To perhaps clarify, you're more than likely outputting UTF-8 encoded text. Given that there are unicode characters that represent a non-breaking space, or in your example, the copyright symbol, kid optimizes. This is a very small win, in that you're saving a few bytes per character. You'll notice that if you use a less expressive encoding, such as ASCII, Kid will use the appropriate HTML entities. Also note that it's smart about this. & is not translated, as it's a control character in some cases. It's a two-way street: if you have the raw character in an UTF-8 template, but wish KID to publish in ASCII, it will convert it to the proper HTML entity.
So, as fredrick is suggesting, Kid is NOT screwing up anything, it's merely converting it to an equivalent and portable encoding that the browser doesn't worry about. If it supports UTF, it won't matter.
