We have written a month ago about Linda Sundro's amazing presumptuousness at the National Science Foundation. In the presence of a series of unsure and uncertain Directors of the Foundation, this woman, in her capacity as Inspector General, has gone far beyond the normal bounds of assuring against mismanagement. Instead, according to a series of people who have spoken to us off record, beginning in the summer of 1994, she has made the organization into an agency where many people are afraid to do their jobs unless they know before hand that the IG will approve. Seven years ago, while Eric Bloch was Director, we have learned that he let her know in no uncertain terms that *HE* was running the agency. Since his departure, she has shown a strong interest in pre empting the rest of the agency's policy makers.

The latest example of Linda's meddling follows. When last summer Jim Fleming started his absurd complaints about mis-doings in the administration of domain names, the NSF IG was obligated to take a look. She did so -- as Fleming proudly told the world. Funny thing is her Report that follows contains no mention of the initial reasons for her examination of the issue and no findings of anything being done improperly. There is also no evidence that she had a thorough understanding of the context in which she was operating and that she showed any congnizance of the actions being taken by the agency to make proper policy.

Rather her report is a naked attempt to make sweeping proposals for action by NSF itself that would entail unprecedented changes in the very nature of the agency. She is in over her head. The policy that she recommends is not only for the operation of the NSF and the turning of it into a taxation agency, but also for the agency to level a global tax on the operation of the Internet world wide. In her own words: "we believe that NSF should evaluate options available to recoup or otherwise protect the taxpayers' investment in network research and development. We have identified several approaches that would ensure fairness to the taxpayers, who funded the early network development programs, and would ensure an income stream for NSF to support peer-reviewed internet-related basic research, service, and development projectsÑthereby reducing the need for taxpayer subsidies in the future. These options expand upon steps that NSF has taken in the past to apply registration revenue to network-related service and development costs under program income arrangements.3 We invite NSF management to consider these additional alternatives, which are designed "to promote economy, efficiency, and effectiveness."4

The last item, footnote four, as readers will see below, seems to be her attempt to justify her meddling under the terms of the enabling legislation for inspector generals. After three weeks when the Foundation had not yet signed on to her far fetched initiative, she or someone else leaked her report outside her agency to the Office of Management and Budget and reportedly to the FCC and Commerce Department as well. During the critical last week in February in which NSF's appropriate plans to end the coperative agreemnent early were derailed by OMB and the White house, Sundro's report, not yet acted on by her own agency, was making the rounds downtown -- probably influencing the OMB's actions. Such leakage outside the agency was a wholely inappropriate action for her to take. One clearly clculated to go over the head of her own director. Thus after the CIX, and Rutkowski in February, and the E.C and NSI in April, she may be thought of as the "fifth interventionist" -- the first four being enumerated in detail in the June Cook Report.

Not only does she propose that the NSF could make $60 million a year after expenses from turning the $50 fee on domain names into an excise tax, she also goes on to propose a ten cent a year charge on every IP number. "Given the large number of internet number addresses, even a nominal fee (such as 10 cents) on domestically held numbers would raise substantial funds for internet-related basic research, service, and development."

"Many of these number addresses are not actually in use, but are held by one of three regional registries. Furthermore, for historical reasons, organizations that received number addresses early in the development of the internet have been allocated many more numbers than they actually use. For example, NSF has been allocated more than 66,000 number addresses, but actually uses fewer than ten percent of them. Imposing number address fees would provide an incentive to conserve this limited internet resource." Cook: never mind the accounting costs and that it would rather difficult to determine which numbers are "domestically held."

For those wishing to be further "dazzled" by (1) her failure to understand the broad context of her agencency's own actions, (2) her presumptuous sweeping policy proposal to create a new taxing authority, and (3) finally proposals themselves that are so badly flawed because she lacks the most rundementary understanding of the internet necessary to make sound judgements -- her report follows.

Thankfully, but unfortunately only after damage was done to the process that was going forward, and the spread of a great deal of misinformation and poor adviuce to centers of policy making outside the NSF, whose affects are still being felt, on April 17th the Director's Office declined her recommendations. Our review of this episode shows that someone else ought to be inspecting the Office of Inspector General at NSF.

National Science Fopundation Office of the Inspector General

Report

The Administration of Internet addresses

7 February 1997

Preface: What's In a (Domain) Name?

