Google Docs, Webspiration, Voicethread, and Etherpad (one of its many manifestations is linked below) are all examples of these. I've used a couple of tools recently that highlighted the limits of collaborative environments, however, and the experience helps underscore a couple of good principles to work from when setting up a collaborative learning activity. When I worked with 8/9th graders in building The Parthenon with OpenSim the project went really well until the last session. There were just two specific tasks to accomplish--placing a few more ceiling pieces and lining the rest up better and doing the same with some columns in the chamber of Athena--and students were concentrating on doing them well, knowing that the appearance of the final product was up to them. But they kept running up against the problem of two people moving pieces around at the same time. Interestingly, there is nothing built in to OpenSim to prevent multiple people from editing the same prim. Because they were moving them into place based on their perspective, both were often trying to move them into different positions, undermining each others' efforts. The problem came down to there not really being enough for everyone to do.
Etherpad (great version at typewith.me--35 collaborators and no registration) is another wonderful tool for collaboration that I've been using with great success in my robotics class as a way for pairs of students to document project descriptions, plans, pseudo-code, and program versions that allows me to see who is contributing what. But before I start students using it I open it up as a sandbox activity for everyone to try for a bit and get their desire to play with it out of their system. It always ends up freaking them out a bit, because with a blank page and 18 students trying to type at once no one can get anything done. For this reason, when an English teacher asked if she could have her class use it to collaborate on a sort of fan fiction activity, I cautioned against her just tossing them in there simultaneously to start writing. They would just get in each others' way. I suggested small groups take turns working on it while the rest are doing something else, or somehow starting with a big chunk of text already there so they would be less likely to undermine each others' work.
So the principles I can think of that help manage collaboration with these robust tools are 1) make sure there is enough meaningful work for everyone to do, and 2) provide systems and procedures to ensure no one is duplicating efforts of anyone else.