The way multicasting works is this. Ordinarily, the net server sends packets from one location to another. Under multicasting, any number of stations can receive the same packets, hence participate in the same session. It works like a multimedia conference call, but differs in the mechanism. In effect, domain servers which are mbone friendly, can create a maximal tree in the subnet that wishes to be connected. Now, each node puts packets on this tree for the other nodes to read and respond to.
At the moment, the voice part works pretty well, the video part pretty poorly (band limited), and is used mostly for face recognition. But one can learn to use the white board. HTML and TeX documents can be served flawlessly, and there are electronic sketchboards which work well with a stylus (poorly with a mouse).
This is what our participation could look like. At several locations in the department there would be an MBone station, consisting of a WindowsNT work station, a 21 inch monitor (or projector) with a suitable video card (for the video camera), sound card (for the two way speech), and graphics card, and pen-drawing tablet.
The location would influence the way this is used.
This is the main location at first. It can use the MBone both for informal and formal commmunication. Informal communication means that mathematicians interact in a private session. Formal communications can range from a single-session multicast of a seminar, to a multi-session series, e.g. a distance education course.
This is a temporary (secure) location during the debugging phase. We need two stations to learn how to use the MBone inside our domain, so as not to create traffic on the global MBone. But it could continue to serve for technical sessions (e.g. editorial conferences to edit and review documents, closed meetings of commitees. )
Here we expect individual mathematicians to visit with each other casually and ``do mathematics'' at a distance as we would do it locally.
This would be for formal sessions.
The project is too expensive to implement all at once. Here is a feasible plan:
First Phase: Two MBone stations, one mobile.
Install the home station at (1) and serve (2)-(4) from an a mobile MBone station. This is an extension of the ``Indy a-la-carte'' project. This would now be modified to be a Pentium-Pro machine on a movable cart for temporary installation in several places, as we test out the feasibility of the location, and get our colleagues hooked on this opportunity to communicate with their colleagues on the coasts.
Second Phase: Install rigid (un-stealable) stations wherever the demand warrants it, the mobile station could continue to serve those that do not warrant a fixed installation.
Third Phase: Install it wherever mathematicisans gather to communicate. There should be one in the chair's office, for example, to insure privacy. The should be one in the commons room, in the library etc.