[Note from Gordon Cook:we have scanned a paper copy of the report and tried to get all the OCR errors out. Apologies for any failures.]

Worldwideweb addresses, such as "www.fastlane.nsf.gov", are now widely used by the general public. They generally consist of a series of letters and/or numbers separated by periods. The part of the address after the last period ("gov" in the above examples) is called the "top-level domain name," and the part of the address immediately to the left of that last period ("nsf" in the examples) is called the "second-level domain name." Currently, top-level domain names are either a two-lettercountry code (such as 'fr" for France) or one of the following three-letter codes: "gov","org", "com", "net", "edu", "mil", or "int". Domain names map to internet number addresses, which actually identify each computer interface attached to the internet and are used to route information over the network. Internet number addresses consist of four l-to-3 digit numbers separated by periods; for example, "128.150.1.1" is a numeric NSF address. Domain names are used because they are easy to remember; the Psmith company is more likely to attract potential customers to www.psmith.com than to 173.75.128.43.

Domain names have to be unique: if "psmith.com" identifies a multinational corporation, it cannot also identify a journalist named Eustace Psmith. Some domain names may have significant commercial value; in our example, the corporation might be willing to outbid the journalist for the use of psmith.com. Currently, all second- level internet domain names in the top-level domains of gov, org, com, net, and edu are administered by Network Solutions Inc. ("NSI") under a cooperative agreement with NSF. Unless extended by NSF, this agreement will expire on 30 September 1998. In this memorandum we address how NSF should administer internet addresses in the future.

Internet Address Registration Fees

The internet can be traced back to a network created by the Department of Defense ("DoD"), and for a long time DoD handled all internet address registrations. In 1991, NSF assumed support for non-military registration services. On January 1, 1993, NSF entered a five-year cooperative agreement with NSI to provide registration services; that agreement was expected to cost approximately $1.0 million annually. In 1994, new registrations began to increase dramatically, primarily in the top-level com domain, and by 1995, NSI's cooperative agreement was costing NSF upwards of $1.9 million annually.

In September 1995, NSF authorized NSI to charge registration fees for second-level domain name registration services. Under the new arrangement, NSI collects fees from the individuals registering in the top-level com, net, and org domains, and it collects $50 annual fees from NSF for registrations in the top-level gov and edu domains. For each $50 fee, $15 goes into a pool "for the preservation and enhancement of the 'Intellectual Infrastructure' of the Internet in general conformance with approved Program Plans."' The remaining $35 is retained by NSI to cover its operating costs and profit.2

The importance of the support of basic research by the federal government is acknowledged across the political spectrum. Through financial support of projects selected according to merit by peer review, NSF supports basic research in a broad range of science and engineering disciplines. THe internet has been heavily funded and promoted by NSF at every stage of its development and expansion; indeed, we believe that NSF deserves substantial credit for the current explosion in the use of the internet. In addition, in contrast to the typical circumstance of NSF funding of innovative research or education projects, internet address registration concerns the allocation of a unique public resource. In this context, we believe that NSF should evaluate options available to recoup or otherwise protect the taxpayers' investment in network research and development. We have identified several approaches that would ensure fairness to the taxpayers, who funded the early network development programs, and would ensure an income stream for NSF to support peer-reviewed internet related basic research, service, and development projectsÑthereby reducing the need for taxpayer subsidies in the future. These options expand upon steps that NSF has taken in the past to apply registration revenue to network-related service and development costs under program income arrangements.3 We invite NSF management to consider these additional alternatives, which are designed "to promote economy, efficiency, and effectiveness."4

How Much Money Is Involved.?

A. Domain Name Registration

Under the amended cooperanve agreement, NSI charges $100 for an initial, two-year domain name registration and $50 for annual renewals.5 In late 1996, NSI anticipated demand for second-level domain name registratiions to grow from 41,700 to 75,000 per month by the end of March 1997. This proved an underestimate: domain name registrations exceeded 75,000 per month in September 1996. As of October 1996, NSI reported cumulative domain name registrations of 740,785, which would result in an annual revenue of more than $36 million.6

More recently, the growth in domain name registrations has begun to moderate. For the nine months preceding the new registranon fees regime, January 1995 through September 1995, cumulative domain name registrations grew an average of 15.8% per month; for the fifteen months since registration fees were instituted, October 1995 through December 1996, cumulative domain name registrations have grown an average of 13.5% per month.7

