IP: Re: Another view on ArpaNet history --

[Dave Farber <farber () cis upenn edu>]

> Date: Wed, 21 Apr 1999 20:28:35 -0400
> To: farber () cis upenn edu, ip-sub-1 () admin listbox com
> From: Steve Crocker <steve () stevecrocker com>
> Subject: Re: IP: Another view on ArpaNet history -- 
> Cc: steve () stevecrocker com
>
> Dave,
>
> Danny presents a crisp view of the three schools of thought about the
> origin of the ARPAnet.  I was a few years younger than the principals and
> wasn't around for the early discussions, but everything I saw supports
> Danny's picture.
>
> Let me add an observation.  The ARPAnet project was driven first by the
> desire to connect computers together, i.e. computer communication.  There
> had been earlier projects along this line.  Three important ones come to
> mind, and there may have been others.  One was the connection between
> Lincoln Lab's TX-2 and ARPA/IPTO, as Danny cited.  Another was a connection
> between SDC's SAGE Q-32 and Lincoln Lab's TX-2.  A third was a substantial
> but ill-fated effort at UCLA to connect three large computer centers.  I
> worked on the UCLA project in 1965-66, and we came close to setting up a
> really interesting network among three politically warring and culturally
> diverse computing communities.  All of the effort was focused on the
> services to be provided to the users.  Essentially none of the effort was
> focused on how to make efficient use of the bandwidth.  For relatively
> small networks, say two to five nodes, packet-switching may be useful as a
> means of allocating bandwidth fairly among competing user streams, but
> there really aren't any interesting routing problems.
>
> These three projects were not the only computer communication experiments.
> I visited all of the early sites on the ARPAnet.  It was not uncommon to
> hear about a project to connect various computers together within the same
> building.  For the most part, those projects did not reach completion
> because the lacunae of connecting computers from different vendors was
> almost insurmountable.  On more than one occasion, the introduction of an
> IMP at a site provided the interconnection the local computers that had
> long been sought.  Thus, a single IMP, without any lines connecting it to
> other IMPs, served as the first local area network.
>
> With this as background, it's easy to see that Bob Taylor's description of
> sitting at ARPA with multiple terminals -- see Katie Hafner and Matthew
> Lyon, "Where Wizards Stay up Late, The Origins of the Internet", pp 12-13
> and much of the rest of the first part of the book -- was a driving force
> in creating the ARPAnet project.  Irrespective of the underlying
> communication architecture, ARPA/IPTO had been poking into computer
> interconnection for several years, and it was obviously going to keep
> pushing on the problem until it got somewhere.
>
> In contrast, ARPA/IPTO had not been pursuing communication technologies,
> per se.  Topics like spread spectrum, phase modulation, etc. were known to
> the senior members of the community, but that wasn't where the money from
> IPTO was going.  Indeed, the political debates inside the IPTO community
> were about the relative importance of research on artificial intelligence
> versus computer graphics versus multiprocessor architectures.  At best,
> interconnection of computers was perceived as a means of serving the needs
> of the research community, not as a means of carrying out research on new
> communications architecture.  Communication was almost completely absent
> from the research agenda, with Abramson and Kuo's seminal work at Hawaii on
> the Aloha system a notable exception.
>
> Thus, I think some sort of "ARPAnet" was inevitable irrespective of the
> underlying communication architecture.  Fortunately, the packet-switching
> ideas Kleinrock and others developed were nicely available at the right
> time and the right place, and the ARPAnet set the direction for
> communication architecture as well as opening the door for broad scale
> computer interactions.
>
> Baran's work on distributed communication was motivated by the need to
> preserve communication in the military command and control system.  As
> Baran found to his frustration, there wasn't any direct way to move his
> work into the mainstream of military systems development.  However, once
> the concept of the ARPAnet was committed to and funded, there was
> investigation into collateral ideas.  Baran's work and the work at the NPL
> became known more broadly.  As the ARPAnet blossomed, communication
> research became important and vied for research funds.  A third budget
> element was added to IPTO to cover funding of post-ARPAnet communication
> research projects like packet-switching in the radio and satellite
> environments.  Local area network projects sprang up everywhere, with
> Metcalfe's Ethernet winning the day.
>
> Steve
>
> ----------------------------------
> Steve Crocker Associates, LLC                   Bus:   +1 301 654 4569
> 5110 Edgemoor Lane                              Fax:   +1 202 478 0458
> Bethesda, MD 20814                              steve () stevecrocker com