ANS Attempts to Mend Fences with the Regionals

O'Neil reports that "twenty-five individuals representing 20 different regional and state network service organizations attended this meeting." When the attendees were asked what they wanted to get out of the session the following common themes emerged: "Some regionals receive services from ANS and would like to understand how to expand service offerings and how new services can be phased in.

ANS's 'identity crisis' confuses regionals and requires clarifying.

Some of ANS's actions are inconsistent with its words.

Unclear if ANS is pursuing a zero-sum or win-win strategy with the regionals.

Some mid-levels work with ANS while other compete with ANS and participants want to understand why each behavior occurs.

Networking community needs to work together to help dispel 'the fog' for ourselves, existing users, and potential users of Internet services by creating and acting on a shared vision.

Need for a single network to support R&E and CO uses.

Regionals and ANS share same interest in supporting the R&E community and want that to continue.

Desire on the part of senior ANS staff to 'listen' to regionals and determine how to mix the growth of the regionals with the growth of ANS."⁹³

This is the complete list of points from the ANS memo. In order to give readers a feeling for the balance of events as described the list of issues from the 'ask Al' session described below will also be given in complete form.

O'Neil goes on to summarize the agenda discussion: "During the course of reviewing the proposed agenda, the participants preferred to work together as a single group on all topics rather than dividing up into smaller breakout sessions. This decision resulted in scrapping the original agenda in favor of allocating blocks of time to each major topic. The initial time called for Al Weis . . . to respond to a few commonly asked questions. . . This quickly became known as the ask Al segment and it will be included in future advisory meetings. Al Weis began by addressing the question of whether ANS is a regional or backbone provider of network services. The discussion that ensued quickly began to cover other topics that had been planned for review later in the meeting. These were the extent to which ANS competes with the regionals, the perceived mixed messages that ANS gives to the regionals about who they are, what they're about, and how they plan to achieve their goals.

The exchange that developed during this segment highlighted many issues that continued to be refined over the two days. Several points are worth noting:

ANS is interested in exploring any and all ways of deriving cooperative relationships that allow regional networks to thrive and grow as providers of local services.

Regional networks want ANS to thrive and grow as a provider of reliable backbone services and to incorporate fast packet technology.

Competition for organizations with existing connections seems fruitless

-- attention should be focused on addressing issues to help promote unattached organizations to become connected to the Internet.

Considerable confusion exists as to cost structures and the meaning of full average cost, and what ANS's prices were; ANS's price list was made available.

ANS was beginning to focus on industry market segments with national account focus and regionals could act as local distribution channels.

ANS should decide what business it is in and become good at it -- this will help the regionals figure out what they want to do with ANS based on ANS's strengths and abilities to provide services.

Commonplace today for organizations to be suppliers, competitors, partners and consumers of services all at the same time; these complex relationships must be managed well to minimize conflicts and maximize cooperation.

Regional and ANS share common organizational form, mission, and goals on behalf of the R&E community, promoting broad connectivity and encouraging technology transfer and economic growth.

Sharing a common vision is a strong argument for developing cooperative agreements."⁹⁴

These topics circumscribed the area out of which several more discussion group grew as the meeting progressed. The memo presents somewhat fragmentary ideas (best understood no doubt by the attendees) in great detail. The most significant can be summed up under four general areas. The first of these centered on problems in understanding and dealing with the ANS agreements some 15 months after they had been outlined in somewhat rudimentary detail by Al Weis in his Proposal to Steve Wolff and eight months after they had been launched publicly at the Montana FARnet meeting of August 1991.

The Agreements Still Present Problems

"We need to understand the difference between fixed and variable charges.

What is the purpose of the Infrastructure Pool?

Do we need the RAC and Infrastructure Pool? (Should we fund common needs instead?)

Can the Pool funds be used to create and develop quality services that add value to end users?

What does it take to allow the regionals to have commercial attachments to the ANS backbone?

How can we make sure that everyone plays and the playing field is fair?"⁹⁶

Under the discussion of combits the questions asked showed no greater focus and as O'Neil added, couldn't be adequately answered in the time available.

