A couple of years back, at a time when the US authorities looked serious about reining in Microsoft, there seemed a real possibility that the company would be broken up. There was a precedent in the 1984 break-up of the US phone giant AT&T into seven 'Baby Bells', and it was argued that Microsoft had similarly grown too big for anybody's good.
    There was also a very good argument for holding the company together. Microsoft gained its near-monopoly because computing needed a standard platform; and, whatever you may feel about its business practices, it has done a pretty good job over the years in providing a mix of stability and innovation.
    I fear, nevertheless, that the US courts did both Microsoft and the industry a disservice by not forcing it to split into two. I got to thinking about this after hearing Palmsoft chief executive Dave Nagel talk about the next-generation PalmOS, codenamed Sahara, which is clearly designed to head off Microsoft's encroachment into mobile devices.
    A major reason for the success of Palm handhelds, apart from their ground-breaking pen interface, was the ease with which they shared information with a base machine.

Contacts and appointments data held on the base and mobile machines would be reconciled on both whenever you slotted the Palm into its cradle. The word synchronisation was misapplied to this operation, but I guess we are stuck with it.
    Nagel boasted that Palms are still better than Microsoft's own pocket software at synchronising with Windows apps such as Outlook. I won't argue the toss on that one, but it begs the real question of how long vendors using Palmsoft, Symbian, Linux and other mobile software can compete with Microsoft if it exploits its dominance of desktop and backend systems to gain a competitive edge for Microsoft-driven mobiles.
    In a normal company this would be a legitimate business practice, but Microsoft's extraordinary position was recognised at last year's anti-trust trial, when it was forced to undertake to open its application programming interfaces (APIs) to rivals. Unfortunately it has almost infinite scope to fudge the issue.
    Earlier this month I went to a briefing on Microsoft's Exchange Server 2003, the latest version of its enterprise-level 'post office' software which, among other new features, has been tailored to provide remote synchronisation with a variety of mobile devices.
    Microsoft's UK mobility business manager James Langridge told us that Microsoft was 'considering what to do' about opening the Exchange APIs to allow rival vendors equivalent facilities to Microsoft client software.
    Pressed about this later, he said: 'The trouble is that there are no standards for this kind of thing. Nokia and others agreed SyncML but that is their proprietary standard.'
    I suggested that there was no global standard because Microsoft had avoided signing up to one. Langridge pointed out that Microsoft is a member of the Open Mobile Alliance, the industry body that is supposed to sort out these matters. I recalled the hollow laughter among representatives of that body at Comdex last year when I raised the question of its participation.
    Now, for all I know, Microsoft may be behaving entirely honourably in this matter; but the longer it obfuscates and delays, the greater the edge it can give its Pocket PC software, and the stronger the suspicion that it is doing so on purpose. You can be sure that it would already have those standards set, and the APIs open, if it saw this as being in its interest. If, for instance, its mobile operations had been sold off as a completely independent company.
    There is a natural demarcation between fixed infrastructure and the mobile; desktop and particularly back-end systems are older and necessarily set in their ways because they are too important for rapid radical change. The mobile, in the broadest sense of an untethered networked machine, can use any hardware or software so long as it can communicate with the global infrastructure in a standard way.
    Microsoft would argue, as it has in its subtle campaign to stifle Java, that it can produce more powerful software by using non-standard features in client devices and servers. This would be a valid point for just about any other company; but when Microsoft goes proprietary, even (saints preserve us) with the best of intentions, it is necessarily locking people into its software.
    You may think that I exaggerate Microsoft's strength. Exchange leads its market, but with a share of only 32 per cent. IBM, which has a respectable 25 per cent with Notes, has a collaboration deal with Palm. So does Novell, which claims 30 million users for its Groupwise. There is also a promising Opengroupware.org open-source messaging project.
    Nevertheless it seems to me that both Microsoft and computing would be better off if the company drew a big white line at the desktop and sold off everything beyond it. Microsoft does not need the mobile market in order to make money. It will earn more than enough by allowing mobiles of all makes to link with its desktops, and use its back-end systems and its vaunted .Net services (which also favour Microsoft-driven mobiles).
    Both Microsoft and its spin-off would be freer to innovate, because they could do so without being accused of anti-competitive practices, and the entire industry could share the innovation.
    Of course any split-up would be a complex operation, with much to sort out in the small print (a laptop, for instance, is both a desktop and a mobile) but Microsoft could choose the terms if it split voluntarily; otherwise, there may come a time when others will choose the terms for it.