Predicting the future is always risky for an author, especially one who contributes to internet sites, where your words are often instantly accessible to the curious. But I'm going to put my money on the table and take some guesses about the future of Linux. Here, in no particular order, are six theories that I believe are inevitabilities. Keep in mind that although I've been liberal in tone, nearly everything in this piece is speculation or opinion and is subject to debate. Not all of these theories are necessarily entirely original thought, but all arguments are.
1) Major Linux distributions will collapse into a small, powerful group.
"Major players" in the Linux market, until recently, included Red Hat, SuSE, Mandrake, Debian, and Slackware. Some would argue more or less, but now you have a number of popular distros making inroads into the community, Xandros, LindowsOS, and Gentoo to name a few. Another fringe including Yoper, ELX, and TurboLinux are making plays for corporate desktops. I'm coining a new term for this era of Linux computing: distribution bloat. We have hundreds of groups offering us what is essentially minor tweaks and optimizations of a very similar base. This cannot continue at this pace. There will from this point on, be a growing number of Linux installation packages as people become more skilled, but there will be fewer distributions on a mass scale as commercial Linux stabilizes.
I think we'll see the commercial Linux market boil down to two or three players, and this has already begun. I expect it to be a Ximian-ized Novell/SUSE distribution, Red Hat, and some sort of Debian offshoot - whether it's User Linux or not remains to be seen. Sun's Linux offering, Java Desktop System, will be deployed in Solaris committed companies and not much more.
2) Neither KDE nor Gnome will "win;" a third desktop environment will emerge.
The KDE/Gnome debate is a troll's dream come true. People are often passionate about their desktop environment. I believe they both have strengths and weaknesses. However, a third DE, with a clean and usable base, will emerge in time, its sole mission to unify the Linux GUI. Only when there is true consistency of the look and feel of the desktop, or close to it, will Linux become a viable home OS for an average user. Currently, we see this consistency forged by common Qt and GKT themes, and offerings like Ximian Desktop which attempts to mask the different nature of each application. This is not about lack of choice - it is, however, about not allowing choice to supercede usability of the whole product.
Features that a desktop must include are obvious by now: cut & paste must work the same way throughout the OS, menus must be the same in all file manager windows, the same shortcut keys must apply in all applications, and all applications must have the same window borders. Many seemingly basic tasks that haven't entirely matured, or in some cases, been accomplished at all, yet.
In any event, the DE's importance will lessen once greater platform neutrality exists. This will doubtlessly cause many to argue that I am wrong - admittedly, it's a tall order especially with Gnome and KDE becoming established and accomplishing so much. I maintain that unless there is some sort of merging, not a set of standards like freedesktop.org, but rather, a common base for development, that there will be a fragmented feel to Linux that simply doesn't exist in Windows today.