Take a dig through the archives; we've discussed this specific topic
before. (Indeed, not long ago -- it should be relatively near the
end.) Your opinion here is essentially the same as mine: the "VRMUD"
engine is a separate beast from the "rooms" or the "viewer"
itself. (Indeed, I've been arguing that you can easily have multiple
"VRMUDs" that run off the same "rooms" as their base datasets.) A
sophisticated viewer would synthesize the rooms being uploaded with
the data from the VRMUD into a more interactive scenario.
However, note that this isn't the only way this could be handled, and
I don't think we've hit consensus yet. As I recall, we sort of tacitly
decided not to decide this issue quite yet -- first, we're going to
worry about getting the basic viewer technology up and running, then
worry about the harder problems like this...
-- Justin
Random Quote du Jour:
"The day after tomorrow is the third day of the rest of your life."
-- Dfan