Information Blog on Everything

Friday, January 05, 2007

Top four reasons Windows wins and Linux loses

Opinion -- Today, you can do everything you want with a Linux desktop, except play the latest games. Even there, Linux is catching up. So, why do only a handful of people run Linux instead of Windows? Here are my top-four reasons why Windows wins and Linux loses.

Before I start, though, let me say -- because people always assume I'm anti-Linux when I write pieces like this -- that I use Linux desktops every day. I'm writing this on a SLED (SUSE Linux Enterprise Desktop) 10 system, I run MEPIS 6.01 and Xandros Professional Desktop 4.1 on my laptops, and on my other desktops that I use at least weekly, you'll find Freespire 1.0, Fedora Core 6, and openSUSE 10.2. In short, I use Linux. I love Linux. But, that doesn't mean I'm blind to business reality, Windows virtues, or Linux flaws.

So, without further adieu, the number-one reason why Linux trails in the desktop races...

#1: The installed base

There are, what, hundreds of millions of Windows XP and 2000 systems still out there and working? That's a lot of systems. That's a lot of people who know nothing but Windows.

Canonical CEO Mark Shuttleworth claims that there are at least 8 million Ubuntu Linux desktops alone out there. I wish I could believe that number, but I don't.

I could believe that there are 8-million total Linux desktops out there. If we accept that there are 8 million Linux desktops out there, based on IDC market-share Linux marketshare estimates that would mean we're talking over a billion Windows desktops out there. Ouch.

The installed base, however, may turn out to be a blessing in disguise for the Linux desktop. That's because the analysts think that Vista is going to have real trouble pulling users away from older versions of Windows.

Al Gillen, vice president of research at IDC, recently told eWEEK that one of the biggest threats facing Microsoft in 2007 is its own installed base. Even Rob Enderle, principal analyst at The Enderle Group and a consultant to Microsoft, said he doesn't see much demand for Vista both because of its abysmally late launch and users who will stick with their legacy systems.

In addition, Gillen thinks that Microsoft's focus on reducing piracy with its invasive authentication systems, "may accidentally accelerate the option of Linux as a client operating system. Microsoft's client operating system anti-piracy efforts may well backfire and that very anti-piracy campaign could drive customers toward Linux."

So, maybe this will be the year that the Linux desktop market doubles, or even triples, in size. That will mean great business for Novell, Red Hat, and Ubuntu, but that will still leave Linux hundreds of millions of users running Windows.

#2: PC vendor support

If you want to buy a Windows system, go anywhere and you can pick up one. If you want to buy a Mac, you'll need to do a little looking, but your local yellow pages should point you to a dealer in your area without much fuss. If you want to buy a Linux desktop... well, prepare for a long hunt.

Yes, they're out there. Linspire, in particular, does a good job of partnering with smaller PC manufacturers and distributors. For example, Koolbox's Mini koolbox line of Mac Mini-style PCs are fine low-end computers for a decent price.

But, you do have to actively look for a Linux-powered PC. It also doesn't help any that even the big-time vendors that offer Linux desktops, like Dell and Lenovo, make it darn hard to buy them.

This isn't going to change anytime soon. I've been talking to a lot of vendors lately, and it's really very simple why we're not going to see many more pre-loaded Linux desktop PCs anytime soon: there's almost no demand for them.

If you want to see more Linux PCs, you're going to need to ask for them; again, and again, and again, because the big vendors aren't hearing a peep. What demand there is for Linux PCs is coming not from consumers but from enterprise customers. So while I think we may see an HP or Dell come out with a low-priced business desktop line this year, you can pretty much give up on the fantasy that CompUSA will have half-a-dozen Linux-powered PCs in its aisles come Christmas 2007.

#3: Hardware vendor support

One of the things that everyone complains about in Linux is that it doesn't have enough hardware equipment support -- WiFi cards, iPods, high-end graphic cards, scanners, whatever. You know what? They're right.

It's not Linux's fault, but, repeat after me: users don't care. All they know is that they can't connect to their WiFi access point, or that their all-in-one scanner/printer/fax machine can only print.

Yes, with Linux, 99 percent of all hardware works with 95 percent of its functionality. Again, users don't care. All, they know is that their WiFi card doesn't work, therefore Linux is trash.

