[ih] DMARC [was "Father of e-Marketing"]

[Jack Haverty]

On 6/3/19 1:23 PM, Brian E Carpenter wrote:

> Maybe we can talk here about the history of anti-spam techniques.

The mail system I wrote at MIT-DM in the mid 70s allowed each user to
create a file of instructions that the mail daemon would execute
whenever an incoming mail arrived for that user.  Those instructions
were simply a program fragment that would be loaded and run by the
daemon with the incoming mail available to it for whatever kind of
processing the user might want.

So it was easy to do things by rules something like "if FROM = <spammer>
or SUBJECT contains "SALE" then discard", and the user would never be
bothered by such a message.   One common use of the feature was to
automatically categorize incoming mail into folders based on things like
addressees, source, words in the subject or body, etc.   One of the
destination categories could of course be the recycle bin.  

Users wrote interesting fragments of code to do such stuff and shared
them with others; each fragment was typically just a line or two of code.

Users also used the feature to do things like automatically
re-distribute incoming messages matching some rule to other users, e.g.,
to other members of a particular project team.   One clever use somebody
coded was in response to the chronic shortage of PDP-10 memory and
cycles.  It was considered anti-social to run a compile job while
someone else was running one, so someone gave the compiler a mailbox and
to compile your program you'd simply mail it to the compiler which would
handle it when it could, and email you when complete - a form of "batch
processing" by email.

Licklider was very keen on this kind of "Man Computer Symbiosis" where
the computer would do things for the human on its own initiative.

Sometimes that led to trouble.....

One specific case I remember was a rule that looked to see if an
incoming mail matched the subject/from/body of any recent previously
received message.  If it found a match, the incoming message was
deleted.   This was a technique to deal with "routing loops" that could
occur if anyone created a loop in their various distribution lists - a
phenomenon related to what was later known as "cross-posting".  That
rule was quickly coded after the second or third time our disk filled up
with looping messages and I figured out what was happening.

All of this happen in the mid-70s, certainly before 1977 when I left MIT.

/Jack Haverty

(MIT-DM 1970-1977)