Microsoft Research and the University of Washington have been doing a lot of research into the way we will browse our photos in the future. The end process analyzes the picture pool and places the photos into a virtual world that the user can navigate, automatically choosing the best resolution picture for the given view.
This allows the user to focus on the picture content, not on guessing which keyword might match up to the desired photo. Beyond that, the process can also suggest the best "band" of viewing for the most available photos and even link it to any available video for a given viewpoint.
Obviously the process works best when there are a large amount of photos to provide the different angles and viewpoints, so it might seem a natural fit for community based online photo repositories. It would also be fantastic for online encyclopedias, where any of a multitude of photos could give you that particular visual data of a given area. Nice.