I'm just starting to think about the avenues opened by Google Base, but this brings a question: why doesn't Google "simply" index metadata attached to web pages?

Look for example at what PiggyBank does, any web page having a link like

would have its metadata indexed.

And of course it doesn't have to be RDF. Talking about this with Sylvain the other day, he came up with

Where Google might accept the same formats as the Google Base bulk upload does.

Simple enough, or not? I wonder why they're not doing that already, it would be a nice way of making the web more semantic, incrementally, with little additional effort on their part (well, a few more containers maybe ;-)

The motivation for people to add this metadata would certainly be high: who doesn't want their web pages to be better indexed?