The future of Linux
(IDG) -- I'm going to take a break from my usual format this month, due to reader response to my comments on the future of Linux.
Several readers thoughtfully disagreed with my prediction that over the next few years the number of Linux variants will shrink to one or two major distributions, and that these changes will cost Linux some of its characteristic vitality and flexibility. My disagreeing compatriots predict that, if anything, Linux distributions will grow and diversify as Linux becomes more popular.
They may be right. Historically, though, that's not the way markets have developed.
For example, consider what has happened with language compilers, especially for popular languages like C and Pascal. Here, the language specification plays roughly the role of the GNU license in keeping the underlying system the same. (Note the word roughly. The cases are cognate, but not identical.)
The various ISVs compete by adding value around an underlying model. Some ISVs are better at it than others. You quickly get a situation where the compilers you buy are more powerful, flexible, and useful than the bare language, and the competition over who has the best add-on features eventually produces one or two winners. The process takes several years, and at the end you'll still have some marginal players -- if you can find them -- but the "serious" work is done by the market leaders.
If you accept my analogy, some of today's Linux distributions may be the equivalent of Alice Pascal, PowerC, or Lattice C.
Now look at Linux as it stands today. The core OS is good, but for the person who just wants to get the job done, there are problems. Installation can be intimidating to the newbie, hardware incompatibilities aren't handled gracefully, the GUI is a problem to set up, and so on. Often, too much is made of these difficulties -- but the fact remains that they are barriers for beginners.
Hurdling the barriers
The obvious way to gain advantage in the Linux market is to deal with those barriers. If I were doing a commercial distribution, I'd have teams working like madmen (and women) on designing a really bulletproof installation system -- complete with error messages that mean something to the uninitiated, suggested fixes, specific advice on handling hardware incompatibilities, and a help file that's even better than the how-tos (which are very strong, by the way).
The goal would be something the user could take out of the box, pop into the system, and see up and running in a matter of minutes. If I have that, I have the beginnings of market leadership.
Don't forget applications
The next part would be to make it easy for vendors to hook their applications into my installation system. That would encourage them to write applications for my version of Linux first.
Of course, under the GNU license, the source code for any add-ons has to be publicly available. But having the source is only half the battle. It's how well you integrate the features source code provides into the rest of your distribution.
Fundamentally, I'm betting that I not only have a superior feature set in my "frosting," but that I can execute better than anyone who tries to copy me. (In the case of offering help, there are ways around the GNU license as well, such as separating the code from the text and copyrighting the text. Not recommended, but it would probably work.)
This is only one of five or six possible value-adds to Linux. In order to make them work, I don't have to violate the GNU license. All I have to do is make the frosting on my cake more attractive -- and sufficiently different -- so people will choose me and it won't be as easy for those who choose someone else.
Maybe I won't be the one who succeeds. But someone will. This strategy is so obvious we can expect a lot of people to try it. It's in the nature of competitive enterprise that most of them will fail, but one or two of them will succeed.
Next, market momentum takes over. It takes work to gain the advantage, but once you have it, it snowballs.
Marketshare alone becomes an advantage, and it's a self-reinforcing advantage. The natural result is that anywhere from one to three companies quickly build a nearly insurmountable lead over the multitudes of competitors.
Because of Linux's history and culture, I don't see the emergence of a Linux-based incarnation of Microsoft, with absolutely commanding dominance and the ability to sway the entire course of the market by fiat. For one thing, the terms of the GNU license works against that. For another thing, so does the nature of the product. Finally, competitors and users have learned from experience.
However, I still think the process will progress far enough that we will see a severe winnowing out to two or perhaps three major Linux players in the next four or five years.
I know this kind of concentration is antithetical to the spirit of Linux, which glorifies diversity, but to the extent that Linux becomes widely popular, that spirit doesn't matter. From the very first microcomputers on, the spirit of these products have been changed (or corrupted) by success.
The other side of this equation is that for most users, this is a good thing.
The vast majority of users are, by definition, not experts. They want something that will work. Science fiction editor John W. Campbell used to say that what most people want is magic, which he defined as "product without process." Campbell was right. Most people aren't interested in mucking with the process, they want the results. And the trade-off for becoming a major force in software is that you must give those people what they want.
So, if Linux is to become really popular -- and it's well on its way -- it's going to have to acquire a lot of additional features. And those features must be contained in coordinated, well-designed, and thoughtful packages. Those will almost certainly be the province of distributors, and the distributions that do the best job of providing what all those new users want will be the ones to flourish and ultimately dominate.
This isn't necessarily a bad thing. Most programmers prefer the latest versions of C++ and Pascal, with their integrated development environments, built-in code profilers, and all the other wonderful stuff that's been added on top of the basic language. In fact, one of the things they complain bitterly about with Java is that it doesn't have the same level of tool support as do modern C and C++.
In the same fashion, I think most Linux users are looking forward to really painless installations (and no, a Windows installation is not painless in my book), superior GUI, and all the other nifty stuff that will come out of this revolution.
But we have to realize that we will inevitably lose something in the process. What we will lose as Linux settles down is diversity and vitality. We will lose distributions, as they'll be either completely eliminated or pushed into marginal existence by the roaring success of one or two players.
That's sad, especially for those of us who enjoy the cheerful anarchy of Linux as it has been. But I think that's the way it will be.
Rick Cook has been covering computers and high technology for nearly 20 years for various publications. He is the author of a series of fantasy novels full of bad computer jokes.
Back to the top
© 2000 Cable News Network. All Rights Reserved.
Terms under which this service is provided to you.
Read our privacy guidelines.