A gestalt of recent trends, events, and stories should tell you that this prediction is right on the money. A gestalt, by the way, is a perception or conclusion that goes beyond the sum of its parts. Here are three of the many parts that comprise the gestalt.
Part One: Microsoft ran an ad campaign for about six months that used the results of a Meta Group study. The study compares the cost of Windows running on a dual 900MHz Xeon server versus Linux on a z900 IBM mainframe. The Microsoft conclusion, according to the ad, is that Linux is ten times more expensive to run than Windows. The UK Advertising Standards Authority recently upheld complaints that this ad was misleading, as reported in this Inquirer article.
Part Two: Sean Gallagher wrote in a recent Baseline magazine article that Microsoft's developer base is eroding due to powerful and intuitive languages like Python, which can easily replace Visual Basic.
Part Three: Linux Development studies by Evans Data Corporation over the past few years show a definite shift from Windows to Linux among developers.
The prediction that Windows will become a niche operating system and be replaced by Linux is counterintuitive at best. Some of the problems that plague Windows are caused by the fact that Microsoft has designed it to be a one-size-fits-all operating system, so how could it ever be relegated to a niche status?
The answer to this question begins with part one, and requires a bit of history.
Once upon a time, product-oriented publications earned a level of credibility by maintaining a strict wall between the sales staff, whose job it is to sell advertising, and the editorial staff, whose job it is to tell the truth about products. One of the few times an editorial staff crossed the line was when editors felt it was necessary to censor an advertisement that contained potentially false information or was intentionally misleading. Running such an ad could damage the credibility of the publication and, in turn, the credibility of its editorial content.
Now, it doesn't take a Sherlock Holmes to detect that something has changed. While it is reassuring that the the UK Advertising Standards Authority recently agreed that it is misleading to include the cost of a mainframe in order to prove Linux is ten times more expensive than Windows, surely the only people who were fooled by this comparison were those who can measure their IQ with a tire pressure gauge. What is surprising, however, is that (to their shame) reputable IT publications ran this ad for months.
I will leave it to the reader to decide if the editorial content of any of these publications is less credible as a result. But if I may come to their defense just a bit without excusing their poor judgment, allow me explain why this change in editorial sensitivity and control over misleading advertising has come about.
The Golden Years of Microsoft
Ironically, we owe at least some credit to Microsoft that the media once rejected misleading advertisements without fear that the loss of advertising income could harm the publication's bottom line. More than a decade ago, IBM, Microsoft, and others had created a wide-open market for PC applications. If a publication refused to run a misleading ad from Ashton-Tate about its dBase product, so what? The publication would remain filled with ads for other databases like Paradox, Dataease, Clarion, FoxPro, Clipper, R:Base, and countless more. It was a market rich with healthy competition.
Then came the '90s. Microsoft sold these publications a dirty bill of goods with the promise of Windows 95. Microsoft promoted Windows 95 in such a way that publications became positively giddy with excitement over the possible explosion of new advertising revenue. Surely everyone would be eager to sell their new 32-bit versions of Windows applications, hence more advertising income would come rolling in like a tidal wave with the release of this new 32-bit version of Windows. As a result, the IT media (in general) was eager to whip the consuming public into an orgasmic frenzy over the introduction of Windows 95. They believed it was in their best financial interests to do so.
Except it wasn't. Instead of creating a market for a new breed of applications, Microsoft leveraged its control over Windows 95 to eliminate all major competing products in almost every category. By the time the media realized this, it was too late to do anything about it. Almost all products that competed with Microsoft products began to fail in the market, even when the non-Microsoft products were clearly superior. With almost all competition out of the way, Microsoft quickly became the primary source of advertising income for the media.
That's often the reason why even reputable IT publications now run blatantly misleading ads that claim Linux is ten times more expensive than Windows. Many of them can't afford to offend the provider of their greatest source of income.
Don't expect things to change anytime soon. Microsoft has managed to see to it that there is no longer a market for new third-party commercial Windows applications. Some existing applications (Lotus Notes, Oracle, DB2, to cite a few examples) are hanging on in the Windows market, but primarily because it is often painful for the installed base to migrate away from these applications to the Microsoft equivalents. With each new version of Windows, Microsoft is hard at work making it painful not to migrate from existing third-party applications to Microsoft equivalents. For example, one of the design goals of Longhorn is to make third party databases irrelevant.
Developers, developers, developers
That brings us to part two of the gestalt. In the article referenced above, Sean Gallagher points out that Microsoft is losing its developer base. Personally, I think he overstated the influence of Python in his argument. Python is one of my favorite languages, but Python will overthrow the Microsoft hold on developers when someone turns down a date with Angelina Jolie because she has a zit that day.
Python mania aside, Gallagher's basic premise has merit. Gallagher rightly points out that Microsoft built its empire on the backs of developers. But what he fails to make clear is that these were primarily third-party commercial software developers. When Steve Ballmer once chanted "developers, developers, developers" to his employees, people were buying Windows because that was the target for all the popular applications, namely Lotus SmartSuite, Lotus Notes, Borland Paradox, WordPerfect Office, Visio, Corel Draw, and so on. Microsoft has since eliminated, bought, or relegated these applications to niche status.
