obrien@Aero.org (Michael O'Brien) (04/09/91)
This thread about fingering Coke machines (I was previously only aware of the installation at SAIL) has me fascinated. Hence, I'll proceed to collect entries in the form of answers to the following question: What other oddball devices have been hooked to the net? Ideally, entries should not be scientific instrumentation (unless it's a real Dr. Wacko production), but more common things - such as the reputed front door buzzer at the Tech Square building at MIT, where one could supposedly press control-shift-meta-cokebottle-P (this was on an ITS machine, after all) to open the front door, no matter what application you were in. I'll summarize the best responses. Warning: I may also publish entries in my column "Ask Mr. Protocol" in Sun Expert, unless requested not to by the author. Credit will be given, of course. -- Mike O'Brien obrien@aerospace.aero.org
MAP@LCS.MIT.EDU (Michael A. Patton) (04/13/91)
Here is my contribution for the unusual devices discussion. This is
not all strictly network related, but the connected history is itself
interesting. These are my personal recollections of the last 18 years
at Tech Square, home of the MIT AI Lab and Project MAC (now LCS).
When I first arrived at MIT (1973), the AI Lab's Knight TV (KTV) front
end was already hooked up to control several unusual devices in the
building. The KTVs were special video displays connected through a
PDP11 operating as a front end for the KA10 (at that time MIT-AI,
later AI.AI.MIT.Edu, now scrap metal). The keyboards had several
extra keys for special uses (most were available for special use in
programs). One of these was labeled ESCAPE (there was also an ALTMODE
that generated ASCII code 27 octal) and reserved for interacting with
the front end. There were lots of mundane functions connected with
the key (e.g. ESC-F for "finger" or "free" showed the current list of
users), but there were two that were special. ESC-D operated the
electric lock on the door to the computer room (the entire ninth floor
of the building, which also included a number of hackers offices, the
rest were one floor down). ESC-E summoned an elevator, it knew
whether your display was on the 8th or 9th floor and pushed the
appropriate button. The time it took for an elevator to arrive from
the first floor in the dead of night was about the same as the time to
walk out to the lobby. Initially these were only available from the
special consoles and worked directly off the front-end system, the
host had no real way of affecting this.
Then, in the late 70's, the AI Lab was developing a new style of
operation that used special purpose individual workstations (nobody
had invented the name workstation yet, as I recall, but that's what
they were) that were the predecessors of the Lisp Machines later
manufactured by Symbolics, LMI, TI, and others. These workstations
needed some communications media that was faster than what we commonly
used then and the CHAOSnet was developed. Based loosely on the Xerox
3Mbps Ethernet developed a little earlier, it ran at the blinding
speed of 10Mbps! CHAOSnet was CSMA/CA (that's Collision Avoidance, as
opposed to Ethernet which only does CD, Collision Detection). The
routers used in this net ran on LSI11 systems and the door controls
and elevator controls were moved to one of these boxes so that the
workstations could get the same functionality as the old displays
(they initially used the same keyboards and later an expanded version
referred to as the "Space Cadet" keyboard). As the CHAOSnet expanded
around campus this meant that anyone could open the door or summon an
elevator, even outside the building and these were eventually
disconnected.
Another addition around this time was the Weather Computer and the
Weather Reporting Service. The Weather Computer was an LSI11
(seperate from the routers, but only connected to the network) with a
parallel interface to a HeathKit (I think) weather station. By using
a datagram protocol over the CHAOSnet any system could interrogate the
Weather Computer for the current info about inside and outside
temperature, pressure, and wind speed (actually, I think the "inside
wind speed" was hard coded as zero, we didn't buy two of these). The
Weather Computer didn't have any state, it just engaged in a datagram
exchange where it read the current values and returned them. Then a
daemon was written to run on one of the Labs timesharing systems which
interrogated the Weather Computer frequently and assembled longer term
information. If you fingered "weather" on this machine you got back a
summary report of current and recent past readings, I believe daily
(today and yesterday) and weekly high/low/average for each value.
__
/| /| /| \ Michael A. Patton, Network Manager
/ | / | /_|__/ Laboratory for Computer Science
/ |/ |/ |atton Massachusetts Institute of Technology
Disclaimer: The opinions expressed above are a figment of the phosphor
on your screen and do not represent the views of MIT, LCS, or MAP. :-)