I find that problems of ignorance tend to stem from conception rather than capacity. It's not that people couldn't understand, just that their mental model is sometimes askew as, because they don't want to "waste time" reading, they reason the nature of how something operates by its external effects that they're most familiar with. It's one of the reasons many people have trouble flipping from Windows to Unix: The idea that the desktop is really just an everyday app like any other twists their conceptions to extreme if they've only ever known Windows.

On the other hand, you get "actions of necessity". Most people don't know, or don't try to, because most people don't need to; hence why we're paid. As you say, though; it comes in useful at pay review time.

I'd only meddled with 'subversion' in the past (personal tinkering, hobby projects, that kind of thing ...), but for the past eight months we've been using it seriously on a fairly major project (8,000+ revs on the main branch so far), for which I configured the server.

I know it has its detractors and it doesn't do some things well but, I have to say, as a replacement for CVS, I found it pretty damned decent and would recommend it over certain others. There are a wide range of front-end "plug-ins" and GUI tools for it too; from Eclipse, from KDE, from Windows Explorer, stand-alone apps etc., so people can use whatever they want as their personal client (I tended to use the "official" command-line client in unix and TortoiseSVN on Windows) and this prevents project-wide dogma from messing with people's personal development preferences.

The only real problem we found was that, to use it effectively, it really helps to understand it at the command-line level. We had one dev who hadn't quite got his head around the update->work->update->commit cycle and, as a consequence, they were pulling everyone else's changes but not properly commiting their own. This was then compounded by them deciding to distribute patches via email (instead of implicitly, via commit etc.). Basically: As long as the devs understand it (which, IMO. any decent dev should be able to), it works great. I prefer the distributed repository model philosophically but, as far as the centralized repository model goes, you can do a lot worse for real-world purposes than SVN.