IBM's versions, from 2.0 onwards, extended that with better GUI support, real multi-platform interoperability and backwards compatibility. I was an OS/2 developer from 1.0 through 2.1, and a participant in IBM's OS/2 developer's program (I even purchased several big PS/2 machines, very nice for their day, and gave presentations to groups of bored scientists about it). I was also heartbroken when Gerstner pulled the plug on OS/2 and effectively doomed it to obscurity. But, I've since moved on to Linux and FreeBSD, and I haven't looked back.

If OS/2 was going to be open-sourced, I think the parts of real interest would be the pre-2.0 code and some of the 2.x kernel extensions. But, so far as I know, 1.x is mostly MS code, and will never see the light of day. I don't know if there's enough of the 2.x that isn't legally encumbered to be of interest to anyone.

It's a pity, though, because it could have been what Windows wanted to be, and Linux could be. But, that could be said for NextOS and BeOS as well. Just because something is a good idea doesn't mean it will enjoy success in the real world.