Currently Time Is placed in the attribute field, making it associated with all of the geometry of the object. In the case of a single point, this is ok, but for more complex geometry, mulitpoint, polyline, polygon, it is not. We should be integrating time into the geometry of each node. Each node should have X, Y, Z, M, B, D and where B and D represent the time of Birth and Death (Or start/end time) of that vertex.
Where would this be useful? My immediate requirement is for georeferencing video. I currently put time (or a derivative of it like a frame number) in the M column, but when the video stops at one place (vehicle stopped), then there is a problem. Since the video didn't move, it should't move spatially, only temporally. So the M value should change.. But I wouldn't want to be storing multiple points at the same location (besides in some cases it is difficult because of XYZ and M tolerance). Thus if we add the Birth and Death of the Vertices then I can have a temporally correct representation of the location of the video. We have video represented as a line, because it makes it made an efficient way for figuring out where to start the video based on a user click (reverse geocode).
If we are going to model the physical world, then we should strive to embed the fundamentals of that world in the models that we create. Time complements space, and so incorporating it at the most fundamental level would lead to additional capabilites for the analysts building the next GIS.