> now you're moving from the realm of specifying 'features' or > 'default handling' of the object into specifying layered encodings. > and this was something that we were very unwilling to do when > we defined MIME. The feeling was that having an arbitrary > number of cascaded encodings was likely to be detrimental to > interoperability.
I thought this too until I realized these were simply strawman names in Martin's example.