I have setup WordPress blogs for a few people now and am hitting a common problem in that a lot of people copy content from MS-Word straight into WordPress.
The problem is that Word has already converted many ' into â€™ and " " into â€œ â€?. This makes the page not validate as it's not in the UTF-8 characterset anymore.
It is fine if you enter the ' and " characters into WordPress and allow WordPress to convert them to #8217; ... etc. But if you paste from Word, you hit this problem. Sometimes you even get a little question mark character show in the blog when using FireFox, for where this non UTF-8 character is supposed to be.
Maybe, would it be possible... if there are higher level (non UTF-8) characers in your post, to convert them straight to things like #8217; #8220; #8221; to keep it valid.