[ih] Early email systems

[Jack Haverty]

Hi Ian,

I know what "Notifications" means in 2018, but I don't recall hearing
that term back in the 70s.  Does it mean messages you get somehow from
"the system" about significant events - "You have mail" or "Your message
was delivered" or "Your message failed."?

Back in the early 70s, I was in Professor Licklider's (aka "Lick") group
at MIT project MAC.  One of his passions was using computers to assist
human communications, and we built a rather elaborate "Communications
System" as part of that.  It ran on a single machine, and therefore its
functionality as visible from the outside was constrained by the
limitations of the network protocols at the time.

Many of the things that could be done "inside" that system weren't
applicable when a message went to or from other machines on the ARPANET.
 E.G., there was a way to tell the system to capture any particular
message for posterity by archiving it on the Datacomputer - but no one
outside our group had any software to do so.  Much other functionality
along those lines was there, reflecting what people did in traditional
business and personal postal/fax situations as well as military command
and control (it was all funded by the US Dept of Defense...).

The system sent lots of "notifications" as it did its work, e.g.,
informing a sender that a particular message had been received, read,
approved, forwarded, or answered by one or more addressees.   These were
effectively new messages created by the system "daemon" as the author,
referencing the relevant messages(s) that the notification concerned.
BTW, I believe that was the genesis of the "Message-ID" field surviving
in today's email - I fought very hard to get that included in the
protocol so that our system could have a fighting chance of tracking
messages as they traversed the ARPANET.

Some of those notifications could escape our system, e.g., error
notifications that a message had not been delivered.   Those
notifications would be used to manufacture a (rather poor) "natural
language" mail message to the relevant parties.

So people would get a message from "COMMUNICATIONS-SYSTEM at MIT-DMS",
with a text something like "An error has occurred in processing a
message which may be of interest to you.   You sent the message at ...
and its subject was ...   The message failed because..."

When you sent a message, to other users within our system, you could
specify under what conditions you wanted to be notified - e.g., when the
message was in the user's mailbox, when it had been actually read, when
it had been answered or forwarded, etc.  But network protocols had no
way of conveying that detail so it didn't work for outside mailboxes on
other ARPANET machines.  Nevertheless, the daemon tried very hard to
send critical "notifications" of failures by manufacturing a message for
delivery through the ARPANET.

A common cause of such a notification was when someone "out there" sent
a message to a mailbox at MIT-DMS which was actually a mailing list, and
one of the mailing list members no longer existed, or their machine was
down, etc.   So the sender would get an unexpected message from our
daemon, telling it that their message to some user they had never heard
of had not been delivered.  A lot of the things we tried generated a lot
more questions about how to "do it right"....which I don't think have
ever been answered.

I used to get lots of such messages from my daemon, addressed to
"COMMUNICATIONS-SYSTEM-MAINTAINER at MIT-DMS", which was of course an alias
for my regular mailbox.  When it got into trouble, it yelled for help.
It's how I watched to make sure the system was acting properly, and get
notification that there was some problem that needed to be addressed
(usually disk space running low...)

My "natural language" algorithms weren't very sophisticated.  But
perhaps this was the first instance of "Robo-email"?  If you have access
to old mail archives of things like HEADER-PEOPLE in the early/mid 70s
you may be able to find some of these messages sent by the daemon by
searching for "COMMUNICATIONS-SYSTEM".

I'm not sure if this is the kind of situation you're seeking.  I did
recently find an ancient internal MIT document "MIT-DMS Communications
System Overview" from May 1975 which talks about the architecture of the
system, but doesn't have much detail about how it was used.  If it's
useful I can email you a copy.

/Jack Haverty
Doing Internet at:
MIT-DMS 1970-1977
BBN 1977-1990
Oracle 1990-1998


On 07/15/2018 08:57 PM, ian.peter at ianpeter.com wrote:
> 
> Hi,
> 
> Just following up on the recent discussion here on early email systems:
> 
> Does anyone know the origins of mailer Daemon notifications? I guess 
> they have been around for sometime but if anyone knows when they were 
> introduced I would be interested.
> 
> Also, were there any systems in email preceding this which notified eg 
> when a message was not sent, when a system was offline, could not be 
> delivered etc?
> 
> 
> 
> Ian Peter
> 
> 
> 
> _______
> internet-history mailing list
> internet-history at postel.org
> http://mailman.postel.org/mailman/listinfo/internet-history
> Contact list-owner at postel.org for assistance.
>