[By Matthew French] For people who don’t use Windows, the day that Microsoft’s grip on the desktop is broken can’t come soon enough. The reason is simple: the technical divide created by the dominance of one operating system means that people who choose to use something different face constant discrimination.

For the past decade, Microsoft’s position has been unassailable. Numerous predictions of the imminent demise of Windows have proven to be wishful thinking. However, isolated pockets of resistance have started to appear in the desktop world. For the first time in years, Microsoft’s market share is being whittled away by amounts that can no longer be explained as statistical noise.

You are not alone if you think Microsoft has little to be concerned about. However, if you dig into the numbers, home users are defecting. Unlike people who use the computers provided by their companies, home users have a choice. Businesses tend to lag adoption by home users, so it is only a matter of time before large organisations start to look at their options. The more adventurous ones are already doing so.

Now, most companies will want a business case before they change something that works. Usually the first step is to do a financial cost-benefit analysis. In IT this inevitably means a study of the total cost of ownership (TCO). If you have been involved with TCO calculations you probably know that it is easy to make free look expensive and expensive look cheap just by changing a few assumptions.

Nowhere is this more evident than in the choice of desktop operating system. Linux might cost nothing to buy but if you listen to Microsoft it can cost a lot more than Windows to keep it running. Ironically, advocates for Apple use exactly the same argument against Windows. They will say it costs more up-front to buy an Apple but a dramatic drop in support costs makes up for this.

So, looking at the financial costs alone is not very helpful.

The next option is to do a blow-by-blow comparison of the benefits and risks of each operating system. Unfortunately, this path can lead to madness. To understand why, let’s look at one of the common debates.

The relative user friendliness of operating systems is usually one of the first points picked up when this debate starts. The sad truth is that apart from basic calculators — and arguably the very first Macintosh — computers are not user friendly. The mistake is to confuse familiarity with ease of use. This doesn’t stop user friendliness, or many other highly subjective measures, being used to compare operating systems.

So if we can’t use features or costs, how does one rationally choose the best desktop operating system for a business? The answer is, it’s the wrong question.

To make sense of this cryptic response, we need to look back at the marketing disaster that was Windows Vista. When Microsoft first announced it was ending support for Windows XP, many commentators viewed it as a cynical attempt to force businesses to move to Vista in order to improve Microsoft’s sales figures. What it also achieved was to drive home the point that businesses have a critical dependence on Windows. It doesn’t matter if the vendor is Microsoft, Eskom or Telkom — if there is one thing a business doesn’t like, it is being dependent on one supplier.

Unfortunately for businesses, there is a problem. One of the key reasons why Microsoft was able to achieve and maintain its dominance over the desktop is because of operating system economics. Operating systems, especially for networked desktops, are a natural monopoly. What this means is that once a critical level of market penetration has been reached, it is far easier for people to stick with the status quo. One of the key reasons people use Windows is, well, because everyone else uses Windows.

To see how far the effects of the natural monopoly go, we can look at one common argument against Linux: lack of device drivers, especially for printers, scanners and wireless network cards. The drivers don’t exist because not enough people use Linux for the vendor to justify writing Linux drivers, but people don’t use Linux because there are no drivers. Microsoft doesn’t have to lift a finger to hobble Linux, while the Linux community has to work extra hard to support other people’s products — and at the same time run the risk of legal challenges from hardware manufacturers who don’t like people reverse engineering their products.

The natural monopoly also affects applications, especially those applications aimed at specific industries such as hotel management or medical practices. The software vendor usually doesn’t have the resources to write applications for every platform, so they target the majority operating system and end up reinforcing the lock-in.

There is even a monopoly on experience. People hate change and for many a computer is something with a green Start button in the lower left corner of the screen.

There are a number of reasons monopolies are bad. In the case of operating systems, a monopoly not only removes the incentive to innovate, but it makes it harder to fix known issues because so many applications depend on the workarounds. Virus writers also use the same logic as application vendors and target the majority platform. And, of course, the monopoly means that if Microsoft does say jump, it has millions of customers who have no choice but to ask how high.

The obvious solution is to replace some of the Windows desktops with alternative operating systems. But like so many obvious solutions this is not so easy to do, nor is it practical in most cases. The thought of having two extra operating systems to support would be enough to get many helpdesk managers to change careers.

Since the problem has its root in economics, maybe the solution lies there, too. Instead of forcing people to change their desktop operating system, just remove the disincentives that prevent them from changing platforms by themselves.

But before looking at disincentives, the first step is to ensure the IT department is behind the idea. The beauty of the economic approach is that IT departments don’t actively need to support alternative platforms; they just need to tolerate them. However, the IT department still has some work to do. It has to ensure that licensing and security policies are adhered to. It also needs to be sensitised to the issue so that when negotiating on behalf of the enterprise it doesn’t inadvertently increase the single operating system dependency.

Once the IT department accepts that this is an issue it has to deal with, the next step is to identify the disincentives. The most common disincentives are usually business applications that only run on Windows.

In small businesses, one typical culprit is the accounting system. Bigger organisations will often have an eclectic mix of custom and specialised software. Some software will be written for Windows only. Other applications will be Web-based, though surprisingly often these will need Internet Explorer 6 or use components that only work on Windows clients. There are even quite a few Java applications that will only run on Windows.

To work around these applications, users of alternative operating systems can use a remote desktop connected to a Windows server. For some applications it is also possible to use Wine, which fakes Windows on Linux and Mac OS X with varying levels of success. However, since the idea is to break the operating system dependency this should be seen as a short-term solution.

The longer-term solution is to persuade vendors to support other platforms. This might not seem practical for applications that have been written to run only on a Windows desktop, but already some of the more innovative vendors have found solutions.

For example, some vendors actively support their applications running under Wine. While this is still a short-term solution, it is important because it means that they are willing to consider alternatives even if they don’t want to invest in them just yet.

For Web-based or client-server applications that do most of their processing on the server side, it should not be a big issue to support more than one desktop platform. Unfortunately, many vendors see supporting a single deployment environment as a shortcut to reducing development costs. Customers need to be wary of these vendors because this thinking leads to those applications that in 10 years time will still only run using IE6 on Windows 2000, in the same way that today there are still applications out there that only run on Netware 3.11 or SCO Unix. These legacy systems introduce so many problems that we cannot even begin to cover them here.

And, unfortunately, that is the biggest issue with arguing that breaking the operating system monopoly is important. It requires decision makers to think about problems they will come across in five or 10 years. Unless one is staring at a million-dollar, mission-critical business software system that has to be replaced immediately because it is no longer possible to get hardware to support it, it can be difficult to appreciate how deeply an organisation can be tied to its operating system choices.
Before they get too smug, proponents of alternative operating systems should be mindful that this argument cuts both ways.

There are serious issues with tying a business into Windows, but it is unlikely an Apple monopoly would be any better. And without something for the Linux community to focus against, Linux could easily end up catering for the lowest common denominator. Replacing one monopoly with another is not going to solve anything.

In the meantime, the good news for Microsoft partners is that Windows is not going away, not for a very long time. But now it has some real competition, and that should benefit everyone.