David H. Crocker, Brandenburg Consulting
in regards to the
Department Of Commerce
National Telecommunications and Information Administration
15 CFR Chapter XXIII
[Docket No. 980212036-8036-01]
Improvement of Technical Management
of Internet Names and Addresses
22 March 1998
It has taken 30 years for the Internet to become an overnight success. Over the last 10 years it has been fulfilling its technical and communications promise, incrementally moving from a centrally-controlled, specialized tool within a restricted community, to a global resource available to all. The subject plan ("Green Paper") considers what is perhaps the final step in this 10 years sequence, final termination of direct United States Government involvement in funding or directing Internet processes.
I welcome the opportunity to comment on the proposal and look forward to its revision in line with its goal of Internet self-governance.
I have been a participant in the Internetís development for most of the its 30 year history, first in the research community and more recently in the commercial sector developing products and service, including founding an Internet startup. In addition to contributions to various Internet technical standards, I had the privilege of serving in a management capacity for the Internet Engineering Task Force (IETF) standards process, including direct oversight of Domain Name Service (DNS) development. Recently I was named by the Internet Assigned Numbers Authority (IANA) to participate in an emergency effort to enhance administration for "generic" domain names, through the International Ad Hoc Committee (IAHC). The IAHC produced the gTLD Memorandum of Understanding (gTLD MoU). My participation was as unfunded volunteer work over the course of one year and was performed out of the typical concern for proper evolution of the Internet. As for any consultant engaged in such professional volunteer work, the time spent covered not covered by an employer. In other words, the donation is from the individual and not an employer. Further, I have no past, present or planned financial interests in DNS administration or operation.
The United States government has made a seminal contribution to global communications, through its initial and continuing funding and direction of technology which produced today's Internet. Since 1989, the Internet has been moving from being the tool of a limited research and academic community, to a critical piece of global infrastructure essential for commerce. The United States government has been divesting itself of funding and management responsibilities over Internet development and administration. The Green Paper (GP) represents a proposal for completing the last stage of that divestiture, namely independent funding and oversight of Internet administration involving and derived from the Internet Assigned Numbers Authority (IANA). The GP copies many ideas and details from prior work by IANA and under the gTLD Memorandum of Understanding (gTLD MoU). However it makes basic changes to the plans in that prior work, but does not adequately justify them.
Rather than the current approach of fine-grained intervention and direction taken in the GP, the government needs to work as a participant-among-equals in the global Internet community's process of evolving administration naturally. Rather than arguing that the history of U.S. government funding provides the government with special authority, the GP needs to recognize and honor the true basis for authority of Internet administrative organization, namely the consent of the Internet community.
There is extensive and well-established demand for new "generic" top-level domain names (gTLD). The current limited set of gTLDs imposes a significant hardship on organization seeking to register new names. Unfortunately gTLD administration touches many, disparate and important issues, ranging from Internet technical structure to international trademark law. The gTLD MoU was produced through intense, extensive negotiations among many stakeholders and represents a carefully-crafted set of compromises. The GP essentially ignores that considerable and constructive effort and tries to start afresh, but without conducting true, open and incremental negotiations as was done -- and continues to be done -- for the gTLD MoU. The GP provides concessions to a minority of the trademark community and to those seeking excessive commercial profit through exclusive control over gTLDs. In contrast with these special interests, the gTLD MoU represents a balanced set of comprom! ises among formal trademark associations, end-users, and provider organizations. Development and pursuit of additions to the domain name service is a continuing effort, with incremental revision. It is reasonable and appropriate to consider and pursue refinements for the work that is underway. However such refinements must be achieved incrementally and without delaying existing efforts.
Internet development and administration
From research to commerce
The United States Government (USG) has had primary responsibility for the original development of the Internet. Beginning with seminal research in the mid-1960's, through early experiments with the Defense Advanced Research Projects Agency Arpanet and CATENet, into formal, production operation of the Internet in 1983, continuing into creation of a second, parallel backbone with multiple regional networks, funded by the National Science Foundation. In recent years, this obscure effort has emerged as the embodiment of long-standing dreams for global, personal and commercial data communications.
It is essential to appreciate that the USG did not make its seminal contributions in isolation, even from the earliest days. A number of other countries participated in the development effort. Today expenditures from other governments and from the private sector vastly outweigh total expenditures by the USG, never mind current annual budgeting.
The Internet is based on "open" standards. These permits companies to compete against each other, while offering the same underlying technology. Further the technology interworks, so that different organizations can buy products from different vendors, yet the organizations can use those products to talk with each other.
