one small chink of light for users from an unlikely quarter: computer viruses. Viruses are a good index of success, since nobody bothers to write them for unpopular software. The basic DOS code that still lurks inside Windows has long been a major target for the virus-writers. The appearance of 'pseudo-versions' of DOS and Windows in emulators should make life harder for them, although in most cases they will only see it as a fresh challenge. So you can either opt for an off-beat emulator crashes occasionally because it has fundamental disagreements with the software around it, or you can go for a more mainstream product.

The latter will sooner or later attract viruses, capable of disabling your system just as dramatically. The unpalatable nature of this choice is alleviated only by the prospect that defences against viruses are improving all the time, whereas no one has yet figured out an easy way of resolving conflicts between software products written and devised by entirely different people (particularly if they are commercial rivals).

'Beware of imitations' used to be a tag-line on branded goods. It may be that Microsoft will resurrect it. The only problem is, given the speed at which the computer industry moves, it is going to become increasingly difficult to decide what, if anything, has a real claim to be regarded as the genuine article.