For those, this Wikipedia article should make it more clear: http://en.wikipedia.org/wiki/Collaborative_real-time_editor

Alfresco is very “industrial strength”, but it does everything asked for here (plus a whole bunch of other stuff). They have a commercially supported version, and a GPL’ed “community version”, which has all the same functionality as the commercially-supported version. A key strength is that it has CIFS/SMB and WebDAV access. This lets you easily copy documents in and out of the Alfresco, with no need for any special “thin client” (in both Linux and Windows clients). You can even “map a drive letter” to the Alfresco server in Windows.

It has lots of Open Source technologies under the hood, check out their “technical specs”

It is probably overly complex for most small groups’ needs: it does all kinds of fancy workflow management, enforcing of fine-grained permissions, etc. It’s very “all-singing, all-dancing”. Their target market is large enterprises.

I too hope someday there emerges a sort of “Alfresco-Lite” that just does ODF document collaboration in a wiki-like fashion.

For transport, both p2p and client/server should be possible. I guess each participating application would need to understand the concept of Edits for the pieces of information that are exchanged as fast as possible. So the transport layer sends/receives Edits as opaque elements and the apps have to make sense of them.

Edits should be viewable in a log format like IRC chat. In collaborative editing, the same flow from synchronous to asynchronous should be expected.

There needs to be versioning, branching, merging and annotations. Edits should be tracked with information Who did it When.

Users should be able to ‘share’ (make available, expose) on the version level (private/public versions …)

There should be a central ‘Person’ or Contact type of object to bundle address information, to represent state (availability) and for associating Edits/documents/activities.

[Mark: I touched alot of this in my "Rethinking Desktop Computing" document, send to Ms Newman]

Also, with jingle we would have already a telephone and a video conferencing software (ok, some bits not ready yet, but that can be done).

A good start would be to first have a good jabber/jingle client (maybe in telepathy with an example client to not disgruntle gaim and kopete) with full audio, video and file sharing support. The next step would be to add a collaborating sketch board (Coccinella is a client which can do this already, I think, there should be some code usable; also the kdevelop guys develop something like that for their development suite) to be able to collaborate on simple drawings and sketches. After that a simple text editor and a simple wiki should follow.
With the experiences gained from these projects the next big step would be full integration into OpenOffice/KOffice – and together with this integration the project could settle on stable APIs which other app developers could use to also integrate this new awesome app with their apps.

Last but not least there would follow a collaboration management server for collaborational work.

I think that this sounds like a cool plan – the only thing left now are some developers and some people with a realy vision to drive the project and the developers into the right direction.

It is true that similar funcionality as OSS is quite desirable