Allow me to go out on a limb. I'm not claiming to know what the next big thing
in Linux will be. I'm thinking of what will arrive by, say, 2006: Operating Systems.
OK, I've stated the obvious, right? No, not really.
I either smuggly smirk or bury my head in my hands when Linux Evangelists state that Linux is an OS. It's a kernel. FreeBSD is an OS. Debian is an OS. Gentoo is an OS. It happens that Debian and Gentoo run the same kernel, and a different kernel than FreeBSD.
In other words, the emphasis is going to shift away from what Linus, et. al., are doing with Linux to what others are making from Linux.
Why? The Linux kernel is a groovy, funky piece of technology, and it's the heart of a movement. But hearts don't live outside of rib cages. Kernels don't run without OS's. Companies don't migrate high-end, mission critical servers to OS's that barely run the super-fast kernel beating at its center. They want -- scratch that, they need a full OS that does the job. Whether the kernel is trendy or not doesn't matter in the end.
FreeBSD has shown that a free, stable, solid Unix-like OS system is possible. If not for its license (sorry, BSD license lovers), it might have stood a chance at the top spot in the Free OS world. Debian and Gentoo have shown the first real movement toward something like a complete OS on the Linux side, especially Debian. Deb was first, and it's still around, but it's stodgy to the point of ridiculousness (from the POV of a power user). Thank God for Gentoo.
Sure, Gentoo may not be ready for mission critical servers simply because it offers you the latest, untested code. But power users get their candy and their popped-up engine. And how sweet it is.
For anything that must stay up, that's when Debian wins points for its stodginess. And here's the kicker: you get to choose your kernel.
This is the development that turned on the little light-bulb that floats above my head. This is the future of Linux.
Think about it: Debian runs on the Linux kernel, the Hurd kernel (no chuckling, please), and the NetBSD kernel. So, which OS runs on the most hardware in the world?
Debian! (10 points.) What does this mean? That we're moving away from a kernel-centric universe. It's not which kernel to choose from, it's which OS. A savvy sysadmin can just install Debian everywhere, choosing the kernel that fits the situation. The key phrase won't be: "I must run Linux." It will be: "I must run Debian." Choosing the kernel will secondary to getting the right OS. I doubt it will be long before Debian is joined in this effort by Gentoo or a similar project
So, how does an OS-centric universe differ from a kernel-centric? For one, Richard Stallman might get the recognition he feels has been wrongly given to Linus. For another, "GNU" will be just as important a word as "Linux", which again will make RMS a much happier camper. On a technical level, the emphasis will shift from the sophomoric question of "Do you run Linux?" to "Which OS do you run?" Debian with a 2.2 Linux kernel. Debian with NetBSD. Gentoo with a development kernel. FreeBSD, modified with OpenBSD security, running a NetBSD kernel. Whatever. Hackerdom may offer near unlimited possibilities.
The point is, the whole OS will finally be greater than the sum of its parts. Watch for the Linux kernel to lose prominence (slightly) as OS's that offer specific features (stability, the latest-and-greatest, etc.) begin to move to the forefront of user consciousness. Watch for a port of Gentoo to include a non-Linux kernel; watch for Debian to support a fourth kernel; watch for a commerical product that produces custom OS's based on Free and Open Source software that emphasizes the Linux kernel without excluding other options.
Yes, Linux Evangelists will kick and scream, but for the wrong reasons. If this scenario comes to pass, the world will be filled a much better breed operating systems than we have now.
[Here's one idea: strip down a Linux kernel and create an OS to replace Minix. How much better would a book like Tanenbaum's be with a real OS to study?]