I did a search on this but I did not find anything much relevant yet. I was wondering how come there is no GIS software that would take a sufficiently accurate 3D surface model of the world (such as based on SRTM's) and put geographic features on it instead of a simplified spheroid.
This would obviate the need for flattening the world using projections, hence remove all distortions, enabling us to make much more accurate measurements. We then wouldn't need to use different projections for different purposes or for different regions of the world.
I'd imagine when we look at the interface of this software, we'd be looking at a curved map, flattened only at the screen. When we zoomed out enough, we could see the entire world as a globe (as an option). The major problem with this approach would be the relatively large size of the 3D surface grid, demanding computation power and a need for developing new libraries for computations.
I think the most of these issues could be solved by getting only the portion of the grid necessary for one's mapping needs.