Interesting comments, Gandalph and JPWhite. So what would be your guesses about the proportion of installed PCs that are too-new-for-Knoppix, and where do they have problems?

(My guess would be that 10% of desktops are too new for The Community->Debian->Knoppix to have played around with and adapted to, but half of those 10% work anyway. And what breaks in those cases is almost exclusively video-related.)

One of the reasons these numbers are important is that testing applications and supporting them on Windows 95, 98, ME, NT, 2K, XP and their various patch-and-service-pack combinations is a serious burden on a lot of organisations. With Knoppix derivatives we could skip the installed software environment altogether and just test and support our products in a single setup.

But that only works if they run out of the box for a significant proportion of the installed base. My company, like a lot of others, currently ships software that will probably only reach about 80% of the Windows installed base. (No 95, no NT, and 98 only works for Second Edition.) And when a new service pack comes out, there's always a chance that another 40% (all the XPs that install it) will suddenly go south as well. So the bar for a useful Knoppix derivative may not be all that high.

I take it no-one really knows for sure what proportion of PCs out there Knoppix derivatives can reach. But what are our best guesses?