[1391] in Commercialization & Privatization of the Internet
Re: CSnet
daemon@ATHENA.MIT.EDU (Dave Crocker)
Sat Sep 21 14:30:19 1991
To: Stephen Wolff <steve@ncri.cise.nsf.gov>
Cc: Robert Ullmann <ARIEL@relay.prime.com>, ietf@isi.edu, com-priv@psi.com,
In-Reply-To: Your message of Fri, 20 Sep 91 14:13:25 -0400.
Date: Sat, 21 Sep 91 11:28:17 PDT
From: Dave Crocker <David_Crocker@Pa.dec.com>
Steve,
The original CSNet proposal came to NSF from, I believe, a consortium of
4 schools. None were on the Arpanet but had the understandable desire
to join it. Their model was to provide useful, interactive service
for the second-level schools and below. (First level, by definition, meant
you were already on the net. Hence, "level" has nothing to do with
quality or size.) My recollection was that Larry Landweber, at UMich,
definitely was a/the prime instigator. A southwestern school (Arizona?)
wa s amember of the proposal and I'm afraid I can't remember whether Prudue
was or even guess at the 4th member.
With luck, Dave Farber can fill in the detail about the other personalities
that were involved.
NSF sent the proposal around for review and received consistent feedback
that this was A GOOD IDEA. To improve its chances at success, a new
proposal was formulated, to include some folks that already had networking
experience. Hence, Purdue and University of Michigan were the two members
not already on the net. University of Delaware and The Rand Corp were
the members that did have experience.
UDel's involvement was something of a sucker punch. I already had core
software running (and in use by some groups in the Army) which would
permit dial-up email access and that was deemed by NSF, et al, to be
quite useful. So we were able to add the first CSnet site (probably
UMich and Purdue) essentially within days of having the project
authorized.
That is not to say that my software was "complete." The experience was,
in fact, my first exposure to the challenges of scaling a system up and
having it widely used by non-expert folks. (expertise has nothing to
do with intelligence; it relates to direct experience and none of
CSnets customer sites had networking experience.) Can't remember the
early numbers, but I think I added one site per month for about
six months and then doubled the rate. In those days, Mike O'Brien,
at Rand, and I were the entire support staff, doing both addition of
sites and enhancements to the software. Unfortunately, the software
need quite a bit of work and, therefore, the sites needed quite a bit of
help.
Permission to "extend" the Arpanet net was, indeed, a major advancement and
Kahn and Cerf, at DARPA, deserve considerable credit for allowing it
formally. (It should be noted that Berkeley had just begun to do
some back-door relaying, via uucp, but it was a source of discomfort.)
DARPA had predictable concerns about technical and behavioral safety
of allowing other groups to have access and the CSNet project had to
make a committment to pursue good rules of conduct among its members.
For most sites that joined, I continued to be amazed at the degree of
enthusiasm they showed to the opening of the new pathway to their peers.
Dave
