Re: metamail

[Karl Kleinpaste <karl@justresearch.com>]

I'm well aware of where metamail originated, and in fact I
experimented with the very first public release he made of it (with
related tools) in ¿1992?.  It is still the case that it's quite ugly
in some ways, and is incomplete, as we've found when looking for MIME
comparison points against Gnus.

Borenstein published a fair amount of commentary about having spent N
years working with things like CMU's Andrew Message System (a
seriously spiffy and capable system which supported advanced concepts
like embedded animation, as far back as the mid-1980s; I recall
watching him demo this kind of thing at the January 1988 Usenix Conf.)
and yet, much later, found himself back-porting MIME-related hooks
into ancient crufty mail handlers like UCB Mail because, as he
related, composition of such messages is Hard -- that one can spend a
great deal of time putting together even a short message when one is
trying to get the right presentation of graphic, animation, and text,
and that there is a "diminishing returns" factor involved, which
limited the AMS' ability to achieve penetration and acceptance.

I consider the deficiencies of metamail in particular to be an example
of the realization that effective rendering is apparently equally as
Hard as composition.