Print 

This is a special issue about the rapidly declining ability of the NSF to cope with increasing crises that will affect the outcome of network commercialization in very serious ways.


Slashed NSF Staff Faces "Catastrophic Situation" Endangering Creation of New Internet Architecture, pp. 1-24

At the most critical moments of Internet commercialization extremely important components of government oversight within the National Science Foundation have ground almost to a halt. Crises affecting the commercial viability of the Internet are beginning to blaze out of control. We describe four of them in separate lengthy articles. A source inside the Foundation calls the present situation "catastrophic."

Nearly a year to the day after the award of the vBNS to MCI, that cooperative agreement has still not emerged from the Division of Grants and Agreements which, out of fear of the NSF Inspector General, is tending to take a month to clear paper work that it used to clear in two weeks. What is worse a veritable rule by the IG since the departure of Eric Bloc as Director (as described to us by 4 different sources over two year's time) has created both an environment where "people are afraid to do their jobs" and cliques that have impacted the ability of the Division of Network Infrastructure and Communications Research (DNCRI) to function well. As a result, and unknown to almost all on the outside, staffing on the DNCRI Infrastructure side is down by almost 50% from its level of two years ago. Due to bureaucratic haggling and cliques outside of DNCRI, a serious search for Steve Wolff's replacement as Director did not get started until October - more than 6 months after he announced his departure.

Without a permanent director and working with half staff between now and April 30 DNCRI must execute (1) a roll out of the NAPs and shut down of the ANS backbone, (2) implementation by MCI of the vBNS at the 5 supercomputer centers, (3) debut the routing arbiter (4) transition the regional nets to the NAPs, (5) revamp the InterNic after the termination of the General Atomics component, (6) issue a new solicitation for an International Connections Manager, and (7) develop a new program for research connections to the vBNS.

This would be more than most people could be expected to handle. However the commercialization process has brought on new crises. It is the addition of these crises that have caused some of our sources to use terms like catastrophic. They will have several effects. Among them: increasing costs of usage for small ISPs and individual users, aggregating authority and consequent power in the hands of the of the telcos - both IXCs and RBOCs, and in general creating a rising danger that DNCRI's ability to maintain a level playing field through out the next critical months of commercialization may be lost. We describe four such flash points in great detail.

(1) Roughly two weeks ago DNCRI was the indirect target of a lawsuit filed by KnowledgeNet, a small IBM mid range consulting company in Illinois. This company lists as defendants, David Boone who runs his own consulting service using KnowledgeNet as an association and has registered it as a domain name, DIGEX Boone's service provider, and Network Solutions which assigned the domain name to Boone. The suit levels 5 charges against Boone including theft of trade marks and racketeering under the 1970 RICO statute!

Sources close to the case feel it is likely that the suit will go to trial. If Boone looses case law will be established linking domain names to trade marks. Regardless of what happens the next year will very likely see the imposition of charges for commercial domain names. A loss of this suit could mean that the registering authority would have to employ trademark attorneys to check a flood of new commercial domain name applications, and buy legal liability insurance. This would be likely to increase the annual cost of a domain name by an order of magnitude.

(2) For the first time in its history DNCRI has suspended a 5 year cooperative agreement before the second year was even up. The General Atomics part of the InterNic has gone nowhere since it began in April of 1993. In less than two years three different people (Susan Estrada, Kent England, and Bob Randall) were in charge. Things that GA had promised to do under the agreement were simply not done. At the end of last year 13 of 16 review panel members recommend to DNCRI that there be no funding to GA for years 3 through 5 beginning on April 1, 1995. We present interviews with Kent England and Susan Calcari, and a detailed chronology by Susan Calcari.

Karsten Blue, the young son of GA owner Neil Blue (observers report unanimously that Karsten appears to be in his mid twenties) was installed last fall as President of a newly created subsidiary that ran both CERFnet and the InterNic. When Karsten received notice from NSF in late January that the GA Internic was being suspended April 1, he proceeded to immediately lay off Susan Calcari, the remaining co PI on the project. Unfortunately Susan was governed by a key personnel clause that stipulated that she could not be laid off without the prior consent of the National Science Foundation. In doing so GA committed a serious breach of the agreement. As a result they received a letter from NSF this week notifying them that the agreement was suspended immediately with some transitional processes running until February 28 when total suspension would take place a month earlier than other wise would have happened. These events have robbed and will continue to rob DNCRI staff of time they simply do not have.