"What are the responsibilities of commercial service provision (non economic) ?

Who will ensure adherence to ground rules in the Internet [once the guiding hand of the NSF is gone]?

Can or will the successor [to combits] be a flat charge algorithm (annual) based on 'capacity'? Independent of use restrictions?

What's 'capacity?' How do we define the base level of service?

Can the transfer of funds occur via the NSF rather than the combit mechanism?"

Combits is not workable for some regionals because it is not understandable or predictable in terms of costs to end user organizations.

Why do we have to worry about usage at all?"⁹⁷

A perhaps more significant group of discussion points emerged over the question of ANS's role. Glenn Ricart remarked that "ANS is unsure if it is in the wholesale or the retail business -- being in both businesses simultaneously is confusing." As an unfortunate result "we are all fighting over the 2% of the customers already connected rather than the 98% that are unconnected." An unresolved issue Ricart concluded was "what is ANS going to be good at."⁹⁸

John Rugo had a suggestion for ANS. "He called on ANS to consider other roles that might include introducing new technologies, the tools to manage them and user services which regional networks cannot easily provide. This would be similar to the way in which Bellcore serves the RBOCs."⁹⁹

On the final day a significant conclusion emerged: "ANS is probably untenable only as a backbone provider."¹⁰⁰

Other questions raised were: "Are the regional to be 'first' line service providers for ANS? Are all value added services to be provided by ANS? What consistent set of parameters can ANS specify and the regionals fulfill to determine the boundaries on service?"¹⁰¹

At the end of the first day the regionals had decided that they wanted a "menu" of "models for CO+RE use pricing that includes:

• retaining combits,
• charging a flat fee based on the size and classification of pipes behind the gateway,
• a dedicated commercial gateway price that assumes full use without regard to the pipes behind it,
• the flexibility to change from one of these options to another."¹⁰²

"Regionals and ANS share the same vision of supporting R&E, promoting technology transfer across all sectors of society and economic growth, and providing broad access and connectivity.

Desire to develop win-win model hat allows both to grow and be financially sound.

ANS to work national accounts.

Regionals to act as distribution channel for national accounts and ANS services.

ANS to introduce and develop management tools for new technologies and common user services (Bellcore model).

Regionals and ANS work to supply answers and clear the fog

[Both to work to] find the proper mix between production network versus introduction of new technologies.

ANS to clearly state its mission and goals (and then act accordingly) to help dispel the fog."¹⁰³

"It was agreed that a small group be formed to work on a draft cooperative agreement between ANS and the regional networks."¹⁰⁴ At the meeting's end the "consensus was that the candid comments, suggestions and ensuing discussions were difficult but productive. . . . . The feeling was that the next meeting should be triggered by the results of the small working group on the cooperative agreement and other issues mentioned in this meeting. Several participants expressed a sense of urgency for moving forward together as quickly as possible."¹⁰⁵

Translating Good Intentions into Action?

In an article on ANS's application for 501 (c)(3) status in the September COOK Report on Internet -> NREN, the author emphasized the infrastructure pool and suggested that enough time had passed for ANS to be expected to account for its promises and to report to the network community on the status of the pool.¹⁰⁷ On October 23, 1992 the author posted a query to com-priv asking about the status of the infrastructure pool that ending with the following words: "ANS may be a private company but it runs the infrastructure pool as a public trust. Or if it does not, then who does run the pool? And how much money has the privatization of the NSFnet backbone raised for the good of the network's research and education members?" The author asked the same question in a slightly different form on October 27. On November 2 when no answers had been posted by either the NSF or MERIT or ANS the author began to ask the question somewhat more insistently. On November 6, in view of silence from Steve Wolff that was hard to understand, a reader of the list suggested privately to the author that the time had come to use the Freedom of Information Act to extract the data from a strangely reluctant National Science Foundation. Since the reader included detailed instructions on how to execute such a request the author was able to file it the next day. To its credit, the NSF was quite prompt in its response with several hundred pages of data on which much of this report is based arriving at the author's home on November 30.¹⁰⁸

