When you think about the state of Linux today, and then compare what you see around you to what you see people writing about Linux, it becomes obvious that there's one of those proverbial elephants in the room - something everybody knows and nobody mentions.
The elephant, of course, is the missing Linux market momentum on the desktop.
Three years ago the Linux desktop was the next big thing and maybe it still is, but meantime it hasn't happened. Lindows, for example, was going to revolutionise consumer focused Linux and I think their product today, now nominally called Linspire (but note: http://www.linspire.com/lindows_products_aisle_landing.php), is more than adequate for the role - but it's not the run away best seller it should be. So why not?
I believe that the Linux phenomenon of the late nineties was fundamentally as much a social movement driven by the hot breath of mass media adoration as a technical one, and that the absence of mass media push over the last few years accounts for most of the missing growth. But not all of it - so the real question is: what's the rest of the story?
I think that "rest of the story" comes in two forms: attrition, and absence of shared direction.
The story on attrition is pretty clear: there are lots of forces, with the fear of lawsuits induced by the IBM/SCO mess being number one and Microsoft's anti-Linux propaganda number two, causing serious commercial developers to think twice about any commitment to Linux - and Sun's openSolaris based CDDL is growing like a wildfire in a high wind in part because it's a safe haven for those looking to hedge their bets.
The biggest and worst factor, however, is the absence of clear direction. In the 90s Linux had both an obvious hero and a committed purpose: as Torvalds put it, he wanted to create: "a free Unix for the 386", "a better Minix than Minix." Laudable goals indeed, but these goals have been achieved and now it's other times, other goals.
And many of those goals are unclear. It's not clear, for example, whether someone considering a new contribution to Linux should now think of this as the ante for joining an important community of like minded individuals united by a common purpose, or simply as uncompensated labour for companies like IBM, Red Hat, and Linspire.
It's not clear either whether desktop Linux is even a challenge worth pursuing or, if it is, who benefits from success. Scientists and developers, after all, need small, fast, cheap compute servers with easily modifiable source - and Linux has that aced. They don't need fancy desktops and technical gimcrackery: basically anything the science community can't do using SuSe 7.1 with KDE they can't do on Linux at all.
Look at what's nominally holding back the desktop and it's mostly stuff, like music players and games drivers, for kids or other technical illiterates who don't contribute to Linux. So why should Joe, physics post-grad cum accidental sysadmin by day and brilliant open source programmer by night, want to work on it? If he wants that kind of thing for his use, he can buy a Mac - and get a first class Unix with the wheels already invented, debugged, and ready to roll.
So bottom line? the community doesn't know why it should do desktop Linux - IBM knows, Red Hat knows, Linspire knows, but they're not the Linux community and that community is suffering from a failure of leadership. The early, inspiring, goals have been met; the early leaders have moved on with their lives; and no new goals have been set, no new leaders coronated, no new myths created.