[Cook: we have ommitted both her charts.] The above chart shows the total number of domain name registrations and the growth rate over the past two years. The growth rate has varied, and obviously there is no way to know for sure how the growth rate will change in the future. For the purposes of this discussion, we wilt presume that the rate of growth will continue to decline at the same rate that it has since fees were instituted in 1995. As the chart below shows, this approach predicts that the number of domain name registrations will reach a plateau in mid-1999 of approximately 4 million, generating annual revenue from registration fees of approximately $200 million. We regard this as a conservative estimate. At the current return of $15 to the "Intellectual Infrastructure7 pool for each $50 fee, the current rate scheme for name registration at that time will be generating more than $60 million per year in revenue for infrastructure development of the internet.

B. Number Registration

Currently, fees are not charged at all for internet number addresses. This means that the costs of the services supported by the fees falls only on those registering names. An NSF official discussing the imposition of the current registration fee structure explained:

"While it is difficult to describe the 'typical' Internet user, generally speaking, they fall into three categories: there are large organizations and corporations, which have a single, or relatively few, domain names. For those users, the costs [of domain name registration] are insignificant. There are those who have Internet access through on-line services such as America Online, Prodigy and CompuServe. These services each have a single domain name for all the end users. That means America Online will pay a single $50 fee annually, regardless of how many subscribers the service has. The third group are small business and individuals who have their own domain name."8

Because internet number addresses, like names, must be unique, number addresses must also be registered. Some customers, such as internet service providers, large companies, and government agencies, have only one second-level domain name but use a great many number addresses, reflecting a greater use of the internet than other customers with only one or just a few number addresses. It seems reasonable that single domain-name users that desire numerous internet number addresses should pay for them. Imposing fees on number as well as name address registration would more equitably distribute the cost burden associated with supported services through the internet community.9

To date, 2 billion internet number addresses have been allocated world wide.10 A flat rate fee for internet number addresses would reflect the &ct that from an end- user perspective number addresses, unlike names, are equivalent. On the other hand, because internet number addresses, unlike names, are used to route information through the network, number addresses are not equivalent from an engineering perspective. The fee structure could be chosen to provide financial incentives for a best practices" in assigning number addresses within computer networks. Given the large number of internet number addresses, even a nominal fee (such as 10 cents) on domestically held numbers would raise substantial funds for internet-related basic research, service, and development.

Many of these number addresses are not actually in use, but are held by one of three regional registries. Furthermore, for historical reasons, organizations that received number addresses early in the development of the internet have been allocated many more numbers than they actually use. For example, NSF has been allocated more than 66,000 number addresses, but actually uses fewer than ten percent of them. Imposing number address fees would provide an incentive to conserve this limited internet resource.11

C. Projected Above-Cost Revenues

Under the Government Performance and Results Act, starting in fiscal year 1997 NSF must submit a strategic plan for all program activities covering a period of at least five years. On the basis of the assumptions discussed above, we estimate that by retaining the above-cost revenue from name and/or number registration over a five- year planning period beginning in 1998, NSF would be able to allocate more than $300 million in additional h~nds for network-related basic research, service, and development costs.

III What Should Be Done

Federal Oversight of Internet Addresses Should Continue

No one disputes the importance of the internet to the government and society. In announcing the "Next Generation Internet" initiative, the White House noted that the internet provides students ready access to remote library collections, brings market information to business entrepreneurs, and allows citizens ready access to the records of their elected representatives.l2 That initiative calls for Ucreative" investments to "enable a new generation of applications that support scientific research, national security, distance education, environmental monitoring, and health care."13 NSF was identified by the White House as one of the implementing agencies for the initiative.

The growing dependence of the nation on the internet underscores the importance of its continued availability to the public. For the internet to operate, the origination and destination points for information routed over the network must have unique addresses. Just as using the telephone system requires a phone number, using the internet requires an internet number address. The people, through their elected representatives, have historically exercised control over these internet addresses. In our view, public administration of this unique public resource should continue.

Government authority over internet addresses, which continues to be administered under federal awards, is a matter of historical fact, but some have questioned the government's legal authority in this area. Some have complained, for example, that NSF has given a "monopoly" to NSI. NSI is acting pursuant to a legally binding agreement with NSF, in a manner that NSF has scrutinized and determined to be acceptable. The NSF/NSI agreement has not conveyed any authority to NSI that extends beyond the duration of the agreement. The agreement expires on 30 September 1998, and the agency has not yet decided upon a process for overseeing internet addresses after the period of the cooperative agreement ends.

