Okay, here is the deal: I set up a brand new blog (private). The default character set encoding for the blog is UTF-8 (as is standard on new installs). So the data is saved in the posts database is UTF-8 format.
However, when I pull the feed and check it with magpie, it comes back as:
[feed_type] => RSS
[feed_version] => 2.0
[encoding] => ISO-8859-1
Which means that if the feed is republished on other sites using standard tools, they end up with extra characters in the posts (the imfamous Â instead of a double space, example) because the feed was decoded as ISO but the content was actually UTF-8.