[ih] A paper

[vinton cerf]

TCP/IP benefitted from the NCP work on ARPANET and the application
protocols that were developed such as TELNET, FTP and eventually SMTP among
others. The first paper on TCP was published in May 1974 and detailed TCP
specifications were the product of a team working at Stanford with visitors
from Norway, France, Japan and Xerox PARC and were released as RFC 675 in
December 1974. Implementation started in 1975 at Stanford, BBN, University
College London and SRI International. In the Fall of 1976, I joined Bob
Kahn at ARPA and implementation and refinement of the protocols continued.
During these formative years leading up to the TCP/IP flag day on January
1, 1983, I do not recall anything but pragmatics driving the development.
We were all focused on getting internetworking to work with different
packet networking technologies and different computer types and operating
systems. The original Network Working Group philosophy permeated this
effort as did the subsequent International Network Working Group which
later became IFIP 6.1. The protocol architecture was deliberately kept open
to new ideas and participation expanded from those under research contract
to DARPA to a broader community as the capability became more widely
available through research and education networks, commercial sources, and
open implementations for operating systems from IBM, Digital Equipment
Corporation, Hewlett-Packard, Berkeley (ATT/UNIX), BBN's TENEX, 3COM, SUN
Microsystems and others.

Attempts to retrofit political motivations onto what was a pragmatic
engineering effort strike me as strained and artificial.  Speaking for
myself, in the late 1980s, I felt a personal motivation to see the
technology commercialized and welcomed INTEROP and the eventual permission
to interconnect the commercial MCI Mail system (along with others) to the
Internet because I hoped an economic engine could be ignited to support
access to the Internet by the general public and private sector. I was
enthusiastic about the founding of the Internet Society, partly as a legal
home for the IETF/IRTF/IAB and also as a means to promote a philosophy
(policy) of open networking. Perhaps more political elements can be
discerned especially with the arrival of the World Wide Web and the
numerous applications that capability spawned.


On Sun, Jul 18, 2021 at 9:48 AM Andrew G. Malis via Internet-history <
internet-history at elists.isoc.org> wrote:

> I WAS there in the 80s, at BBN, where I worked first on the ARPAnet writing
> IMP code, and then later I managed, at DARPA's direction, the ARPAnet
> transition from NCP to TCP/IP. I also wrote reports for the DoD to help
> them plan their "eventual" transition from TCP/IP to ISO/GOSIP/CLNP/TP4,
> which as we all know never actually happened. I was also writing RFCs
> before the IETF was even established. So I've got some amount of personal
> knowledge here. :-)
>
> IMHO, NONE of the decisions made at the time were "political" in nature.
> TCP/IP won over OSI because it was designed by a group of people (largely
> grad students at the time) that were interested in creating something that
> worked, and once they had a protocol design, they wrote the code for
> prototypes and tested it out both in locally in their labs and using the
> ARPAnet as a testbed for making it work over a WAN, which they had access
> to as they were students. They saw what worked and what didn't, and then
> refined the specs and implementations to match. All of this work happening
> in the open arena resulted in the relatively rapid development of freely
> accessible specifications and implementations that anyone could obtain and
> play with. As has been said elsewhere, that resulted in it being taught to
> other students, and them playing with it in their labs, resulting in a
> broad knowledge base entering industry. Meanwhile, the free code was
> adopted by and incorporated into the major OSes of the day, especially the
> various flavors of UNIX, or was available as add-on implementations, either
> free or commercial (such as FTP Software's stack for DOS and early
> Windows). And much of this happened even before the IETF existed!
>
> Meanwhile, OSI kept plodding along, but you had to pay for the
> specifications in order to just read them., and at the time whatever
> implementations that existed weren't free, and couldn't hold a candle to
> what was freely available for TCP/IP.
>
> The folks working on OSI were very aware of TCP/IP and had a lot of good
> ideas for improvements, and could have won if their specifications and
> implementations had also been freely available, and if their timing had
> been better (earlier rather than later).
>
> Cheers,
> Andy
>
>
>
> On Sun, Jul 18, 2021 at 2:55 AM Patrik Fältström via Internet-history <
> internet-history at elists.isoc.org> wrote:
>
> > On 18 Jul 2021, at 1:08, Bob Purvy via Internet-history wrote:
> >
> > > Are there some decisions you think were *not* made for technical
> > reasons? Particularly in the 80s.
> >
> > I was not present in standardisation in the 80s. I started around 1990.
> > Over time, I saw, specifically when later being area director and in IAB,
> > arguments be other parameters visible in market economy. From my
> > perspective, calendaring was the first real situation when technical
> > arguments did not have any role at all in the IETF. It was just market
> > economy (and the ability to expand the market the sponsors of the
> > technologies had).
> >
> >    Patrik
> > --
> > Internet-history mailing list
> > Internet-history at elists.isoc.org
> > https://elists.isoc.org/mailman/listinfo/internet-history
> >
> --
> Internet-history mailing list
> Internet-history at elists.isoc.org
> https://elists.isoc.org/mailman/listinfo/internet-history
>