In the glorious future, when we have a general outdoor AR tracking system, our mobile devices are going to need data. A lot of data. Apart from all the content we’ll be downloading to use to augment the areas were in - tasty 3D, animated, shiny, fat content - we’ll probably need some kind of fairly dense 3D model of the surroundings we want to track against.
Both model and content density should logically become proportional to the popularity of the real world space. The center of major cities will be more densely mapped because in tightly packed cities every inch of space becomes important. Similarly there will be more data positioned in those spaces because there’s a much greater potential audience and there’s a much greater density of things to augment, which also motivates denser models for finer scale tracking.
Of course, both the importance of space and the size of the audience are straight functions of how many people occupy (rather than inhabit) a space at peak times over the course of a day, and that’s why I think it’s inevitable that AR content becomes P2P distributed. The population density/spatial resolution conundrum of requiring more of both kinds of data solves itself, rather than being a burden on servers close to populous areas. It’ll begin with people sharing models of the environment across the tracking platform they all share. From there it could expand to using distributed download to propagate the channel or ‘feed’ style AR data discussed earlier.
As an aside, this makes perfect sense for acquiring data, but pushing new data to the service will probable remain bootstrapped by a boring old direct push to a central server that can function as tracker.
