The fact we're seeing fonts (ab)used in this manner (and for icon fonts) says that we're sorely missing a good technical solution.
So how did web fonts end up as the approach? Because that's where browsers first acquired the notion of confluencing semantics and vector graphics. Fonts demand all the advantages of those entities: semantic information to define characters suitable for mapping to glyphs, the ability to apply styling to those glyphs (bold/italic/etc), and raster-independent scalability.
Could SVG do something like this, include semantic hooks in a file for applying styles? (I'm not too familiar with the format.)
The complex part is usually simply working out which country/region/whatever corresponds to which <path> in the first place; but this is a complaint more about the quality of the markup of the underlying SVG of an average map on Wikipedia, than about SVG itself.
I have to say Improvely looks solid by the way.
It doesn't display the entire map like this (so the use case is different), but it's pretty cool all the same.