If, after the period of the cooperative agreement ends, NSI were not operating under NSF direction and were somehow able to continue to provide its current registration services and collect registration fees, nothing would currently prevent NSI from using its de facto control of internet addresses to prolSt from granting access to the internet.14 In light of the significant public interest in the continuing stability of the internet and the large federal investments at stake, we recommend that federal oversight of internet addresses continue.15

Federal Oversight of Internet Addresses Should Generate Income for the Government

The federal government has made major investments in the creation of the internet. Operational networks developed by federal agencies include ARPANET (DoD), ESNET (Department of Energy ("DoE")), and the NASA Science Internet, as well as NSFs NSFNET backbone and regional networks. In addition to developing operational precursors and subsidizing their use by the research and education community, federal funding has supported research and development of related technologies. For example, from FY 1990 through FY 1995, NSF funding for the Networking and Communications Research and Infrastructure division ("NCRI") within the Directorate for Computer and Information Science and Engineering ("CISE") exceeded $230 million.16 During this period NCRI funding supported fundamental research on communications theory and data networks as well as the NSFNET program."

The federal government will continue to invest in the internet in the foreseeable future. NSF's support for networking and communications research and infrastructure was approximately $55 million in FY 1996.18 In its Justification of Estimates of Appropriations to the Congress for FY 1997, NSF identified "networking and communications research focusing on new technologies for high bandwidth communications, and on the convergence of computing and communicanons" as a high-priority activity for CISE; the total CISE budget request for E:Y 1997 was $277 million.l9 In the fall of 1996, the President identified the need for a $500 million investment in the next generation internet over the next five years.20 The contributions of NSF and the other implementing agencies toward this new initiative would total $100 million in FY 1998. In response to the initiative, NSF is expected to include additional hlnding for the High Performance Computing & Communications program in its FY 1998 budget request.

In the increasingly difflcult budget environment that faces us, NSF must be able to find creative ways to leverage its assets to further its support of peer reviewed basic research in a broad range of science and engineering disciplines. We believe it would be appropriate for income from the administration of internet addresses to supplement the government's investment in the internet and related technologies. We estimate the above-cost annual income from name registrations will exceed $60 million a year. Income from registration fees could ultimately make NSF's investment in the internet self-sustaining, with revenue to be used to hlnd network-related basic research, service, and development. With this income, for example, NSF could hlnd much of the NCRI budget or the next generation internet initiative.21

Disbursement of the "Intellectual Infrastructure" Pool

Under NSF's cooperative agreement with NSI, as amended, the gross income from registration fees "will be treated as 'Program Income"' and allocated as follows:

"a. 70% will be available to [NSI] as consideration for the services provided[;]

"b. the remaining 30% will be placed into an interest-bearing account which will be used for the preservation and enhancement of the 'Intellectual Infrastructure' of the Internet in general conformance with approved Program Plans. [NSI] will develop and implement mechanisms to ensure the involvement of the Internet communities in determining and overseeing disbursements from this account. [NSI] will also establish and maintain publicly available records of all deposits to and disbursements from the account."

For the same reasons that future above-cost revenues from internet address registrations should come back to NSF, we believe it would be best if the funds going into this pool could be provided to NSF to be used to support internet-relate research, development, and education projects selected by merit-based peer review.22 Alternatively, NSF should ensure that the funds in the pool are used to support only specific NSF-approved network-related basic research, service, and development projects, in coordination with related projects supported directly by NSF.

According to NSI, $12,817,262 has been deposited in the pool through 22 January 1997.23 At a minimum, our recommended approach would result in use of thisS12.8 million for network-related basic research, service, and development projects approved by NSF.24 We estimate that the total amount of registration user fees charged by NSI under the cooperative agreement will reach nearly $400 million by 30 September 1998, when the agreement expires, with a total of $120 million owed to the infrastructure pool at that time.

D. Fair and Open Decisionmaking

To ensure that the internet address allocation rules and fee structure adopted by NSF are fair, NSF should follow procedures for facilitating public participation and open decisionmaking. For example, NSF went through a public process when it was considering how the NSFNET program should evolve. NSF should disseminate the draft policies and requests for comments broadly, on the internet as well as via traditional means, and NSF should accept comments via the internet.

