[ih] OSI and alternate reality: settlements

[John Gilmore]

Greg Skinner via Internet-history <internet-history at elists.isoc.org> wrote:
> imho, what made the Internet a resounding success was the "flat rate
> and sender keeps all cost nature" of connection -- i.e. the lack of
> "settlements" for traffic transited between interconnected networks as
> is the case with the PSTN and was the case with the interconnected X
> networks...  there was A Very Pivotal Moment in Internet history
> where... that almost didn't happen ...

This was something that Mitch Kapor was concerned about too.

Shortly after forming EFF, Mitch joined the Commercial Internet Exchange
and became its Chairman.  This was largely to understand the issues
among the tiny new commercial ISPs, around Acceptable Use Policies and
settlements, and to help the nascent ISP industry make some good
choices.  Here's a note that he sent to com-priv in April 1992, noting
that the CIX had adopted a "no settlements" policy for the interim, and
asking to learn more about how settlements had worked in other networks.
Dave Farber forwarded this inquiry to his Interesting-People list on the
same day.

At the time, CIX had 7 small members.  In the following months, Mitch
also helped to cross-connect CIX with ANS (which ran NSFnet), allowing
the NSFnet regionals to use ANS as an access network to reach the
customers of CIX members (like me on Alternet).  See the excerpt from
his June 1992 INET speech, appended below.

	John

Date: Sun, 5 Apr 92 11:27:08 EDT
From: farber at central.cis.upenn.edu (David J. Farber)
Message-Id: <9204051527.AA01784 at pcpond.cis.upenn.edu>
To: interesting_people at dsl.cis.upenn.edu
Subject: Request for Assistance

Begin forwarded message:

Received-Date: Sun, 5 Apr 92 10:07:50 EDT
Posted-Date: Sun, 5 Apr 1992 09:55:48 -0400
Date: Sun, 5 Apr 1992 09:55:48 -0400
To: com-priv at psi.com
From: mkapor at eff.org (Mitch Kapor)
Subject: Request for Assistance

Request for assistance:

I am interested in developing a better understanding of the economics of
interchange ("settlements") between networking carriers on the Internet. 

There has been much casual, sometimes heated, conversation on the subject
but I have not seen anything yet which persuades me that we even have the
basic Internet settlements problem well-framed, much less that any workable
model is at hand.   Further, given the uneven movement from subsidized to
self-supporting networks, it becomes even harder to imagine which set of
arrangements approaches economic efficiency. (This is a principal reason
why the CIX has adopted a "no settlements" policy for now.)

Nonetheless,  it is becoming apparent (to me, at least) that the entire
community needs to have a principled way to make sure, in a world of
interconnected carriers, that no party unfairly bears costs.  As an initial
step, I would greatly appreciate pointers to experts, organizations, and
printed matter on the logic and practice of settlements in telephone
networks, electronic funds transfer systems, EDI, or other in
infrastructures, as well as generally relevant material from economics.  

Material of a tutorial or general background nature is preferred.   

 


Thanks.


Mitch Kapor mkapor at eff.org
Electronic Frontier Foundation
Chairman, Commercial Internet Exchange
(affiliations for purposes of identification only)

 --------

(Here's a relevant excerpt from Mitch's later June 1992 speech at INET 92:)

I also serve as Chairman of the Commercial Internet Exchange, or CIX, a
not for profit trade association of commercial Internet providers in the
U.S.  and Europe.  CIX has 7 members, all interconnected, all exchanging
traffic with one another without any Acceptable Use Policy.  I was
partly responsible for recent ANS-CIX interconnection by which
U.S. mid-level, regional networks can use ANS as transit network to
connect with CIX.  ANS operates the NSFNET national backbone in the U.S.
With this important first step accomplished, we are now working toward
broadening CIX membership and getting the commercial part of ANS to
join.