My conclusion is that the architectural principle behind mime types provided by web servers is wrong. Web servers have barely any more ability than the rest of us to guess content type, [...]
The design approach of server provided content types is an attempt to avoid guessing. Lucas' assumption that the server is or should be guessing is just plain wrong. If there is guessing going on, then the server should not provide a content type header in a response.
I just saw that Lucas saw my post I made & I apologize if what I said came off as flippant - reading it now I shouldn't have said 'silly' - that was unprofessional. Sorry.
My only excuse was that it was really really late...