Here's the problem. Linux and free/open source software is not going to restore the third party Independent Software Vendor market to its former glory. Free/open source software cannot promise to restore this market because the best alternatives to Windows and Windows applications are free-as-in-beer, like Linux, OpenOffice, Apache, PostgreSQL, Ximian Evolution, and so on. It is safe to say that the demand for third-party commercial applications in the Linux market is never going to reach the heights once enjoyed by the Windows market.
The New Market
So here is the current state of the industry. On the one side, we have Microsoft working toward the goal of making sure that everything you run in the front and back office is Microsoft software. On the other side, we have free/open source equivalent solutions to almost all of what Microsoft has to offer. In neither case is there a promise of a revival of the third-party commercial software market. In both cases, there is the promise of making money by adding value to the existing solutions, or by providing vertical commercial applications. (A vertical application is one that targets a special market such as medical systems, etc.)
The new market is therefore comprised primarily of value add resellers (VARs), system integrators (SI), vertical applications development, and internal development of corporate applications. This is the reason why what Gallagher says is true. Microsoft really is losing its developer base to free software. The Evans Data Corporation Linux developer surveys include hard data to back up that assertion, although it currently indicates developers are far more inclined to use languages like Java, PHP, C/C++, and Perl rather than Python.
One of the most interesting trends in the most recent Evans survey is that the reality of what developers are doing is beginning to match their predictions from past surveys. With each survey, developers predicted a major shift in their primary focus from Windows to Linux. Each new survey contained the same predictions, but the shift was not actually occurring. That is, it didn't occur until this latest survey.
The most recent survey shows a significant drop in developers who expect to upgrade to Windows 2003 server. It also shows that a significant percentage of developers are moving directly from Windows 2000 to Linux. In other words, when taken together, several pieces of the survey indicate that developers are finally jumping off the Windows upgrade merry-go-round. This accounts for the fact that they are finally beginning to catch up to their prior predictions of switching focus from Windows to Linux.
That's not to say other factors aren't involved. The most recent survey also revealed a dramatic increase in dissatisfaction with Windows security, as a large percentage of developers moved from ranking Windows as "somewhat less secure" to "much less secure" than Linux.
More important than why Microsoft is losing developers to Linux is which developers are defecting. Microsoft is losing the developer base that belongs to the new market, not the old one. These developers are primarily VARs, systems integrators, commercial developers of vertical applications and the like. The glory days of selling word processors is over. The glory days of selling custom solutions is just beginning to flourish. And Microsoft cannot possibly compete effectively in this new market.
As Microsoft creeps closer to its goal of providing the entire solution from desktop to server, Microsoft runs into the sticky problem of how to perpetuate its revenue streams. The obvious solution, and the one Microsoft has clearly chosen, is the renewable license. Neither developers nor end-user customers can buy Microsoft software, they can only rent it. And when the lease is up, they have no choice but to do without support or renew the contract. Failing to renew is not an option, because they can't afford to leave systems open to the latest worms, viruses, and flaws. No doubt they will also pay substantially more each round, because Microsoft can't afford to keep its revenue stream constant, it must increase its revenue.
Imagine, therefore, that you are a VAR attempting to sell a custom solution to your customer. Which solution will put more money in your pocket? The all-Microsoft solution includes the cost of each copy of Windows at the server and desktop, including user licenses for Windows itself and any licenses for additional Microsoft applications that are part of the final solution. The VAR must pass on this cost to the customer, thus inflating the price of the solution. In addition, these costs are recurring, due to Microsoft's shift to a subscription based revenue model.
The VAR who chooses a free/open source solution can easily underbid any Microsoft-based solution, and still pocket more money when the sale is done.
Microsoft currently has three leverage points against the free/open source solutions. First, Microsoft still offers arguably better tools for the VAR to build the final product. Second, most customers suffer from some degree of Microsoft lock-in, whether it is dependence upon Office document formats or a customer's unwillingness to retrain its users. Third, Microsoft has a Fort Knox of funds with which to advertise propaganda, grease palms, and influence customers in other ways.
On the other hand, as Microsoft tools improve, prices rise. As free/open source tools improve, they remain free. Nor is it cost-effective to remain locked into Microsoft. At some point, many customers will not be able to afford to remain locked-in, and they will have to take the open source plunge. This trend has already begun. Microsoft has been using its cash to stave off some defections to open source, but Microsoft's supply of money is not inexhaustible. With each new defection, another log is added to the dam blocking the river of green flowing into Microsoft's Fort Knox.
As Microsoft loses more customers to open source, it has to yield to pressures to find other sources of income, and one of those sources will be existing customers. That, in turn, places more pressure on existing customers to defect to open source.
In the end, Microsoft has nowhere to go but to become a "niche" player, where its only remaining customers are those who choose to stay locked-in to one or more software products available only from Microsoft. Allison Singh is right. At some point in the not-too-distant future, Windows and Microsoft software is doomed to a niche status while Linux and free/open source software becomes the de-facto standard.