Documents received under FOIA from the National Science Foundation reveal that on October 15, 1992 Peter O'Neil, ANS Director for Regional Networks, sent by email to Resource Allocation Committee (RAC) members an invitation to schedule the first meeting of the RAC by conference call. The meeting took place from 1 to 2pm Eastern Standard Time on October 23, 1992. Those contacted were informed that their invitations as members of the RAC came because they were either a signatory of the ANS Connectivity Agreement or ANS CO+RE Gateway Agreement, or as a representative of the National Science Foundation.¹⁰⁹

The invitation stated that: "It is now time to develop a process to combine our experiences and understand the infrastructure impacts of providing commercial traffic along with research and education traffic. The purpose of this meeting is to identify the work and tasks that need to be performed prior to a face-to-face meeting. . . . . [that] can be held in conjunction with other national meeting that RAC members plan to attend. . . . . In addition to developing the appropriate criteria for distributing the pool funds, we propose that the committee members each undertake the task of gathering usage data, and the relevant member attitudes and behavior so that technical, economic and organizational considerations can be taken into account."¹¹⁰

On November 3, 1992 ANS Vice president Guy Almes sent the following message to the RAC: "Brock Meeks of Communications Daily has expressed interest in the activities of the RAC. Our sense is that the RAC would work best if the details of its deliberations including the detailed minutes were kept confidential. I have just posted a draft of these detailed minutes -- please do review them for corrections or improvements. I'll be reviewing them as well. If the RAC generally agrees that such details should be kept confidential, then that raises the issue of how best to report to Meeks and others. One possibility would be to review and edit the following two documents drafted by Jim for this purpose. The first labeled 'Background Information,' discusses the RAC as an organization. The second labeled 'Report of a Meeting' summarizes the first meeting. Please comment on the basic approach to internal versus external reporting."

The Background Information Draft is very similar in content to what was released to FARnet on November 6, 1992 -- with one exception. In the text of the draft ends with the statement: "the National Science Foundation was kept apprised of the various discussions and agreements by which the Infrastructure Pool and the RAC were conceived and implemented." This sentence is missing from the document was released to FARnet.

One can only speculate as to why. The answer seems to have something to do with the position of the NSF vis-a-vis the RAC. Questions at the NSF were escalating. By November 5th, the author had become rather insistent in the questions that he posted to the com-priv list. Professor Dave Farber (one of the most senior and respected members of the Internet) had sent the author's lengthy posting of November 3 to either Steve Wolff or Chuck Brownstein - - from the material released to the author it is not clear which. Early in the morning of November 5th Brownstein replied to Professor Farber's message which contained the following statement from Farber: "I have not distributed past versions of this from Cook waiting for the NSF or ANS to reply so I could send both at once. No reply has been forthcoming to any list I monitor." Brownstein's reply to Farber - copied to Wolff was laconic - "this seems to be in the class of questions, 'When did Gordon stop beating his dog?'"¹¹¹ That evening Farber posted the author's unanswered query on the infrastructure pool to his "Interesting Person's" mail list.

Meanwhile it is impossible to understand why Steve Wolff would have refused to answer the author's questions to com-priv unless he lacked any information with which to answer them. When Brownstein refused to become involved on the morning of November 5, Wolff may have gotten on the phone to ANS and insisted that he be told what was happening with the pool and the RAC. For at 4:20 PM on November 5, Kristen Mortensen (Al Weis' secretary) sent the following message to Wolff: "Steve, I sent quite a bit of msg from WEIS-ID to you. All are on the first discussions that have taken place on RAC mailing. [Meeting was probably meant.] I believe you have everything today. Take care, Kristin."¹¹² If this line of reasoning is correct, since the author announced on com-priv that evening that he would make a FOIA request from the NSF, Wolff may have decided simply to wait for that to arrive rather than post a reply to the com-priv list. In any case, the material about the NSF being fully informed was removed from the draft announcement perhaps at the request of the NSF.

