NSF TO IMPOSE PERFORMANCE CRITERIA ON VBNS PAYMENTS PP. 1, 19-22A guaranteed throughput under high stress conditions of only 5 megabits per second on the 155 megabit vBNS service is a fall out of pushing the envelope by the NSF and MCI in a new technology where performance is difficult to measure and hardware and software interfaces are not yet complete. While practical service will be much higher than the minimum, some in the super computing community are saying that the vBNS will not be really useful until MCI expands capacity on its commercial ATM network to OC12 (622 megabits per second) - something not expected to happen until the second half of 1995. The vBNS will be simply a group of PVCs overlayed on MCI's commercial ATM backbone with tail circuits running to the Super Computer centers. The cooperative agreement with MCI for vBNS implementation also is not yet signed. DNCRI thought it best to recover from five months of protests by putting the agreements for the remaining parts of the new architecture in place first. The paper work for the vBNS is expected to go to the Foundation's Office of Grants and Agreements next week. Don Mitchell, the DNCRI Staff Associate who is the negotiations specialist for Steve Wolff's group, in a lengthy interview, explained that NSF hoped to have an operational vBNS in place on April 1 and stated that monthly payments to MCI for the vBNS would not begin until the vBNS with tail circuits to the NAPs was turned on. He also stated that payments to MCI would be dependant on MCI achieving mutually agreed on performance criteria. Such criteria would be set on a quarterly basis with throughput metrics rising throughout the project. Once set the criteria would be public as would measurements showing MCI's compliance. He explained that setting the criteria would be difficult because of the complex ways in which the performance of an ATM network can be measured. We include a technical note which Peter Ford kindly wrote for us and which explains in general terms why these metrics are so complex. We applaud DNCRI for taking steps which should bring more fiscal responsibility into what has been seen as a cooperative agreement process that gives the awardee too much of a blank check.
STEVE GOLDSTEIN DISCUSSES PAST AND FUTURE OF INTERNATIONAL CONNECTIONS PROGRAM, pp. 1-5The NSF's International connections program is little known outside the inner circles of the Internet policy community. In an interview, Steve Goldstein describes the programs accidental beginnings and explains how Sprint as the first International Connections Manager worked with NSF to help make the US with its low international circuit rates one of the most desirable places in the world for other countries to come for their Internet connections. Steve also explains that the new solicitation will include plans to phase out NSF support for the major trans Atlantic links on the grounds that these links are becoming commodities that no longer need government subsidies. Other components of the new solicitation are likely to include support for high bandwidth connections to experimental networks like the vBNS and support for the delivery of network services to remote areas such as ships at sea and places like the rain forest.
LITTLE GARDEN, pp. 6-8Tom Jennings tell us how a few in the San Francisco Silicon Valley hacker community created and extraordinarily successful home brew internet service provider designed on the premise that resale of network connections could work to everyone's advantage.
A CRITICAL LOOK AT THE UNIVERSITY OF MICHIGAN'S ROLE IN THE 1987 MERIT AGREEMENT, pp. 9-17Recent University of Michigan graduate Chetly Zarko has spent much of the last two years digging in the University of Michigan archives and serving the University authorities with FOIA requests for the contents of computer conferences and other University documents. He sheds much light on the way in which Dr Doug van Houwelling was brought in to shepard a major technology push at the University one that including ousting Amdahl as the University mainframe and bringing in IBM who quickly became a joint study partner in the 1987 Merit Cooperative Agreement for the NSF backbone. Zarko describes the complex series of accounting arrangements between Merit and the University of Michigan as tempting University administrators to identify all possible loop holes in the cooperative agreement and manipulate them to the University's advantage. In early 1992 Robert Moore an accountant in the UM Information Technology division filed a False Claims Act against the University that resulted in an out-of-court settlement between the University of Michigan and the government. U-M agreed to repay nearly $3.1 million in computer processing overcharges that involved federally funded research. In exchange, the government accepted U-M's explanation that the charges were "inadvertent," and therefore dropped the case. While this affair was not apart of the Merit NSF cooperate agreement, Zarko shows it as part of the same institutional system that appears not to have enough checks and balances in it on behalf of the public interest. It is in this respect that this primarily historical research we believe has meaning for the present and future. Anything that the NSF in the future can do exert more effective oversight with regard to the way its money is spent is desirable. Zarko's article helps to underline the importance of DNCRI's movement toward a fixed unit price arrangement and away for payment for costs incurred with the vBNS Cooperative Agreement. We publish the University's rebuttal.
ISF FOREIGN OPERATIONS DIRECTOR ACCUSES LARGEST RUSSIAN NETWORK OF FRAUD, pp. 18-19In a letter to the Editor Dr Alex Goldfarb accuses Relcom of plan to appropriate for its own use $1,000,000 in ISF and Russian gov't paid for hardware. Allegations against Relcom include involvement in a secret protocol, surrogate non profit entities, & behind the scenes deal as part of fraudulent scheme. We publish Goldfarb's letter and offer a rebuttal based on our own knowledge and research.