Devil's Advocate: A standardised PC on every desktop?

The last two decades have been the era of the PC but, argues Martin Brampton, this platform is another example of computing failing to deliver...

It is really rather odd. Almost every large organisation puts personal computers on every desk. But then they go to enormous trouble to defeat many of the key features of the PC. Achieving this takes enormous effort and a significant slice of IT budget. Why do they do it?

There were many problems with early PCs. Notable among them was the inability of the processor to access very much memory or to multitask properly. But there was a reason for these deficiencies: the Intel processors used in the first couple of generations of the PC had never been designed for general purpose computing. They were intended as embedded processors for peripherals.

Once Intel saw how their products were being used, they started to design processors suited to the purpose. Of course, they were constrained to maintain compatibility with what had gone before because by then there was a huge installed base. With software, though, it seems that even after a couple of decades, we are still tied to old ideas.

It may be that the idea of a personal computer is one that can be controlled by its user, who is free to configure it how they like and to select whatever software they want. There might be benefits in that approach. Yet most organisations try very hard to impose a quite different principle. From a management point of view, it is desirable that the PC should run with a totally standard set of software and contain little or no local data.

Even the users, deprived of the freedom to configure the PC as they please, often come down in favour of standardisation. When the choice is between choice and reliability, most people opt for reliability. And most organisations have at least some degree of free seating these days. However keen one may be to personalise one’s own system, somebody else’s quirky choices are anathema.

If we stand back for a moment from current concerns, it becomes obvious that the development of desktop computing needs something different. As networking became commonplace, it was apparent that the requirements for a network workstation were entirely different from those of a standalone machine.

For a brief period, this could be handled relatively easily. While DOS remained the mainstream operating system, it was relatively easy to relegate the network workstation to be a slave of the network. Practically all of the software and data could be held on network servers. With graphical operating systems, this has become progressively more difficult.

Simple expedients, like creating standard software images, are rendered exceedingly complex by a huge range of hardware dependencies. Desktop operating systems lack layering, despite having exceeded the size and complexity of mainframe systems some time ago. So IT support staff struggle to keep systems to some degree of consistency, without which neither reliability nor efficient rollout of new software are possible.

Citrix has pushed hard and somewhat successfully with a server-based solution. Hardware and software costs tend to increase, although management and support costs can be driven down. Well-designed installations can provide a good and flexible user experience. Yet this is still a patched solution rather than a root and branch reform.

Really, this seems the clearest indication that in IT, the market can often fail to deliver what is actually needed. The various entrenched commercial interests remain reluctant to deliver genuinely innovative models of computing that would take advantage of the ubiquity of networking and the organisational goal of intelligent and consistent management of information. With a truly constructive new approach to the problems, perhaps we could start to see real benefits from the amazing developments in hardware technology.