Proliferation amongst the masses?
A common theme in many Linux vs. MS posts, including this one and several comments, is the need Linux advocates have to "beat" MS in the non-geek market -- to completely blow away Windows. This hyper-competitive attitude is counterproductive and only can inflame MS to more extreme attempts of wiping out Linux.
The fact is that we don't necessarily need Linux to be the desktop of choice for average Joe User. This is not to support the ridiculous notion that Linux should be kept as an "elitist" OS or that we shouldn't develop user-friendly tools that "real" sys admins scoff at. As a developer running Linux at home, I didn't really want to become a sys admin as well. To me, configuration tools, such as linuxconf or kPPP, are an invaluable aid. I still make some configuration and installation changes by hand, but at least I have the option to go either way.
What we do need for a healthy relationship with average Joe User is simple interoperability. We need application file formats that can be easily used cross-platform. I'm sure that I'm not alone in avoiding firing up StarOffice to edit an Excel spreadsheet just because it's such a bear. We should be investigating ways to get Gnumeric (,etc.) running under Windows. Installing a new OS is beyond most users, but installing a downloaded application (a la InstallShield) is no problem.
There are those who argue (rms?) that porting applications to Windows will only serve as a crutch and remove the impetus to switch to an open OS. My take: average users will never install a new OS (upgrade, yes; install, never). Our best bet is to get the users familiar with open applications -- then, the next machine they buy will be preloaded with Linux or another open OS. After all, it'll have the apps they want and it'll be cheaper. (And probably one of kids will want to play around with tweaking the kernel :)