Ubuntu Linux has become very popular in the open source community in only two years, last confirmed by the 2005 LinuxQuestions.org Members Choice Awards. But being number one in a community of computer literates, nerds, and geeks is not enough to gain a significant market share, as shown by their #1 bug, entitled "Microsoft has a majority market share".
So here's a couple of cents worth of musings on why open source (and especially Linux) is not much used, what can be done about it, and where it's headed.
Ignorance, or being stuck with what you've got
AKA, "users don't know that there are alternatives." This is a popular reason, but I believe it is a bit off the mark: Users don't know that there exists alternatives which are both free and good. Most people believe there is no such thing as a free lunch, and free (as in speech and beer) sounds dubious to people who have learned not to trust anything on the Internet.
The "family nerd", the poor soul which has to fix everybody's computer problems, has become the forerunner in this mission, because of the empirical evidence of security, stability and usability of some brilliant new software. Already it seems Firefox is spreading around like a wildfire, helped by a host of extensions such as AdBlock and SessionSaver, the latest of which will be part of Firefox 2.0.
The trench wars on the legality of file sharing has also sparked a lot of open source development, such as RevConnect (Direct Connect client) and Azureus (BitTorrent client). These can easily replace the old way of having to jump through elaborate web page hoops to download free software and media files, from web pages you don't know whether to trust or not. With P2P software, trust is like viral marketing: If something is popular (i.e., many people are sharing / downloading the same files), it's probably good stuff. That's also part of the beauty of open source: Since the program and source code is scrutinized by many, any malicious behavior by the developers is likely to be detected, discussed, and resolved, either by a forking of the code or by massive abandon by the users. Closed vs. open source security is a much too big discussion to take on here, suffice to say that both camps have released software with horrible bugs.
Laziness, or work space pragmatics
AKA, "users don't want to learn something new." Also not completely true: Users are, for the most part, primarily interested in getting the work done. If Microsoft Office was as broken as Internet Explorer, users would be downloading OpenOffice.org in droves, but as it turns out, current (and older) word processing and email programs are more than adequate for the common user. Hell, I could probably be using WordPerfect at work without anybody noticing.
When your primary goal is to get a bunch of bulleted lists in a presentable format in two hours, it is hardly pragmatic to start thinking about whether you'll be able to view that same presentation in five years, or convert it into a PDF file in a flash. And from what I've heard and read, a lot of people seem to think that keeping to the same level as everybody else is the optimum. Hacking the system to make it conform to your work style will invariably break it sometimes, and that is likely to be more visible than the x% overall productivity gain.
Openness not usually considered, don't want to "fiddle" with system settings; developers don't want to put in the extra effort to make software cross-platform
Stupidity, or not thinking (far) ahead
Except among nerds and geeks, the freedom of open source software seems to be ignored. It's difficult to find good analogies, but that won't stop me from trying: Food. "Closed source food" would be sold without ingredient or nutrition descriptions (source code & bug databases), it would be physically addicting (vendor lock-in mechanisms), and you would not be allowed to share it, take it with you to a new kitchen or eat it at work (license).
Other points, which are not inherent in closed source, but seem to be the norm in the industry: It would be sold in sealed boxes (retail distribution), nobody actually producing it would be available for help (customer support), for the most part you would have to use special cutlery (OS / hardware) for it, and every so often your recipes incorporating this food would go haywire (incompatible upgrades, bad standards support).
Open source food, on the other hand, would come with a recipe for how to make your own (source code), virtually no expiration date (license allows for porting), and often free help with your recipes (community support).
OpenDocument is a great example of the superiority of open source: If you uncompress an OpenOffice.org Writer file, you'll find the original images, application settings, contents, and text styles in separate files. If you want to change some detail in the text or an image, you can manipulate the file in a suitable editor, and then compress the files again to have a fully working document. Most of the files are XML, so you can also automate data collection or manipulation in several documents easily. No such luck with Microsoft Word: Everything is saved in a proprietary, secret, binary format, which the OpenOffice.org team has used years to decipher, in order to be able to import MS Office documents.