Google's Matt Cutts tweeted a few days ago that he didn't understand  
why Twitter and similar services don't simply resolve short URLs to  
their long form and store/display them that way.

Things like that have been on my mind for a while, but I've only just  
put some of those thoughts to words:

And from the perspective of linked data, making our applications query  
the URLs that users submit to them just makes sense. It might seem  
like science fiction to suggest that Twitter resolve a URL to identify  
its canonical version and RDF that enriches the tweet, but Facebook's  
link sharing actually does that (though it looks for meta tags rather  
than RDF).


> ...rather than creating a new link rel="canonical" and BTW their  
> strategy
> only works in HTML, it doesn't work in RDF, JSON, XML, etc...