E. Options for Federal Administration of Internet Addresses

We suggest three different administrative options: traditional agency administration, administration through a performance-based organization ("PBO"), or administration through an independent commission. Each of these options could accommodate different ways of registering domain names. For example, registranon services could be performed by several different organizations in competition with one another for customers; alternatively, a single organization could be compennvely selected to provide the service at a reasonable profit for a fixed period. All of these options would, however, ensure that federal oversight of internet addresses continues and that income generated from the administration of internet addresses would be used to supplement federal investment in network-related basic research, service, and development.

Under the first option, either agency personnel would administer the internet addresses or agency personnel would oversee the administration of the internet Administration Bv a Performance-Based Organization addresses by a private contractor. Under either the PBO or independent commission options, a new governmental organization would be created to administer the internet addresses. We briefly outline the pros and cons of these options, assuming oversight by NSF. 25

1. Traditional Agency Administration

NSF's technical expertise, history of involvement, and continuing investment in internet-related research, development, and education make it a natural choice for the role of oversight agency. NSF possesses the requisite understanding of the important technical issues and the confidence of the research community to apportion the funds derived from the administration of the internet addresses wisely through its peer review process among its research programs to the benefit of the nation as a whole, as well as the internet community.

However, without additional hiring authority, direct NSF administration of internet addresses would divert its existing staff into administrative tasks at the expense of the very projects that qualify NSF to be the oversight agency. Alternatively, NSF could oversee the administration of internet addresses by a private contractor. Because substantial NSF staff time would be required to administer the internet addresses or to oversee the contractor performing that duty, these options would only be acceptable if sufficient (I) personnel (FTEs) and (2) budget authority can be allocated to this function.

A benefit of choosing to oversee a private contractor's administration of internet addresses is that it allows NSF to seek a third party to perform a function that it has no direct experience performing. Under this alternative, the contractor's role would be limited to providing the registration services specified by NSF; responsibility for developing the rules for allocating the address space and the fee structure would remain with NSF. This restriction would allow NSF to fully realize the effciency gains that could be realized through a competitively bid contract for registration services. NSF could also encourage competition by allowing several contractors to provide registration services in different parts of the name and/or number address space.

2. Administration by a Performance Based organization

In March 1996, the Vice President proposed the creation of PBOs within the federal government as part of the ongoing reinventing government initiative. The purpose of PBOs is to restructure agency activities so that certain functions that "resemble the activities of a business and can be measured by business standards" are provided on a business-like basis.26 PBOs are headed by chief executive offlcers who are hired under contract and are held accountable for results. In order to achieve those results, PBOs are exempted from many government rules, such as personnel rules, that preclude federal agencies from operating like private sector organizations.

PBO proposals generally concern the transfer of existing government functions that are separable from policy-making and regulatory activity to a new organization within an existing government department. In considering whether to adopt a PBO structure for some of its functions, the President's Management Council emphasizes that PBO functions must be separable from policy-making and regulatory activity.

Internet address administration could be considered for PBO status within NSF, because the "business" of allocating internet addresses is fully separable from NSF's policy-oriented function of providing financial support to research and education in science and engineering. At the same time, however, in our view the public interest requires internet address administration to remain a governmental activity, to ensure continuity and fairness, and to provide revenue to support research and education in science and engineering, in particular, research in next-generation computer networking. Transferring this service to a PBO would serve to separate this "business"activity from NSF's research support activities, and encourage NSF to concentrate its attention through the PBO on: (1) the internet community receiving the service and (2) the revenue produced from the service, rather than viewing its relationship with the incumbent registration services provider in the context of its research support activities. The PBO should be authorized to select an efficient method of delivering registration services. The revenue received from the internet address administration would not be retained by the PBO, except to the extent required to support its administrative duties, because deciding how best to allocate these funds to internet- related research is within the scope of NSF's statutory mandate to support basic research.

The President's Management Council suggests using three criteria in evaluating PBO candidates: (l) "Does the [candidatel have a clear mission and provide a service that is measurable?. . . [(2)] Does the [candidate] have the capacity to measure progress toward meeting its mission and objectives? . . . [(3)] Is there a clear line of accountability to an agency head and does he or she support the transformation of the unit to a PBO?"27 We believe these criteria are met in this case, and NSF may wish to seek legislative authority to create a PBO. The internet address administration service is well-defined. Because the service has been provided pursuant to agreements overseen by NSF, the agency has the capacity to measure progress toward meeting measurable service objectives. Finally, because the PBO would return revenue to the agency in addition to providing a measurable service to the client community, a basis for performance accountability is assured.

