Right now if you go to a page (including a content page) with a query string that query string goes into the canonical url string. For example:
<link rel="canonical" href="http://en.wikipedia.org/wiki/Schadenfreude?cat" />
<link rel="canonical" href="http://en.wikipedia.org/wiki/Schadenfreude" />
This has caused issues where search engines have indexed bad urls (with people's name or libel threats in the query string) which should obviously never have gotten into their system in the first place but when they check to see if it 'exists' they are getting an answer of 'yes' (as far as they are concerned).