Title: Editorial

Have you noticed the dramatic fall in the hardware costs over the last couple of years? Some parts are hard to explain. A couple of weeks ago I went out to buy an extra hard-drive because I had some software that I wanted to look at but did not want to destabilise any of my existing setups (that says something about what experience has taught me to expect). I found myself buying a 17Gbyte drive for a price that I would have been hard put to have matched for a 3Gbyte drive two years ago. Leaving aside the question as to why anyone needs this storage capacity, ask yourself why the price has plummeted while the capacity has soared.

Now look at the development and cost of CPUs. For years Intel has largely defined the market while others competed on cost. To be honest chips such as AMD's K6 were fine at the low cost end of the market but if you needed SMP (symmetric multi-processing) and x86 compatibility Intel was your only choice. Intel's competitors had to ensure compatibility at both software (i.e. machine code level) and hardware (plug compatibility) levels. Then Intel (in my opinion) made a serious tactical mistake, they defined slot 1 for their chips and denied their competitors access to it. This created a divide in the market, AMD etc. could no longer remain hardware compatible and that opened the door to their providing added value in other ways such as enhanced code sets. Their success in the low cost end of the market began to scare Intel who promptly released the Celeron as a cheap and emasculated alternative that would work on slot 1 motherboards. Too late, the damage was done. Why do you think that the Celeron now has so much added functionality and such a high clock rate?

Much worse was to come. As AMD no longer needed to consider plug compatibility with Intel they could take a more radical look at the design. The K7 promises to be a radical departure because such things as the bus architecture (particularly for SMP) are not tied to the past. Intel has the problem of remaining plug compatible with itself. AMD never supported SMP and so has no such constraint.

The curious feature here is that head-to-head competition has been less effective than the version where the user has to choose which branch to go down. One side effect seems to have been the almost complete demise of the other competitors.

None-the-less we seem to be getting more for much less money because there is now real competition between David and Goliath. The same thing seems to be happening in other hardware areas. The competition in the graphics board market can only be described as frenetic with the market lifetime of products down to less than a year.

So let me turn to software which is what you are more interested in. Now we have a serious problem. A single major producer is responsible for the most widely used OSs, development tools and desktop applications. These have an unhealthy inter-relationship that seriously inhibits competition. That, I believe is something we should feel strongly about.

Microsoft applications are both feature rich and bug ridden. Fixing bugs does not generate revenue. Would you go and buy the next release of MS Office if all it promised was to fix the bugs in the current release? Of course not, you would expect (I believe, rightly so) that bug fixes should be provided as a service. At most you might accept a small service charge. In order to generate further revenue Microsoft must provide some perceived enhancement. That only adds more sources of bugs. Without a real competitor in the market there is nothing to persuade Microsoft to reduce prices and bug counts.

At the other end of the scale Microsoft own the most extensively used OSs. It is hard for competitors at this level because the consumer is generally unconcerned with the OS as such, they want their software to run (rather like not caring what video recorder system you have as long as it plays the available recordings). Microsoft are hardly likely to support their competitor's OS by making their applications available on it (the exception is the Apple Mac, but I don't think Microsoft view the Mac as a competitor).

To make things worse new releases of Microsoft applications will use the latest 'enhanced' version of the OS. This makes the task of providing emulators hard work even when that can be done legally.

The final stroke is where many of us have the greatest concern. Microsoft is using its enormous user base to control people's expectations. Visual C++ is not C++. Worse still, commitments to its own legacy code make it very difficult for Microsoft to produce a Visual C++ that has an ISO C++ compilation mode. Even worse, their major library, MFC, will not compile with an ISO C++ conforming implementation. The wide availability of MFC, and the way that much Windows source code depends on it makes Microsoft's competitors in the development tools market pervert their own tools to work with MFC. Microsoft has learnt that controlling the playing field is the way to destroy the competition.

The lack of real competition in the desktop market has resulted in billions of dollars being wasted in lost time and lost work. Microsoft can make as many mistakes as it likes and produce software as buggy as it wishes so long as it ensures that the applications they produce are plug compatible with their operating systems and that development tools must be compatible with MFC.

For our long-term benefit we need real competition at all levels. Until then we will continually spend (waste) ever more money buying the latest upgrades to our software. All the alternatives (Linux, BeOS etc.) require too much change from the general user to be considered serious competitors. I would love to be proved wrong but I am not going to hold my breath.