This works for information that's physical or pictorial. The examples cited in the article (e.g. Google Earth) fit into this category. Other forms of information (literary, musical, social, mathematical, you name it) aren't nearly as easy to represent this way. There's a reason why humans use symbolic notation.

This article is OK. I thought it was unfortunate that there is no mention of the late Jef Raskin, who covers zoomable interfaces at length in his book, The Humane Interface.