Bruce Lilly wrote:
One item that could be improved in next-generation email is better
orthogonality between envelope, header, and body.
If you're thinking at that level, you have to go further.
1. Paper mail doesn't have to be the reference model to follow (or
to selectively avoid following).
2. The US model for paper mail is not the only one.
3. The words "envelope", "header" and "body" do not contribute to
worldwide understanding.
By these I mean:
1. There are traditions in paper mail that we just don't need in a
new thing based on our experience with first-generation e-mail; and
other behaviours we do need for which paper mail gives at best poor
prototypes. For instance, we don't need to replicate the agreements
for international service or for interworking between multiple service
providers (I didn't say we didn't need any, just that we won't copy
those); but delivery of the same message to multiple recipients is
just not part of paper mail.
2. "Mailbox" is a good example. It was a long time before I worked
out that the concept just doesn't exist in the UK postal service,
and that North Americans have a big collection of expectations
about it that were foreign to me. And I'm an English speaker from
Western Europe; I have to assume that the gap for some people in
(say) Asia or West Africa is greater still.
Then there's "envelope". We have in the UK, for instance, no
convention that you record an authoritative source or return
address on the outside of a letter (though it happens sometimes;
I do it when I believe the recipient's mail service will expect
it). I'm not commenting on whether that's good or bad; just that
it's different.
3. I do not believe there is consensus in a sufficiently wide
community on the semantics of "envelope", "header" and "body"; nor
that if we make them exclusive (or orthogonal) they will necessarily
be exhaustive and cover all the required behaviours in a sensible
way. The words have become jargon in the worst sense, that although
there is a community who probably do agree on a few concepts and
need words for them, there is a wider community who have a different
set of meanings for the words we have chosen; see 2 above on
"envelope".
If there are three concepts, it should be possible to describe them;
but please, not by further overloading of names. It would be better
to describe their functions (eg management of transmission,
management of authentication, management of user agent behaviour
etc) rather than drop hints about their physical representation.
Bruce is otherwise right, of course; the more nearly independent
the chosen base elements are, the better. And he did give several
examples of functions and their grouping.
I don't know how many people have the time or energy to look at
the philosophy of e-mail, nor whether it would result in
next-generation services and products appearing sooner or later. It
should make them better, of course, and perhaps more enduring; and
we should not forget how long the stuff has lasted that we now
regard as not-so-good and are trying to replace.
Rodney Tillotson, JANET-CERT
+44 1235 822 255.