(3) During the last three months CIDR blocks of network addresses have become increasingly difficult to get. Net99 describes their problems in an interview. Network solutions is taking the position that unless a strong case to the contrary can be made new blocs of addresses (needed by every ISP to sell and run SLIP accounts without which WEB browsers cannot be used) will go only to the largest providers with national backbones. These large providers are then expected to parcel them out to their resellers. And as long as the web surfing craze lasts they become like a fuel without which the smaller ISPs simply cannot do business. Yet the stocks of "fuel" are under the control of their larger competitors. This situation is complicated by the fact that there is a valid technical reason for handing the addresses out this way - namely to keep the size of routing tables under control so that the memory requirements for routers do not get out of hand.

(4) At the Chicago and California NAPs technical problems caused by the lack of router to ATM switch interfaces (and warned about last August by Milo Medin) have forced the imposition of FDDI LANs. The Ameritech NAP has been further marred by a controversy with NET99 where it lost this customer's $173,000 5 year contract for T-3 service in December. After being reamed out by Net99, Ameritech proceeded to loose the replacement contract a second time in January! The result is an approximate 60 day delay for NET99 in connecting to Chicago. We have on record statements and interviews with NSF, Bellcore, Joe Stroup of Net99 and George Clapp, Mark Knopper's boss at Ameritech. Clapp's remarks indicate to us that responsible parties in AADSnet, the Ameritech unregulated subsidiary in charge of the NAP had every possible reason to believe that Net99 was trying to connect as early November. A cynic might wonder if AADSnet's misplacement of Net99's applications had anything to do with the fact that Karl Denninger the proprietor of MCS is Net99's VP of Engineering, and is by far the strongest provider in the Chicago area. Denninger is connecting himself as a part of Net99 to the NAP, and is therefore the most significant competitor of AADSnet's planned Internet service in the largest city served by Ameritech. In other areas that we explore AADSnet's presentation has not been very professional. When a source inside Ameritech told us that Ameritech's internal security was investigating AADSnet business practices in both Michigan and Illinois, Clapp replied that the investigation actually was directed against incursions into AADSnet machines coming from outside. These events indicate that RBOCs in control of NAPs in the current frenzy is fraught with possibilities for conflict of interest.

Some general policy considerations: DNCRI is in need of urgent help from the highest levels of management of NSF and the White House. Appointing a new director who is beholden to no particular research or infrastructure constituency should be the top priority. Filling staff vacancies the next. Finally, having once criticized the vBNS, we have come to support it. We are told that some in the foundation would like to merge DNCRI into oblivion and deliver the vBNS completely to the supercomputer centers who have no real interest in networking research. However DNCRI has won a commitment from MCI for a parallel but totally separate test network announced so far only in last month's COOK Report. It is believed that this test network will be the only viable opportunity to explore theories of high speed data networking that do not depend on ATM into which the telcos are sinking tens of billions. Some in the community are claiming valid reason to believe that there are better ways to do data networking than ATM. The test net is seen as an opportunity to find out and pull the telco's chestnuts from the fire should the researchers be right - something that is certainly in the national strategic interest. It is an opportunity that will be lost, insiders believe, if bureaucratic infighting at NSF merges DNCRI into Advanced Scientific Computing.

We have taken these concerns by telephone and email to Tom Kalil of the White House Economic Council and Mike Nelson of OSTP. These are Vice President Gore's top two staffers on these issues. Last evening Kalil replied. His message contained among other things assurances that "senior management at NSF is working hard to fill the vacancies at DNCRI," and that "we have indicated to NSF that the White House would be happy to help them in their recruitment efforts."


Additionally

In addition to a general introduction to the situation, our February issue contains articles on each of the four crises described above. We also publish in full the Bellcore progress report on the California and Chicago NAPs sent to NSF on February 3, and the complete text of the KnowledgeNet Domain Name lawsuit described above.