[ih] Internet-history Digest, Vol 69, Issue 11

[Jack Haverty]

After 50 years it's hard to remember...

Lick's group at MIT was, in retrospect, pretty unique.  I never knew 
anything about contracts per se.  But our group's name changed several 
times over the 8 years I was there.  We were Dynamic Modeling.  Then 
Dynamic Modeling and Computer Graphics.  Later we were "Automatic 
Programming", and "Office Automation" at one point.

The Office Automation phase was probably when we were involved in MME.  
My suspicion is that the group name changed to reflect the current 
priorities in ARPA.   But throughout that period we were always working 
on some aspect of Lick's vision of lots of computers communicating with 
each other in support of whatever humans were doing.  If there was any 
kind of separate contract for MME work, we grunts writing code didn't 
know about it.

Lick's vision involved computers interacting with each other, which 
motivated our interest in computer-friendly protocols for use between 
machines on the ARPANET.  For example, RFC713 was a first step at 
defining computer-computer interactions.   We never got very far in 
generating "rough consensus" though, so the multiple-machine 
configuration wasn't possible for the experiment.  IIRC, the actual MME 
experiment was basically comparing apps running on a single machine, 
somehow connected into Autodin.

Our vision for Message-IDs was to capture a message at its inception 
(yes, it was vague exactly when that was).  That message could then be 
available on the 'net to anyone with authorization to see it (all to be 
worked out of course).  Messages could be stored in what we would today 
call "the cloud", but at that time it was the Datacomputer.   Its entire 
terabit of memory should clearly be sufficient for the entire network!

The mail server I implemented had the ability to store and retrieve 
emails from the Datacomputer.  In addition to being an archive, such 
"cloud" storage might also serve as a trusted third-party, who could 
confirm the content, authorship, security, and delivery of an email.   
That was important for replicating functions commonly used in both the 
business and military worlds, such as escrow agents, verified delivery, 
approval workflow, etc.

Lots of metadata could be attached to Message-IDs, also kept in the 
cloud rather than all being contained in the header of each message as 
it travelled around.   Databases weren't on our radar in 1975, but today 
I'd put all the metadata into database tables and use all the same data 
analysis tools that businesses use to track inventory, sales, etc.   
It's all just data.

One attractive feature (to me at least) was that a sequence of messages 
could be related to each other by using IDs, and users' email apps could 
use that structure to provide a friendly user interface.  Today's 
approach, with a confusing mix of headers, questionable content, and 
unclear authorship especially in long forum conversations (like this 
one) could be much better presented to the users.

Anyway, that's the kind of system we were envisioning, but it depended 
on getting some kind of computer-computer interaction protocols and 
message structure in place between the cooperating computers over 
whatever network they were using.  MTP (message Transmission Protocol) 
was postponed while SMTP (Simple MTP) was pursued as an interim step.  
That was now 50 years ago.

Jack


On 8/16/25 19:19, Dave Crocker wrote:
> On 8/16/2025 6:59 PM, Jack Haverty via Internet-history wrote:
>> I was in Lick's group at MIT during that time, and we were 
>> implementing electronic mail, so we got involved in the MME. 
>
> My understanding of the MME effort:
>
>  1. ISI had a contract to produce a system to be used in Hawaii for
>     the experiment
>  2. Apparently ISI didn't make enough progress to please the funding folk.
>  3. A competition developed -- with funding for each?  as a
>     competitive bid? -- between ISI, BBN (Hermes), and your MIT effort.
>  4. ISI won the competition and was fielded.
>
>
>> "Message-ID:" field
>
> I don't recall the decision to have this field being controversial.  
> What it exactly referred to was a different matter.
>
> Given the handling realities of email, there are quite a few potential 
> applications for a message ID.  Author creation, vs mail handling 
> system origination, vs. each transit hop, for example.
>
> I seem to recall that, much later, the meaning of the Date: field was 
> similarly not consistently interpreted and there was a desire to 
> resolve this.  I also seem to recall going around and asking various 
> folk -- I don't remember my sampling methodology -- what moment they 
> thought it referred to.  The very strong consensus was posting time.
>
> I even vaguely recall that X.400 had multiple message IDs, which 
> suited their 'toss everything in' philosophy.
>
> d/
>
> -- 
> Dave Crocker
>
> Brandenburg InternetWorking
> bbiw.net
> bluesky: @dcrocker.bsky.social
> mast: @dcrocker at mastodon.social

-------------- next part --------------
A non-text attachment was scrubbed...
Name: OpenPGP_signature.asc
Type: application/pgp-signature
Size: 665 bytes
Desc: OpenPGP digital signature
URL: <http://elists.isoc.org/pipermail/internet-history/attachments/20250817/17b2e496/attachment-0001.asc>