A few remarks about CSCW and Hypertext and web based collaborative tools. I write this using an unusual presentation form (data model / protocol networks) that I could describe as the infrastructure parameters.
Groupware systems may impose more or less structure to their underlying data model, and they may be dependant or independant on network protocols.
The Coordinator (Winograd and Flores) was a system to help a group of people handle their emails. It used email network protocol (maybe SMTP) and a kind of hypertext model to describe relations between emails (proposition, counter-proposition, denial, etc). Its purpose was to kept track of on-going email conversations. Its utilisation cost was that each new message had to be tagged in one of the argumentative category. Categories and possible conversation graphs were drawn from Searle's speech act theory of conversation. It has been much critized (sometimes excessively as a fascist system that blocked social innovation in organizations). That was a commercial PC product in the late 80ies.
Terry Winograd et Fernando Flores. Understanding computers and cognition: a new foundation for design. Ablex publishing corporation, Norwood, New Jersey, 1986. Ð 207p.
I think Group Decision Support Systems (which were formerly built into electronic meeting rooms but have migrating towards client-server architecture) are more or less descendants of The Coordinator. I don't know much of gIBIS but that's a model to describe such argumentation networks.
Commercial GDSS (I think with client-server propietary network protocols):
I didn't find references but GMD has published articles about SEPIA and HYPER-SEPIA (see ACML DL).
Have a look at Synchronous Group Editors which have been experimented at DGP Toronto in the early 90s (ref: Baecker, Posner, Mitchell)
There is no data model as the textual document is just a flow of bytes replicated with a low granularity level (character if no locking, otherwise the size of the block of text which has been locked by an author I guess).
A commercial product on the same principle: Aspects
Teamwave Workplace (client-server architecture) is a spinoff of work made by Greenberg & al around a groupware toolkit (Groupkit)
It implemets a shared workspace where you find all classical groupware widgetery (file sharing, blackboards, URL sharing, calendars, to do lists, and so).
These guys have published an impressive amount of paper which has always impressed me when you start to consider usabilities issues concerning their tools. I think this is more a seminal work to experiment many groupware widgets (to provide awareness and so) and to test toolkits.
I would put in that category all the successful basic applications which put together:
Some of them use propietary solutions (Hotline, ICQ) but other systems are web based (html and http):
BSCW illustrates some classical groupware requirements like provision of awareness, versioning and direct communication between users. However they had to use manz tricks and develop code for that purpose. If Hypertext was mature for collaboration, building a CSCW cite should not require programming but only declarative description of groupware functionalities. This could be achieved maybe using special hypertext links (to upload annotations, to send an instant message to any reader of that page, etc.).
In my opinion what is missing to HTTP protocols and web server architecture for collaborative stuff is:
WebDAV is a possible solution.
I once imagined a solution that would consist in defining bnegative pending links (that is one author could leave in a web page some empty links that users would fill with links to their annotations). Thus the author still control where annotations can take place.
Another good illustration of backchannel communication is the pager link on Steven Pemberton's pages.
In that domain the CSCL community makes interesting work (see Simon Shum work)
In that domain it seems that a new use for web servers is emerging: web servers as meeting places to easily setup a real time or a conference between people. Everybody go to the same page at the same time where live information is broadcasted (for instance a power point presentation). This requires some plug-ins sometimes. The commercial argument (which is strong) is to set up electronic conferencing which passes through firewalls (normal with web/http combination).