[2643] in Commercialization & Privatization of the Internet
An Editorial from Issue one of the COOK Report
daemon@ATHENA.MIT.EDU (Gordon Cook)
Wed Apr 1 15:11:33 1992
To: com-priv@psi.com
Date: 30 Mar 92 23:45:27 EST (Mon)
From: cook@tmn.com (Gordon Cook)
<<MESSAGE from>> Gordon Cook 30-MAR-92 23:45
cook@tmn
In the morning I am taking the first 12 page issue of the COOK Report on
Internet - NREN to the printer. Thursday I leave for nearly 2 weeks in
Russia. I'll be offline til April 16th.
I will distribute copies of issue on on the Hearings starting now to those
who send private mail requesting it and assure me that they will not
republish it on other lists. This is the only issue for which I will do
this.
In the meantime here is a six kilobyte editorial from Issue One.
I think it contains the seeds of a solution to the current mess. I'd like
to know what the readers of Com-Priv think.
______________________ An Editorial
Is the NSFnet Precompetitive or Commercial Technology?
A Win-win Solution
The Hearings showed the sharp contrast in the NREN policy debate. Mitch
Kapor and Bill Schrader were adamant that the network represented a
commercial technology no longer in need of government funding to develop
further. Mike Roberts, who testified for Educom, and Doug van Houweling,
who appeared on behalf of MERIT, were just as adamant in their belief that
the network should be viewed as precompetitive technology and a platform
for development of the high bandwidth new tools of multi media electronic
mail and databases that have been spotlighted as among the goals of the
NREN vision. In such a set of circumstances government funding for the
backbone would be expected as a matter of technology development and the
goal of economic competitiveness.
In the opposing view of Schrader and Kapor, the maturity of current
network technology is such that it should be brought, by commercial
providers, to large numbers of users as an enabling technology or a tool
to be used for improved research and education. The result is a win-lose
debate. On the one hand we have a new technology for a small elite now and
new markets later. While, on the other hand, we also possess a
commercially mature technology used by several million Americans for
improving their daily tasks of research and education. Both uses are valid
ones - yet the fact that they were presented as being in such stark
conflict at the hearings reveals the abysmal state of planning for the
implementation of an NREN.
Like it or not, the NSFnet has been commercialized and T-1, if not T-3 IP
internetworking, is not a precompetitive product. However those who want
the federal money to develop a fancier higher tech "bleeding edge" NREN,
say that, if they allow the NSFnet backbone to be treated as a commercial
production entity, they will be left without a platform to move the
technology forward. The central question seems to be that there is only
one backbone that reaches all 32 midlevel networks and that this backbone
is desired by different parties to be both experimental and of production
commercial quality. An impossible sounding task.
The debate between Roberts and van Houweling, on the one hand, and
Schrader and Kapor on the other helped to clarify what the NSF must do in
its solicitation. Choosing two different providers for the same kind of
backbone it now has makes no sense. Rather it must choose to support two
very different backbones.
Schrader and Kapor are correct to assert that, as long as the backbone is
a free good for the mid-levels and their 1200 connected institutions, the
Federal use of taxpayer dollars is holding back the further development of
a commercial industry where companies like Uunet and PSI can offer
production quality commercial networks. To have one partnership
prepositioned for exclusive commercial use of the production backbone
because it engaged in "generous" cost sharing with the NSF in agreeing to
install marginally functional T-3 technology is unacceptable.
It is time for the NSF to acknowledge reality and embark on the 18 to 24
month time line mentioned by Eric Hood on behalf of FARnet so that the
mid-levels may choose some combination of either interconnecting with each
other or selecting commercial backbone services. Someone must be clear
and state that a subsidized backbone ends on January 1, 1994. The NSF
must focus a major share of its time meeting with FARnet and others to
accomplish a coordinated change. Commercial, production backbones with
subsidies to regionals for interconnection should emerge as one kind of
awardee in the recompetition.
However, the other awardee should be a precompetitive research backbone
connecting the major sites involved in the current gigabit testbeds.
Major research universities not in the testbeds could apply to the NSF for
grants for direct connection. Don't put the entire network on the
bleeding edge. Let universities and corporations compete for that honor.
If this backbone can not be brought online before 1995, it would not be a
huge loss. (Indeed just what could be brought online by 1995 seems a bit
uncertain.) However if the gigabit testbeds perished during a never
ending fight between ANS, the mid- levels, and the commercial providers we
would be wasting an important resource.
Congress should instruct the NSF to develop a plan and a budget to work
with CNRI to maintain the testbed momentum and turn it into a small scale
national research backbone by 1995. There is where any federal money
spent for backbones and new technology development should be focused. The
current situation is win loose. Either a commercial network or a
"precompetitive" R&D network is imposed on the entire community.
Supporters of each side spend most of their time struggling to see that
their views predominate.
It is time to stop the warfare and go for a win-win solution. If the NSF
really wants to go for a leadership role in developing the most broadly
defined vision of the NREN possible, it will see the merits of following
both these directions. If it takes more money than would be required by
continuing the special relationship with ANS, so be it. Let the NSF put
ANS on notice: a mistake was made. It will not be continued after
January 1, 1994.
When private industry offers the government a special "money saving" deal,
if it is given in return a unique benefit, no one is well served. If
ANS cannot compete in the market place by 1994, it can certainly not
complain that it did not have a special opportunity that PSI and Uunet did
not have. However the NSF must also make it very clear that it no longer
sanctions attempts by ANS to acquire commercial customers and charge any
mid-level "com-bits" as a result. If ANS is to be paid for provision of
backbone service until 1994, it must accept a quid pro- quo: an immediate
end to its special status.
NREN does have two policy drivers. It is time to implement a separate
plan for supporting each rather than a single plan that leaves no one
satisfied.
