|If you look at the history of open source, I think that it would be hard to argue that it took off because of an effort to lower costs of white-collar programmers. Most open source software and software development made its way through the back door, pushed by system administrators and programmers. Even at IBM, from the stories I've heard Gerstner's decision to embrace open source happened only after a sustained "grassroots" campaign by programmers and lower-level managers had invaded IBM.
The issue of open source and skill isn't straightforward. It depends on what open source software you are talking about. For example, the programming languages Perl and PHP spread like wildfire because they weren't too difficult to learn, there was lots of free code available, and there were huge communities who freely supported them -- if you got stuck, you could usually get an answer from a listserv or newsgroup in a few hours. Similarly, these days setting up an Apache web server is pretty straightforward. Linux is another matter. I'm finishing up a study of nonprofits who use Linux on their servers, and what we have consistently heard from system administrators is that Linux takes significantly more skill to get started than, say, Novell or Windows 2000. However, their sense--and this has certainly been my experience from supervising sys admins in both environs--is that it takes about the same level of skill to be able to handle more complex problems (there are a lot of people who are Novel/MS certified who have no clue what to do if something goes seriously wrong).
But if you were going to make a statement overall about open source and labor costs, I'd say that it's saved companies money in two ways: for software like Linux, it lets companies hire fewer but higher-skilled programmers and admins, it lets the company harness volunteer labor, and for some large projects it increases the amount of code that's available for free (since code is shared).
But I don't think the real savings come in labor costs. The biggest savings are in hardware costs (OS software tends to need less powerful boxes), ongoing licensing fees, stability, security, customization, and reduced risk (for small niche software, where you have to worry that the company that built it will go under or will stop supporting it).
I think where the issue of lowering labor costs is going to become a big issue is in the new surge of outsourcing. Open source development projects have gotten very good at having volunteers from around the globe working together. In other words, they've figured out the processes and software infrastructure that make it much easier to use Java programmers in India, who get paid a fifth of what they do in the U.S.