Threads does support valid output, only instead of shoving in un-needed HTML we use the cdata tag and output actual breaks as they're inserted in the posting. Any decent feed reader should be more than capable of parsing the data.
If you view the actual source of the page where the RSS feeds display, you can see that the text *IS* breaking at the line where needed, however the cdata with some "browser readers" parse the text together because the "additional coding" (like some feeds produce) isn't there...
I do say that RSS feeds need some work (such as providing the author's name of the post in the feed) but I don't really classify the shortcomings of a feed reader as a cause of updating the coding...
