The upcoming battle between Microsoft (Virtual Earth) and Google (Keyhole/Google Earth) is already receiving a lot of attention. What is interesting to me, beside the raw eye candy and fun of the whole thing, is the idea that we are moving to a rich client interaction with web/internet data. Browsers know a lot about a very small space of object data - document encodings. Rich clients will know more about different types of encoding, the way they interact and appropriate ways for users to interact with them. Imagine coming across a table of data in a document - in an html page - and being able to view the data in different ways, sort the data, pull in related data and so on. Adding scripting to clients, as a bolt on to the object level document layer, is not the way to do this horizontally.