[ih] AOL in perspective

[Clem Cole]

below

On Fri, Sep 5, 2025 at 4:25 PM Barbara Denny via Internet-history <
internet-history at elists.isoc.org> wrote:

>
> BTW,  in searching for information on cmua,  I found this write-up on the
> coke machine. I thought people might be interested in reading it if this
> hasn't been posted before.
> https://www.cs.cmu.edu/~coke/history_long.txt


Nice write-up, but it's unfortunate that the history appears to begin in
1982 and lacks some context and history from the 1970s, when we had the CS
Terminal room off the main CS machine room (and the door closer was >>just
timed<< so could go into the hallway were the machine was, purchase at .25
a single a cold 12oz coke in a bottle, and still get back into the terminal
room before it locked and you had to walk around).

The Coke machine itself, BTW, was privately owned and was
basically operated by the same folks that ran the CS "Cheese" COOP.   Since
it was not owned or serviced by the local Coke distributor, it was
hackable, and they could set our own price. At the time, the University had
contracts for other machines around campus from a vendor, and they had
different prices (and selection). Initially, this machine was very much
under the radar, and only CS and EE people tended to know about it.    The
key was that it was an early 1960s-style machine salvaged from somewhere
where it was being discarded.  It was then resurrected (IIRC, they had to
fix the refrigeration system, but otherwise it worked well mechanically).
Note there were six buttons and six choices - all being Coke.

The bottles of Coke itself (along with Cheese) were picked by people
involved in the COOP.  They were stored with other supplies, such as
line printer paper.  At the time, one of the jobs of the operators (like
me) was to refill it.  IIRC, it was Jim Tetter who did the original
instrumentation (he was undoubtedly part of hacking it), as Jim ran the CS
hardware lab in the 1970s.  The original instrumentation was to tell the
operator that it needed to be refilled, IIRC, that was it getting to under
30% capacity.  Adding features like sensors to the different columns was
done early on, because if a column was recently reloaded, users needed to
know not to use the button for that column, or else risk getting a
yet-to-be-chilled bottle.

Also note the Coke machine was connected to the CS Front End (a dedicated
PDP-11/20 with a ton of CMU "ASLIs - async line interfaces and some DR-IIBs
to the different systems), which connected the terminals and allowed access
to CMUA/CMUB/CMUC or C.mmp.  CMUA got the info about the Coke machine via
the front-end connection [which I seem to remember was "hardwired" to it].
Later, when the distributed front-end was developed in the 1977/78
timeframe (originally on Xerox 3M Ethernet), I seem to remember that the
coke machine was run as part of one of the DFE's LSI-11s.  I left soon
thereafter, but I would suspect the LSI-11 interface was replaced by what
is described in that history as being done by John Zsarnay and Mike Kazar
in 1982 (who clearly took the coke machine interface to the next level  —
Tetter probably remembers more of this).

Until the 1980s, when the rest of the world heard about it, it was just
another hack that proved helpful to all of us.

Clem