The idea is to explicitly represent a user's view (or views) of the world with a vobject. (These would be separate from A3dlViewpoints, which just represent possible locations to stand and face). View objects could be created by a server for each client when it joins a world, and perform some view culling on its behalf (so the client doesn't have to download all objects). This could also be used to show personalized views by applying some general selection criteria, or from game logic.
The server could also provide various properties that affect how the client does its rendering. For example, providing bitmap filters -- [A3dlViewRasterFilters] -- to be applied to the final scene, or default 3d shaders.
The idea of a view would also be useful for HyperVos, to provide personalized view when you log in and associate your browser with an Identity, or modified by a CGI query.