Everyone who wants to be a thoughtful advocate of [ free | open source ] software should really go read raph's 10/31 diary entry. It's quite insightful.
Back in my transition from school to industry, I read (ugh) some awful assigned text on Software Engineering. It was awful because of being a classic example of Soporific Prose, but it was worth enduring for the key insight that a software product is more than just source code. A software product consists of the totality of code (source and binary), documentation (online and printed), test suite, build tools, support infrastructure, and (dare I say it?) marketing collateral. And the quality of a software product, as a product, depends on the quality of all of these things, not only on having an elegant design and clean, bug- free code. (Good design and code are necessary but not sufficient conditions.)
This is where proprietary commercial software vendors have traditionally had their advantage. It's a very rare person who has an "itch" regarding test suites and code coverage metrics, or good user guides, and the time and ability to scratch it effectively. This is what companies like The Great Satan can do so well: afford to hire people to take care of these "fit and polish" issues. I'm hoping that the new models of commerical free software will be able to do the same thing, but it's probably too soon to say that they've all proved that they can compete long-term against the proprietary vendors (and this varies in interesting ways by market segment).
The other thing that raph correctly notes is that it's hard to find quality free software implementations in areas that require very specialized application expertise (such as high-end graphics). Proprietary vendors have the cash to affort application specialists and software specialists; free software efforts seem to rely on finding the rare combination of a double application/software expert who is also philosophically a free software fanatic. Not to mention, having a day job which supports writing software to give away. Fortunately, such people do appear from time to time, but the scope of their work is limited.
I'm not sure how to tackle this problem, other than to (a) make sure to support worthy free software commercial enterprises, and (b) work on educating the free software community about the utmost importance of non-coding contributions, and making sure that such people get their share of the credit as well.