As for browser display, i don't believe that recoding how our rss feeds work for IE to display it in a logical manner is really a decent fix... If they parsed the cdata as shown (vs trying to render non-existant html) it'd read it with breaks
Threads is already doctoring the cdata content with html. It contains html to bold the "Posted by", and adds a <br> after the author. How would any RSS parser know when to stop using the html, and begin parsing the rest?
Here's a sample cdata section:
<description>
<![CDATA[<b>Posted by</b> Gary R:<br />
A paragraph here....
Second paragraph here...
Third paragraph here...]]>
</description>
