The website of The Baltimore Sun (http://www.baltimoresun.com) is a logical next step in the progression of web CMSs.
The main feature I want to draw attention to is the escewing of human readable urls and the presence of arbitrary IDs instead (see http://www.baltimoresun.com/news/nation/politics/bal-debate0426,0,803851.story?coll=bal-home-headlines).
I am not crazy about the trade-off, because I don't think it does much for readers. It seems to be mainly for the convenience of the programmer.
On the other hand, I happen to know a bit about the evolution of the site. I am aware that they made a transition many years ago to strongly encouraging users to register with the site. This, plus their efforts with cookies (and maybe other techniques at effecting state) allows the development of ad products that are tuned to users behavior.
It turns out in newspaper publishing, targetted advertising is a major source of interest -- if not revenue. And in newspapers (old media) you have very prominent mutual understandings and relationships between content developers, advertisers, and products. And now, readers themselves are being added to the mix, so newspapers are collecting dossiers of readers' interests and habits and seeking ways to commodify access to these readers.
Another interesting aspect of the site it its selective use of inline hyperlinks to other BaltimoreSun.com content. This is sometimes productive (i.e. related to the story or a series of news events), and sometimes seems kind of parasitic (i.e. seeming to be more motivated by a desire to promote other featured content on the site).