Linux can't succeed as a generic network computing client, only. People will continue to use their PC as a power workstation, even when it isn't appropriate. It's the nature of computer users to do so. For this reason, Linux needs a compelling desktop experience. It already has Compiz Fusion, but even though 3-D on Linux doesn't require nearly the hardware resources as Vista, many Linux users still refuse to install Compiz or turn it off.
The desktop needs a more substantial advance in thinking. The new KDE, KDE4, looks promising in this regard. The KDE developers seem intent upon bringing something new to the desktop experience that isn't just eye-candy. KDE4, or parts of it, will run on Windows and Mac OS-X, but it will be fully native on Linux, and should benefit Linux more than any other platform.
KDE4, the proliferation of Linux on appliances, the trend toward generic network computing, the fact that Linux is free (both as in freedom and as in "free beer"), and other factors contribute to the inevitable success of Linux on the desktop. But Linux still needs more. It needs windows of opportunity to supplant the legacy systems, and it needs to overcome some important obstacles.
Linux's 'Window' Of Opportunity
Both the successes and failures of Microsoft provide a substantial window of opportunity for Linux to seize a significant desktop market share. It is painful, especially at the enterprise level, to switch desktop operating systems, so any legacy system like Windows will always have a huge advantage. But Microsoft has made so many blunders in recent times that one must credit Microsoft itself for encouraging users to seek an alternative desktop operating system. Windows was already a notoriously insecure operating system, but Microsoft has compounded the problem with the expensive, buggy, incomplete, complex license-burdened, DRM-encumbered, hardware-challenged, frequently updated without your permission Vista.
As noted earlier, people are most likely to switch to a new operating system when dollars are on the line. Microsoft would be wise to continue supporting the "good enough" Windows XP, since any move to force people to upgrade to Vista could create the "dollars on the line" scenario. The risk of adjusting to a new operating system becomes much more palatable when it saves you the cost of upgrading to a desktop you know you won't like.
Perhaps the most significant Microsoft failure was its flubbed attempt to use SCO as a proxy to create fear, uncertainty and doubt about Linux. Those who backed SCO are now eating crow. This makes it far less likely for high-profile analysts to make the same mistake, now that Microsoft is attacking Linux directly by claiming Linux violates its software patents.
Microsoft began a catch-22 strategy when it released Windows 95. On the one hand, it successfully leveraged its unique advantage in building 32-bit Windows applications to eliminate virtually all competition in mainstream desktop applications. The catch is that Microsoft has left itself without friends. For example, if Lotus Smartsuite and WordPerfect Office were still thriving competition for Microsoft Office, it would be all but impossible for Linux to break into the desktop market. Companies would be content to collect their Windows applications revenue. There would be no incentive to support another desktop platform.