[ih] History of IoT

[Michael Thomas]

On 2/13/23 12:53 PM, Clem Cole wrote:
>
>
> On Mon, Feb 13, 2023 at 3:01 PM Michael Thomas via Internet-history 
> <internet-history at elists.isoc.org> wrote:
>
>     Hi all,
>
>     I've been trying to understand the history of what we'd now call the
>     Internet of Things. I know about the internet coke machine which
>     was in
>     about 1983 and then the internet toaster in around 1990. 
>
> FWIW - The CMU Coke machine was in the mid 1970s and was the Arpanet 
> before the Internet - so you are missing at least 8-10 years. It was 
> definitely running in '76, but I think was earlier than that - Guy 
> Alms who was there a few years before me, probably remembers when Jim 
> Teter did the original hack (in PDP-11 assembler and BLISS BTW]..
What I've seen is that it was 83, but maybe they are just talking 
post-flag day.
>
> Well the term 'toaster' actually came from the C89 standard work. It 
> was the crude term we used to discuss a cheap embedded device [if you 
> look at the standard C actually has two ways of being deployed - 
> hosted or embedded - but that's not the discussion here].  I also 
> suspect many of those people don't know about/had little first hand 
> experience with BLISS/ESPOL/BCPL etc - which all predated B and C as 
> they were 1960s/1970s technologies - plus few places other than 
> CMU/MIT/Standford would have had 'free' PDP-11 cycles for a hack like 
> that give the cost of the systems at those times (IIRC - the Coke 
> machine ran off the original CS Front-End)
>
> I think IoT community picked up the term Toaster from the C89 work and 
> frankly, many folks were ignorant of the work that predated it when 
> using $10-$30K Minicomputers, much less the $.5K-$1K microcomputers of 
> the 1970s.  Also remember the per host for an IMP connection in the 
> 1970s was huge - in the order of $100K-$150K / yr for each host [paid 
> for as part of your Arpa contract].
 From my scrounging around, it seems that the toaster debuted at the 
1990 Interop to big fanfare and called the first internet of things 
example, even though that phrase didn't come until later and seemingly 
ignores the coke machine guys which I really don't understand. But all 
of these definitions are rather fuzzy. LSI/11's were used extensively 
for controlling and monitoring stuff, but were they "embedded"? They ran 
standard off the shelf DEC OS's (and Unix, I think), so that's not 
exactly what I think of as embedded.
>
>
>     My personal stake -- and the reason for my curiosity -- is that I
>     designed the software for an ethernet enabled laser printer in the
>     mid
>     80's which is very likely to be the first (I'd be happy to hear
>     otherwise) for a printer. 
>
> Hmmmm - the printers at CMU/Stanford/MIT et al were networked in the 
> late 1970s. Also remember of course the network at PARC were also.
> In fact, my own introduction to TCP was when we were working on the 
> CMU distributed Front-End. It had been originally moved from the 
> single 11/20 [which ran the Coke machine] to a 'network of LSI-11 on a 
> 3M Xerox ethernet.  I was part of the group trying to move the system 
> to bunch of Intel 8086 cards in a multibus.  TCP was still yet to be 
> fully defined. Phil Karn (my Lab partner at the time) had started to 
> play with writing a TCP for his Z80 system - that would eventually 
> become his famous release.  [ We had the RFCs from that work].
But were they networked with the network stack on the device itself? I 
think that's the distinction. A networked printer could be moved to 
another room without having drag an LSI/11 along with it.

Mike