The impending release of Windows Vista with its fancy Aero Glass special effects, along with the hasty addition of the similar XGL and Compiz technologies to the latest SUSE Linux release makes me think that programmers have a warped idea of what desktop computing is about. For some reason, many GNU/Linux users are concerned about competing feature-for-feature with Vista, while Apple and Microsoft struggle to add more graphical extras to their already graphics-intensive desktop OSes. It's gotten so that you need a serious 3D video card (with proprietary drivers) and a fairly fast computer just to keep up with desktop environments. Whatever happened to being productive and having fun?

We don't need no stinking graphics

Stop and ask yourself this question: how do the Aero Glass, XGL/Compiz, and the candy-like bouncy/trouncy OS X interfaces make you more productive, or make games more fun? Forget how "beautiful" or "sexy" your computer or software is for a moment -- a computer is neither a person nor a work of art, after all -- and think about what you're really using your computer for. What does a host of 3D effects add to those uses?

My answer is: nothing -- in fact, the 3D effects of XGL/Compiz ruin my work environment because they're so distracting, and they reduce my frame rate in 3D games. Aside from that, XGL has been known to cause system instability, and can be difficult to properly install and configure. So why on earth would I want this?

Graphical effects only seem to be useful when it comes to evangelism, when you need to show your friends all of the fancy tricks your computer can do. It's the veritable equivalent of stupid dog tricks -- they don't improve the animal's disposition, and they don't make them more fun to play with, but boy are they good for impressing your friends.

It's about the programs

What the desktop GNU/Linux (and *BSD) realm truly needs to increase its usefulness and viability is not stupid tricks, but specialized and competent programs that make the operating environment an indispensable tool. Right now GNU/Linux is where the coolest software is, and as a server OS it's the first choice for many sysadmins, but it's not really indispensable. If GNU/Linux disappeared tomorrow, most of the people who rely on it could switch to FreeBSD, OpenBSD, NetBSD, or Solaris without a great deal of trouble. For the most part, all of these operating systems share the same benefits as GNU/Linux -- good security, a large collection of ported applications, system administration tools built in, and access to the source code. You could probably even keep most of your scripts and other custom GNU/Linux hacks, or at least only have to modify them slightly.

OpenBSD and NetBSD have well-defined and focused niches: security and portability, respectively; no other operating systems fit these niches as comfortably. FreeBSD used to have a reputation as a Web server, but it has lost that focus and is now sort of the "Linux" of the BSD world -- and that says just as much about GNU/Linux as it does about FreeBSD. They both do so much that they don't really do anything at all. Specific GNU/Linux distributions do introduce a small amount of focus, but aside from a handful of desktop-oriented GNU/Linux distributions, I haven't found any that really hone in on a specific use. For instance, I'd like to see a distro that is totally dedicated to serving Web pages, with the option to add FTP, rsync, and other Internet services. The kernel, the command line tools, and the graphical interface (if there is one -- and there should be) would all be totally dedicated to configuring, delivering, and monitoring these Web services. Right now all that we have are gigantic, wide-ranging distros that we pare down and customize for our uses.

We also have gigantic, wide-ranging programs like OpenOffice.org that we fit to smaller uses. OpenOffice.org is the king of bloat and copycatism. Was it designed as a suite of tools that people commonly need in an office setting? No, it was designed to copy what was in Microsoft Office, whether or not it makes any sense to do so. If it were truly an "office suite," it would have a program for making charts and other business graphics, a program for writing letters on predefined letterhead, an IM client, and an email/PIM program. It has none of those specialized tools.

If you remember the first few generations of Apple Macintosh computers, you probably knew them as desktop publishing machines. That's what Macs were originally known for -- superb desktop publishing, and at the time there were few or no viable alternatives to the Macintosh, especially for a comparable price. These days Macs are just fancy PCs, and although many audio and video recording engineers and artists prefer Apple hardware and software, Apple by no means has those markets cornered. Like everything else in the desktop computing world, the Macintosh has lost its focus.

It doesn't have to be this way. Like the Macintosh in the mid-80s, GNU/Linux could just as easily establish itself as a must-have platform, but not just for desktop publishing -- for a variety of niche groups:

• Software developers. GNU/Linux is already pretty strong in this arena because of the free software movement, but there are still many ways that it could be made more developer-friendly. For instance, there could be a desktop environment (or an entire GNU/Linux distribution) specialized for software development. As it is now, you have to assemble and customize your own development tools and then arrange the graphical interface (if you use one) so that you can more easily access them. The good news is, there are many excellent IDEs for a variety of different programming languages; excellent compilers and debuggers; and decent virtual machine frameworks. Sun Microsystems' Java Desktop System, which is based on a prehistoric version of SUSE Desktop, does come close to being an ideal software development platform -- especially for Java -- but it still has a lot of unnecessary extras.
• Professional writers. Obviously this one is high on my personal wish list, but I'm not alone -- writers are a large, neglected group of computer users. I regularly get email from other authors and journalists who repeat the same lament: there aren't any really good programs or specialized environments for writers. We're stuck with over-featured software like OpenOffice.org, or barebones tools like Vim and Aspell, and we have to modify our work habits to fit the boundaries of these programs. What we should have is a suite of literary-specific tools -- one for drafting, outlining, and storyboarding; one for editing (spelling, grammar, punctuation); and one for publishing (page layout and design, etc.). Don't forget screenwriters, either -- they should have tools specific for their needs. Programs like JEdit and LyX can come close in many regards, but we're still without heavy-hitting English drafting and editing tools on GNU/Linux.
• Graphic designers. The GIMP is good for computer graphics, but substandard for print graphics -- or at least, that's what the graphic designers say every time someone publishes a GIMP review. I'm not sure that appropriately advanced graphic design tools are necessarily absent in GNU/Linux, but they would be much easier to find and use if they were specialized to certain tasks. This is more or less the same gripe as the one above about professional writers -- what's needed here is a suite or collection of programs that each superbly perform a small set of focused tasks. And, like with programmers, a specific desktop environment or distribution would also help.

Many of these things could be accomplished simply by properly customizing GNU/Linux distributions so that they're already pared down and focused. But who wants to do that? I'll admit that the occasional late night spent watching Sopranos DVDs while casually setting up OpenBSD or Gentoo on a notebook computer is fun and interesting, but only because I have the technical know-how to do it. Most of the writers I know are tech-ignorant, and are stuck with computer setups that are overpriced and inappropriate for their work.

Going further with this idea, a particularly creative company could design hardware to go with a specialized GNU/Linux distro, thereby creating niche-specific computing appliances. I think the majority of computer users really want more of an appliance anyway -- something they don't have to mess with, fuss over, or spend a lot of time maintaining.

Even though pundits and programmers alike seem to disagree with me, I think the future of GNULinux desktop computing does not lie in making existing programs more feature-packed and graphically entertaining. The future is where it has always been -- in achieving a higher degree of usefulness to the people who use computers for more than just programming. Recently, software developers have become shortsighted and distracted by "feature envy" in the months leading up to the Windows Vista release. I can't wait until that's over; maybe then I can look forward to some truly innovative software, rather than workalikes and lookalikes of programs I left the Windows world to escape.

Discuss this article or get technical support on our forum.

Copyright 2006 Jem Matzan.