This is not, however, a problem just for Linux. Windows users, who have become accustomed to the idea that everything always works with their systems, are in for a rude awakening when they start upgrading to Vista. Then, they're going to find more hardware trouble than Linux users have had in years.

But, just because Vista users are going to be in the same boat, won't help the Linux desktop much. Linux companies have to do whatever it takes to work with proprietary hardware. In this regard, Linspire, with its wiliness to include proprietary hardware drivers, has taken a leading position. Other distributions, like Ubuntu, are still fighting over these issues.

From where I sit, it's really pretty simple. You can be ideologically pure and only use open-source software and have distributions that won't work well for many people, or you can include some proprietary drivers and firmware and produce distributions that will work better for most users.

Another idea that could help, which was kicked around at the last Portland desktop meeting, is to set up a program through which vendors could get their hardware certified to work with Linux. Think "Works with Linux," instead of "Works with Windows," as a branding campaign, and you have the idea.

I, for one, would certainly appreciate being able to look at an ad, or at the packaging, and know at a glance whether the goodies inside will work with Linux. This kind of hardware certification sounds a lot easier than it is to actually do, but I think it would go a long way toward making Linux more popular with casual users.

There's also a related problem, but here Linux could gain a permanent advantage over Windows. As open source leader Eric S. Raymond said at last August's LinuxWorld in San Francisco, as PCs make the jump from 32-bits to 64-bits Linux has a chance to become the number-one operating system.

To do that, however, Linux needs to have a lot more 64-bit drivers, and applications that work in 64-bits. One of the biggest problems is that most Linux distributions and the LSB (Linux Standard Base) maintain separate library repositories for 32-bit and 64-bit applications.

What that means, in practice, is that you can't run 32-bit and 64-bit programs together. For instance, if you use 64-bit Firefox, you can't use 32-bit Macromedia Flash or Adobe Acrobat. Can you say annoying?

In this area, Windows is actually in worse shape than Linux. Running 64-bit Windows is much more of a pain than Linux. Now, if Linux can move forward in the 64-bit agenda, we could have an operating system that -- even to the most naive eye -- performs better than Windows.

#4: Software support

Yes, I can run anything I want on Linux today, but then I'm an expert. Most users will do well with Firefox for browsing, GAIM for IM, OpenOffice for office work, and Thunderbird or Evolution for email. But, once you move beyond the basics, though, it gets more complicated.

Part of the problem is that there's no single easy way to install software on Linux. On Windows, you click on the installer, and, wham, bang, you're in business. Vista is going to change that for the worse, but that's not our problem.

Our problem is that we have half-a-dozen very different "easy" ways to install programs, like apt, YaST, and yum. We also have some software that will only install if you know exactly what you're doing with rpmbuild, make, and directory permissions.

Now, the LSB and friends are working on solving the installation problem. Better still, their approach of creating a common, high-level API (application programming interface) sounds very workable. With some work, by this time next year, installing applications may be just as mindless for most Linux users as installing programs on Windows currently is for XP users.

Another sore point is that Linux is still struggling to find common ground for desktop developers. Thanks to the Portland Project, Linux is now well on the way toward making it possible for ISVs (independent software vendors) to build an application one time for any mainstream distribution without needing to worry about whether the desktop environment is KDE or GNOME. For users, this means that they can just get an application simply run it on their distribution, without fussing or fiddling.

Linux is also catching up with Windows software development because it finally has an answer to the outstanding MSDN (Microsoft Software Developers Network): the LSB Developer Network. With these resources, programmers will be able to write Linux software almost as easily as their Windows developer friends.

This is all good news... for software developers. For end-users, having a wide variety of software choices that are easy to use on any distribution is still a ways off. At least, however, Linux is finally on its way toward applications for ordinary, rather than only expert, users.

Giving Windows a run for its money

If Linux can improve in all these areas, and Vista stumbles, as I expect it will, we may finally see Linux giving Windows a run for its money in the marketplace. I certainly hope that will be the case. I already know Linux makes a great desktop operating system -- I'd like the rest of the world to be able to join me on it.

-- Steven J. Vaughan-Nichols