The parser for URL tracking mangles certain URLs. In particular, it seems to do this with URLs that end with a hyphen '
'. In this case, the original URL ('http://www.example.org/topics/this-and-that/news/2010/12/some-people') is mangled and what ends up in the civicrm_mailing_trackable_url table looks like this:
The 'href="' component shouldn't be there, I don't think. And the end of the URL gets cut off (the trailing hyphen). The overall effect is that the click-through link doesn't work and the user gets a page not found error.
I know that having a URL end in a non alphanumeric character is somewhat uncommon, but I don't see why CiviMail's parser should break them.