Andrew Gerrand worked with me on Camlistore too and is one of the Upspin authors.
The main difference I see is that Camlistore can model POSIX filesystems for backup and FUSE, but that's not its preferred view of the world. It is perfectly happy modeling a tweet or a "like" on its own, without any name in the world.
Upspin's data model is very much a traditional filesystem.
Also, upspin cared about the interop between different users from day 1 with keyservers etc, whereas for Camlistore that was not the primary design criteria. (We're only starting to work on that now in Camlistore).
But there is some similarity for sure, and Andrew knows both.
I think both Camlistore and Upspin have promising models.
You mean work is progressing on camlistore? That would make me a little bit happier : )