🚀 go-pugleaf

Here are some of my thoughts on this subject:

1. I consider the lack of a good receipt system to be a major limitation
to internet (small i) mail.

2. In snail mail, the intent of receipts is not to report that the mail
has actually been read by the intended human and comprehended.
It is instead an indication that the recipient or the 
recipient's authorized and identified agent has been made
aware of the message, and therefore the intended recipient *should have*
read it (or otherwise dealt with it).

In other words, these receipts say "We delivered this message at such-and-such
time, and so-and-so at the proper address signed for it,
so it isn't our fault if the intended recipient hasn't read it" or
"We delivered your mail to the proper address at such-and-such time, but
so-and-so refused it" or possibly "We tried to deliver your mail, but no one
could be found to accept or refuse it".

3. Most proprietary Email systems have a receipt system, with similar semantics
to the snail-mail receipts.  They also generally maintain more control
over the mail before final display than the average UNIX mail implementation,
so the receipts are fairly reliable.

4. The invasion of privacy of receipts is frequently mentioned.  However,
under the definition of (2) above, then the purpose of the receipt request is
to absolve the delivery system of responsibility and to place the onus
on the recipient.

By analogy, if an Email UA offers the opportunity for the recipient to
refuse to send the receipt back, it should also refuse to show the mail
to the recipient, and a delivery-refused indication should be returned.

5. Not all internet mail is handled in a single interactive SMTP hop.
Even Internet mail can still go through several hops via the deprecated-
but-not-eliminated mechanism of source routing, and via the tacitly-approved
"%" hack.  Therefore,  presence or absence of certain SMTP indications
does not necessarily indicate delivery to the final mail store, let alone
having been presented to the intended recipient.

6. Two other notes of the snail-mail analog:

a. One aspect of a snail-mail receipt is a signature.  A truly accurate
Email model would require a digital signature system.

b. Snail-mail receipts are nearly always extra-cost items.  I don't know
of any mechanism for modelling this in the internet environment.
-- 
Craig Jackson -- cjackson@mgh.com (Straight Internet)
dricejb@drilex.dri.mgh.com (uucp-forwarded)
{bbn,alliant,redsox}!drilex!{dricej,dricejb}