The "Report of a Meeting" draft by Al Weis differs significantly from what was released to FARnet on November 6th. It states that 12 out of the 16¹¹³ members participated in a meeting where four actions were taken:

1. That "distribution of funds initially would be based on an informal variation of the Grant Approval Process used by NSF. . . Distributions would be determined by use of a philosophical approach of how the Regional Networks and Backbone Network could be improved, and not on quantitative data until enough data is available to indicate where funds will be needed to solve specific problems.

2. Infrastructure would be defined in a broad way to include both physical components and the software used in connection with the networks. This reflected the view that (a) software improvements can be as important as, and sometimes even more important than, improvement in the physical components of the networks, and (b) the most positive impact on the Regional Networks and Backbone Network may be achieved by using funds for improving software.

3. A subcommittee was formed to develop a process, criteria and a draft solicitation model to be used by the RAC.

4. Another subcommittee was formed to develop the process by which proposals submitted to the RAC were to be reviewed."¹¹⁴

Of course one could ask questions about both areas which is probably why she recommended their elimination. One wonders in particular about the elimination of point two on infrastructure and software. Perhaps the focus of point two derived from the fact that the amount of money in the Infrastructure pool is very small. Possibly it would be enough to pay for the creation of some kind of user interface program that could be given to all RAC members? Or perhaps it could be used to pay for improvements in routing software? Software after all is divisible in a way that hardware and bit pipes are not. In any case, the focus on software was a long way removed from the earlier focus on driving the network to gigabits by the end of 1992.

In response to a query from the author about who would benefit from membership in the RAC a member of the RAC replied:

"I am indeed in the group as are the others only because we signed an agreement a long time ago saying that we had no objection to having commercial traffic arrive on our networks from outside. I do not expect to benefit much if at all. Presently there is not much money in the pool even after all this time. The amount of commercial activity has been minimal. I suppose this could change in the future, but none of us are expecting any kind of great windfalls from it." At about the same time the author was shown a recent ANS document stating that the amount of money in the pool was $62,000. While such an amount of money wouldn't buy much hardware, it might buy some software.

Meanwhile on Friday November 6, 1992 Guy Almes sent Steve Wolff a copy of the official release on the RAC and Infrastructure Pool that he was about to send to FARnet. This version, which announced that the next formal meeting of the RAC would be in January of 1993, now had the sentence about the National Science Foundation being kept apprised of developments removed. The membership list of RAC members had also grown by one: Steve Wolff was now listed as a full fledged member.¹¹⁶ Since apparently somewhat after the fact, he had been made a member, there was apparently now no need to say that the NSF had been kept apprised of developments.

From examining the RAC deliberations, one is forced to conclude that there is evidence that ANS' plan to bring commercial traffic into the network and raise a bonanza for reinvestment in infrastructure have not been a smashing success. The amount of money in the pool was only a fourth of what ANS had predicted it would be a year earlier.¹¹⁷ AL Weis, who had never enjoyed broad public scrutiny of his plans, seemed determined to operate the part of his plan that seemed most designed to serve the public interest in such a way that it would be well shielded from public scrutiny. What is somewhat harder to understand is why the mid- level directors who were members of the RAC seemed quite willing to abet his desires. Everyone, it seemed, was developing a siege mentality. And no one unfortunately seemed to remember Mitch Kapor's December 1, 1990 caution to Al Weis about the dangers of making policy outside of the public view.

Consider the following incident. Guy Almes on Friday November 13 sent the following message to the RAC: "We've had some inquiries about the amount of money in the infrastructure pool. We'd like to know what RAC members think about this. Should we disclose it publicly, or would you prefer that it remain a private matter of the RAC? We could go either way with it."¹¹⁸

The following day Mike Staman Director of CICNet replied:

"A conundrum. I do not care. However you have a lose-lose situation here. Fail to disclose it and continued criticism will accrue . . . Disclose it and at least three items will emerge simultaneously:

1. Incredulity at the number (no matter what the size)
2. Calls for accounting (no matter what the size)
3. Re-emergence of the same criticism as if you selected option one.