Dr. Wolff's introductory note:

Gordon,

I have put a lot of vacation time into this commentary and gotten just under halfway through. . . . .

I think you have a number of factual errors, and I've tried to point them out. But I expect you've got them mostly right. However as both mathematicians and historians know, through any (even infinite!) subset ofthe (unknowable) set of all facts, infinitely many distinct histories can be threaded. You have selected your facts and woven about them an historical mesh that supports your point of view and there's nothing I can do about that.

No doubt there'll be more public debate after release.

Thanks for the chance to comment.

-s

a

But the history of the Internet in this country has been one of exclusive gov't awards for network infrastructure that the private sector was unwilling to provide any other way. The gov't (or at least NSF) is not in the business of providing exclusive contracts for network services that can in fact be purchased by individual campuses from whomever they want to buy them from. The reason for the exclusive awards, such as the one to Merit, is to push the technology in ways that are deemed in the national interest. And in all cases, the awards aremade competitively.

b

The reason for the 18-month extension was not a policy impasse, but a result of the rapidly-changing Internet environment (including the additional complexity resulting from overlaying the HPCC/NREN on existing programs). The extension was fully justified inthese terms in our request to the National Science Board in November 1991, and it was approved by them. It was a deliberate policy decision, not an impasse

c

Conditions were imposed to prevent ANS' commercial traffic from degrading service to the R&E community. There has not been any such degradation, since ANS has on its own initiative installed adequate additional network infrastructure to prevent it. Therefore the conditions have not required "enforcing"; I'm not aware of any "difficulty".

d

...as well as a thousand or more secondary schools, hundreds of public and private libraries, and an unknown number of hundreds of commercial organisations (recall that when ANS conected to the CIX, it had to announce only seventy new networks; i.e., all but seventy of the commercial organisations now on the Internet were already accessible via the existing structure.)

Editor's comment: Dr. Wolff speaks from the point of view of his structure. From the point of view of the CIX backbone all customers were accessible. And in fact, as a knowledgable source reported to the author on 12/ 29/92, anyone connected to the network can send anywhere because ANS has for some time not been blocking any CIX commercial traffic.

e

We don't want to control anybody's backbone. We do control the NSFNET Backbone Service.

f

The primary goal of the program is service to the R&E community; the division of expense between backbone services and mid-level support is of at most secondary consideration.

g

If the fulfillment of that dream is in danger, it is not because "the NSF does not have unlimited funds at its disposal". Few people I know dispute the notion that extending access to all the nation's primary and secondary schools will require a cooperative program with the states and the Dep't of Education being major players, and other government agencies such as NSF playing secondary roles.

h

The Internet is already well on its way to being worldwide, and already has somemultinationals connected. I dispute that this outcome is problematic; I think it's assured.

i

Connectivity is certain because of user demand, but absent jackboot state control it cannot be "assured" (and remember what I pointed out above: connecting the CIX only brought 70 new nets aboard - out of the (now) greater than 8500 announced - less than one percent!). There will be no single NREN network because NREN was always intended to be an internet - i.e., a network of networks - and I believe NSF has never said anything different. We have said there will be no single federal agency "in charge", and that's simply a recognition of political reality. I think the argument could be made that if a single network were mandated, it would never happen because no federal agency would give up jurisdiction over its own piece of the whole.

j

We controlled it when it started, but few want the network controlled from Washington now.

k

Clearly not counter to everyones' intuition!

l

There was no relevant public law in September, 1990.

m

I think you mean "peremptory". I certainly didn't read it that way. Van Houweling was as familiar as I with the standard provisions of NSF Cooperative Agreements that state that all subawards require Foundation approval; his letter was an expression of intent and we both knew that the details had to be approved by NSF's Grants and Contracts Division.

n

It wasn't a demand. Some of the 74 days were taken in the subaward approval process.

o

In the preceding eleven paragraphs, it is clearly not supportive of your overall position to remark that (a) MCI and IBM won the right to lose money on the NSFNET project in a free and open competition (it was for example never a secret that they were partners inthe Merit proposal), and (b) AT&T has not exactly been a leader in the data networking business so that quotes from Fred Howlett need to be understood in context.

Editor's comment: what the eleven paragraph's remark on is the belief of the individuals interviewed by Markoff that IBM and MCI were succeding in having the conditions of the Cooperative agreement changed in such a way that they could inherent control of the NREN.

p

"Expertise"?? That's a red herring. NSF is a funding agency. We have ample expertise on tap from various levels of advisory structure, to special-purpose workshops, to hiring experts for short or long terms. All in all, I believe the goals of the networking program at NSF are being very well met indeed.

q

Why do you mention Duderstadt? I think you should be very careful about any implication of conflict of interest.

Editor's comment: Conflict of interest implies personal gain and is not alleged. What is alleged that if Duderstadt did not absent himself from all NSB deliberations on the issue, because of the close relationship between MERIT and the University of Michigan, it raises the appearance of self-interest.

r

We continue to deal with NYSERNET as an awardee even though PSI actually provides, manages, and operates the network. What's wrong with that?

s

That's not true. Why do you think that?

Editor's comment: I think this because I was told this by a reviewer with Federal Network contracting expertise.

t

Absolutely. We had no desire (or right) to control the resources - we only retained control of the service we were paying for (via the numbered paragraphs 1 and 2 in my letter to Van Houweling.

u

Oh? By whom? You have a reference for this assertion? Within the academic and the government communities it was and is well known that NSF makes such awards competitively. The 18- month extension was a move of sufficient import to require NSB approval. A renewal of the original agreement for an additional five years would have been out of the question.

Editor's comment: At 9:09 AM, November 9, 1990 Educom's Mike Roberts stated that he viewed ANS's existence "as a backstop against failure of NSF to renew backbone in 92." OTA Computer Conference on NREN Policy Issues. Item 4 Response 12. The NSF had often said it wished to get out of the business of providing the network backbone. The whole Yellow stamp discussion on com- priv in the summer of 1991 was based on the premise of the NSF not rebidding the backbone.

v

That's demand. Capacity remains at 22.5.

x

The trouble with the above is that a Gb/s network is a matter of public policy - both by Administration priority inthe HPCC program and by legislative intent in PL 102-194. You characterize Weis' behaviour as "infatuation" whereas I would describe it as forward-thinking and aligned with the national goals subsequently expressed by both the Executive and Legislative branches of our government.

y

This simply isn't true. When ANS was spun off, the NSFNET Backbone Executive Committee (with representatives from Merit, IBM, MCI, Michigan, and NSF) was expanded to include Weis, and lower-level (e.g., program, engineering) committees were similarly augmented. The operation of the Cooperative Agreement has always internally been referred to as a "partnership" and all parties have participated vigorously and jointly in the activity.

Editor's comment: This seems to be a matter of changing perspective. Note that Dr Wolff told Communication's Daily that the infrastructure pool was not the NSF's responsibility but rather MERIT's, and that MERIT in turn passed the buck to ANS.

z

You characterize the situation as a tendency for "ANS (or perhaps MERIT and ANS) to go their own way" whereas it was a normal and commendable initiative - we wanted ANS and Merit to expand the network. My reminding them of their obligations was not out of a fear they would disregard them, nor that they were running out of control, but simply so that the record would show that they had obligations, and nobody could in future say that "NSF had let ANS run without controls."

aa

It may indicate that to you, but it didn't to me or anybody else I know of. Instead, as I have written above, ANS was so aggressive in augmenting the infrastructure on their own hook and there was no evidence of any ANS-customer-caused congestion that establishing the "technical means of compliance" assumed a low priority. The teething problems of the T3 network were not related to this issue in any way.