3. Administration By an Independent Commission

Under this option, an independent entity, similar to the Arctic Research Commission ("ARC"), would be created by federal legislation to provide federal oversight of internet addresses and ensure future funding for network-related basic research, service, and development. To ensure broad representation on the Federal Internet Commission ("Commission"), its members would be selected by the President from the academic community, the internet-user community, private industry, and the general public. The Director of NSF would serve as an ex offlcio member of the Commission.

In order that the Commission be self-sustaining, administrative costs associated with its oversight duties would be recovered from the address registration fees. To minimize administrative costs, the Commission would be authorized to use NSF staff services and facilities on a cost reimbursable basis and NSF would budget for the Commission's activities under the Research and Related Appropriations heading, as is the case with ARC.

If this option is pursued, we recommend that the enabling legislation charge the Commission with the selection of an efflcient method for providing registration services and the development of a fee structure for internet address registration that would make NSF's investment in the internet self-sustaining, with revenue to be used to fund network-related basic research, service and development. We further recommend that the legislation require the Commission to remit revenue from those fees in excess of reasonable administrative costs to NSF for disbursement in support of peer-reviewed, network-related basic research, service, and development projects.

IV. Conclusion

The current federal oversight of name and number internet addresses is the natural consequence of federal financial support of internet development. Continued federal oversight of this unique public resource is required by the nation's increasing dependence on the internet, which is being fostered by additional federal investments in this technology. NSF's history of involvement with the internet, its technical expertise, and its continuing investment in related research programs uniquely qualify it to perform that oversight role. NSF's oversight would ensure the protection of the public interest in the resource, the availability of funds to support future network-related basic research, service, and development, fairness to the internet community, and fairness to the taxpayers.

Recommendations

l. We recommend that NSF continue the cooperative agreement with NSI through the end of its term (30 September 1998) to ensure full funding of the infrastructure pool.

2. We recommend that NSF use the income from the administration of internet addresses to supplement direct federal appropriations, with an ultimate objective of making NSF's investment in network-related basic research, service, and development self-sustaining.

3. We recommend that NSF ensure that the "Intellectual Infrastructure" pool which is being funded by NSI be used to support only specific NSF approved network-related basic research, service, and development projects, in coordination with related projects supported directly by NSF.

4. We recommend that NSF: (l) determine which administrative option providing federal oversight of internet addresses (by NSF, a performance based organization, or an independent commission) would best ensure that income generated from the administration of internet addresses supplements federal investment in network-related basic research, service, and development; (2) seek Congressional approval, as appropriate, of its selected oversight structure; and (3) adopt fair and open procedures that facilitate public participation (via the internet as well as traditional means) in the development of the internet address allocation rules and fee structure.

Absent such federal oversight of in~ernet name and number addresses, we recommend that NSF urge the Federal Communications Commission to consider exercising its authority under the Communications Act of 1934, as amended, to ensure impartial and equitable allocation of internet addresses.

1 Cooperative Agreement No. NCR-9218742, Amendment No. 04 (13 September 1995). NSI initially proposed to pay 30 percent of the registration fees directly to NSF. Letter from NSI to Staff Associate, NCRI/CISE (5 May 1995).

2 We have not at this time evaluated the profit NSI derives from this fee arrangement. See also note 24.

3 Cooperative Agreement No. NCR-9218742, art. 15 (l January 1993) (directing that registration fees be used tO pay project expenses of related InterNIC awards charged to NSF).

4 Inspector General Act, 5 U.S.C. app. ¤ 2(2)(A).

5 These registration fees apply only to second-level domain names in the top-level domains of gov, org, com, net, and edu.

6 The cumulative domain name registrations statistic includes 3807 second-level domain names in the two-letter country code domains, which must be excluded when estimating revenue from registration fees. Thus, 740,785 minus 3,807 is 736,978, times $50 is $36,848,900. TLlis and other estimates in the memorandum generally rely on the fact that every registration generates $50 per year.

7 We define the growth rate for December 1995, for example, as the ratio of the difference between the cumulative domain name registrations reported for December 1995 and November 1995 to the cumulative domain name registrations reported for November 1995. Our calculations include cumulative registration statistics from InterNlC's 24 September 1996, Quarterly Review (through October 1995) and from InterNIC News, December 1996 and January 1997 (November 1995 through December 1996).

