I think Mr. Waks is right. VRML needs an infinitely scalable communication
protocol, since VRML worlds can be arbitrarily large.
Any protocol that relies on one central server for anything other than a
one-time startup synchronization just wont scale.
I think we'll need a protocol where the communication paths between clients
dynamically changes, based on the proximity of clients to each other. A
dynamic web of VRML entities communicating with each other, exchaning
information, with information propogating only as far through the web as is
necessary....
So, what are the equivalents of LevelOfDetail and view-volume culling for
client communication?