I absolutely agree. Also, if you want to parse HTML as XML, it's a lot more reliable having it in a known 'good' format to begin with.