the idea of scalabe vector graphics is to describe a picture and not to tell which pixel looks how. This is a bad way for photos and a good way of "artificial" graphics like chesspieces and ratinggraphs. Such is "picture" is just (XML)text, which say from here to here is a line, there is a filled ractangle and these is an ellipse with a gradient (changing color).
Such a text is very fast and easy generated and it's the browser who does the rendering ("computing how the picture looks like"
, so I thought this idea could be interesting for Russ.
For example the rating graph is just a huge number of ractanges. And there are some others...
Got what I mean?