8 Staff Associate, NCRI/CISE, email message concerning "New F[requentlyl A[sked]Q[uestions] (Gen[eral] Public Version) on Fee Imposition" (13 September 1995).

9 For example, organizations that extensively subdivide their domain name space and assign many number addresses within that subdivided name space would also pay thc number address registration fees.

10 This does not include the billion numbers (the upper half of the class A address space) held in reserve by the Internet Assigned Numbers Authority, operated by the Information Sciences Institute at the University of Southern California. See Philip J. Nesser II, An Appeal to the Internet Community to Return Unused IP Networks (Prefixes) to the IANA, Request for Comments 1917 (February 1996) (www.internic.net/rfc/rfcl917.txt).

11 For a brief discussion of the limited number of IP addresses, see Philip J. Nesser II, An Appeal to the Internet Community to Return Unused IP Networks (Prefixes) to the IANA, Request for Comments 1917 (February 1996) (www.internic.net/rfc/rfc1917.txt). Sec also K. Hubbard et al., Internet Registry IP Allocation Guidelines (November 1996). Neither of these documents discusses the use of internet number address fees as a conservation incentive.

12 Office of the Press Secretary, The White House, Background on Clinton Gore Administration's Next-Generation Internet Initiative (10 October 1996).

13. Id.

14 Under the plan we recommend, one option is to ensure some level of competition among private contractors seeking to provide registration services under NSF oversight. Sec infra part E. 1.

15 Absent such federal oversight of internet name and number addresses, we recommend that NSF urge the Federal Communications Commission to consider exercising its authority under the Communications Act of 1934, as amended, to ensure impartial and equitable allocation of internet addresses.

16 This total is computed from data available through the NSF Executive Information System.

17 NSF, Justification of Estimates of Appropriations to the Congress fiscal Year 1992, at CISE-23; NSF, Justification of Estimates of Appropriations to the Congress Fiscal Year 1995, at 113.

18 NSF, Justification of Estimates of Appropriations to the Congress Fiscal Year 1997, at 93.

19 Id. at 94

20 Steve Holland, "Clinton: 'Every Home Connected to the Internet'," Reuters (10 October 1996).

21 NSF should determine how the funds from the fees could be received by NSF, either under current authority or pursuant to legislative amendment. For example, NSF has authority to credit to its Research and Related Activities appropriation "receipts for scientific support services and materials furnishet by . . . other National Science Foundation supported research facilities" and may be able to receive income from the fees pursuant to this provision.

22 NSF should determine whether the funds in the pool could be either (1) transferred by NSI to NSF, under NSF's authority to credit to its Research and Related Activities appropriation "receipts for scientific support services and materials furnished by . . . other National Science Foundation supported research facilities", to supplement the NCRI budget to support internet- related research, development, and education projects, or (2) donated by NSI, without restriction, to the NSF trust fund, and then transferred to supplement the NCRI budget to support internet-related research, development, and education projects.

23 Donald Telage, Alternatives for Disbursing Internet Intellectual Infrastructure Funds, Presented to: The National Science Foundation (30 January 1997). This total reflects collections through December 1996 and includes interest earned. Id.

24 This number is a minimum estimate of available funts because NSI is owed additional fees for domain name registration. David S. Hilzenrath, Master of Internet Domains Says It Loses at Monopoly, Wash. Post, 24 January 1997, at Dl, D3. In our view, this collection issue is separate from the issue of revenues discussed above and will ultimately be resolved. If a person or entity registered or renewed a domain name, then it is obligated to pay, and under the terms of the cooperative agreement NSI is obligated to collect. Accordingly, this revenue is owed to the pool. This situation will be clarified when we audit NSI's costs, revenues, and practices under the cooperative agreement.

25 Unlike NSF, DoD and DOE are mission agencies; their involvement in internet-related research and development has been constrained by their primary goals. The Federal Communications Commission has confronted similar resource allocation issues, as a regulator, in connection with the telephone numbering system and radio frequency spectrum, but does not now invest in basic scientific research.

26 Vice President Al Gore, Address at the National Press Club, in Washington, D.C. 14 March 1996).

27 Reinvention's Next Steps: Governing in a Balanced Butget World, Backgrount papers Supporting a Speech by Vice President Al Gore (4 March 1996).