[ih] Internet-history Digest, Vol 77, Issue 55

[Jack Haverty]

Somtime in the 1980s, I had an experience similar to John's anecdote at 
Bell Labs.  It made me start wondering about how "office politics" 
affected technology development.

I never did much work in the OSI world, but I remember that somehow I 
ended up in a meeting, probably somewhere in Europe, where there was an 
intense technical debate about some protocol or algorithm involved in 
OSI.  I can't remember any details.  There were several camps, each 
arguing for their approach and highlighting problems in the other 
approaches.   But I could see real progress and it seemed that a 
Consensus was imminent.

Then the Chair of the meeting interrupted the discussion and said that 
it was now 2PM and therefore time to move to the next agenda item on the 
schedule.  So the impending Consensus disappeared, with the discussion 
to be scheduled on the agenda for the next meeting, several months away.

That experience never happened in any meeting of Internet projects that 
I remember attending.  Discussions, debates, and arguments would 
continue, sometimes long into the night, until either a Consensus was 
reached by the people still remaining, or proponents of some approach 
would go off and write the code to implement their vision.   In either 
case, progress happened and ideas proceeded to live testing.

Perhaps "office politics" and associated issues of bureaucracy played a 
role in TCP's defeat of OSI and all the proprietary networking 
technologies.   The Internet project may have been "less screwed up" 
than everyone else.

/Jack Haverty



On 5/5/26 09:22, John Shoch via Internet-history wrote:
> Regarding packet switching work at AT&T vs AT&T Bell Labs, Sandy Fraser,
> Spider net, et. al -- some background, and an anecdote:
>
> Background
>
> It's probably true that the parts of AT&T focused on telephony and
> long-distance data had little appreciation for the potential of packet
> switching and the internet.
> But Bell Labs had a long history of work on alternative local data
> communications at the edge of the network, especially via local rings and
> loops for data communication:
> "Newhall Loop" by Newhall, Farmer, and others;  token passing ring.  Papers
> starting in 1969.
> "Pierce loop" by Pierce, empty slot technique.  Papers starting 1971.
> "Spider Network" ring with control via a central switch.  Papers starting
> 1974.
>
> Tony actually wrote a nice note with some of the background:
>     https://dl.acm.org/doi/epdf/10.1145/1015676.1015677
> Many more references are listed here:
>
> https://www.bitsavers.org/pdf/xerox/parc/techReports/SSL-79-5_An_Annotated_Bibliography_on_Local_Computer_Networks_Oct79.pdf
>
> In 1979 I wrote an assessment of the Spider network:
> "One of the major examples of a loop is the Spider network built at Bell
> Laboratories [Hayes,
> 1973, 1974; Fraser, 1974a, 1974b, 1974c, 1974d, 1975]. The system used a
> unidirectional ring
> utilizing T1 technology (1.544 Mbps), controlled from a minicomputer
> serving as a "central
> switch." Frames contain a total o f 386 bits, including 336 usable bits
> plus an additional 50 bits
> required for T1 framing and clock recovery. Connections to the loop at a
> host are made
> through a Terminal Interface Unit (TIU) equipped with a microprocessor
> known as the "Fly." "
>
> I was pretty critical of the design, esp. compared to the Ethernet:
>   separate ring control box required at each station, small block sizes,
> central control via a complex ring controller, virtual circuits needed to
> be established through the controller to another host, reliability issues,
> etc. etc.    And it was not conceived as part of a broader internet
> environment.
>
> But I also had to acknowledge:
> "Spider has been quite distinguished, however, in providing service to a
> real set of users. By
> 1975 it had been used to interconnect numerous machines in the Bell Labs
> environment,
> including single user minicomputers, laboratory instrumentation machines,
> and time-sharing
> facilities. Services available through the network have included access to
> a shared network file
> system and a printing server."
>
> As Len K. said, "Kudos to Sandy."
>
> An anecdote
>
> Several years later I was working at Xerox corporate HQ.  Our CEO had
> crossed paths with the AT&T CEO and they came up with the suggestion, "Why
> don't we have some people meet and see if we can work on something
> together."  So I was dispatched to Bell Labs in NJ for a series of
> presentations on various projects underway within the lab.  There was
> continuing work on alternatives for local data communication -- but
> different groups came through with different projects, and they were often
> critical of the other projects within the Labs!
> I came back to our CEO and reported, "It was so nice to see another large
> corporation that is *more* screwed up than we are!"
>
> John Shoch
>
>
>> Message: 1
>> Date: Tue, 28 Apr 2026 09:16:57 -0500
>> From: Andrew Odlyzko <odlyzko at umn.edu>
>> To: Internet-history <internet-history at elists.isoc.org>
>> Subject: [ih] a few comments about Tier 1 networks, Frame Relay, Sandy
>>          Fraser, etc.
>> Message-ID:
>>          <CAJOs04J1=_
>> Pg8d8AKirBR4vgyTmYDFgM6oECwzHdv3hL_A__-A at mail.gmail.com>
>> Content-Type: text/plain; charset="UTF-8"
>>
>> Jumping into this interesting discussion with a few minor comments:
>>
>> Vint asked about Sandy Fraser's data networks.  My recollection
>> is that Spidernet (or possibly just Spider) was created by Sandy,
>> but it was a local area network, and that it preceded Datakit, which
>> was for long distance links.  I don't know whether Spidernet was
>> ever commercialized, but it was used inside Bell Labs Research.
>> (I joined Bell Labs in 1975, and was located physically and
>> organizationally close to Sandy's group, but was not familiar with
>> their work, just heard some comments here and there in those early days.)
>>

-------------- 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/20260505/233680a7/attachment.asc>