Achieving such interworking requires both competition and cooperation. The nature of that requirement is often misunderstood. Competition exists and is beneficial in terms of software, hardware and value-added services, such as Internet service providers.
This competition has always been predicated upon a central source for the specification of technology. The Internet Engineering Task Force (IETF) is that central source. The need for single authority over technical specifications has long-acceptance within the Internet community. What is often misunderstood is the equal importance for central authority over global Internet configuration and administration information. The central authority for this work is the Internet Assigned Numbers Authority (IANA), which administers assignments for Internet Protocol addresses, Domain Names, and various other protocol parameters.
Both IANA and the IETF began as creatures of the U.S. government, both in terms of funding and in terms of guidance. Over the last ten years, much of this has changed. As the Internet has grown to become a global resource, so has the operational basis for IANA and the IETF. We are currently undergoing the final stage of this transition, rather than the first stage. There is no legislation which requires conformance to the work of IANA or the IETF. Their success and their authority come only and specifically from the support -- and therefore the consent -- of the entire Internet community.
Within the IETF, the basis of this support is known as "rough consensus", with the word "rough" meaning both "approximate" and "difficult". Easy decisions which receive unanimous support are extremely unusual in the IETF. This operational basis for the IETF is well-established and well-accepted. What is generally misunderstood is that it also applies to IANA. We must expect the same reality of "roughness" when seeking consensus for changes to Internet administration. At any time, an alternative consensus could develop to support a different organization. While this would be traumatic for the Internet, there is nothing to prevent it. Nothing, that is, except the continued global support for IANA's work. It is therefore simply inappropriate for any one organization or government to claim special authority over IANA's work. While this may well be an uncomfortable state of affairs for those providing the funding, the current intent of the Green Paper needs to consider and accept it.
Separation of efforts
An important lesson in the development of Internet technology has been to divide complicated problems into separate pieces, so that work can be well focussed and deliver useful pieces of work in a short time. A clear and consistent lessson for the Internet and many other large systems is that trying to solve many complicated problems together usually fails, as well as taking a long time. The Green Paper acknowledges the urgency of making some changes immediately. Hower it then makes the mistake of creating too many dependencies on other actions, thereby ensuring that further delays will be experienced.
The Green paper attempts to deal with a number of different issues:
These can be pursued largely independently, rather than be intimately intertwined as currently proposed by the Green Paper. IANA is already pursuing its own evolution. The USG does not need to direct the details of this, although constructive assistance would of course be helpful. The USG should support existing IANA evolution efforts.
The DNS roots are often characterized negatively as being operated by "volunteers". While a technically correct assessment, it misses the essential fact that all of those volunteers are highly experienced professionals and most perform their tasks within the environment of established, major Internet organizations. The informal nature of the arrangements under which they operate may well need to be modified into a structure more comfortable for those with concerns. However it is essential to understand that there is no operational DNS root server problem and that the changes should be viewed as natural evolution rather than problem solving.
Internet Protocol (IP) address assignment has already achieved a reasonably solid, comfortable global operational structure. It, too, needs to evolve naturally rather than to be "repaired". The Green Paper acknowledges the adequacy of existing IP address assignment mechanisms, although it is inconsistent in taking not taking that assessment and applying it equally to work surrounding the DNS.
An extensive effort has been underway for nearly 1 Ĺ years to develop additions to the set of gTLDs. This effort began with the International Ad Hoc Committee, produced the gTLD Memorandum of Understanding (gTLD MoU), and is being implemented through the Policy Oversight Committee (POC) and the Council of Registrars (CORE). This effort was based on a broad base of direct participation, received formal signatory support form more than 200 organizations around the globe and has the active financial investment and operational efforts of 88 organizations. The process for this work was based on well established principals of Internet openness. Further it was instigated under the established authority of IANA and IANA was an enabling signatory for the gTLD MoU. Although the recipient of strong criticism, the ultimate demonstration of its success has been both the presence of strong support and the lack of any other, equally-organized or sustained, competing effort. The only current barrier to the success of this effort is the U.S. government's assertion of formal authority over IANA and the root servers. If the US government is seriously interested in Internet self-governance then it needs to refrain from blocking extensive, well-organized and well-supported community-based efforts.
Lastly is the question of NSI. The US government changed NSI's operation from one of cost-plus reimbursement into an unbounded commercial monopoly. It did this without following the usual methods of assigning such a largess. The US government needs to repair this problem, and it should do so in a way which does not leave the users of the Internet subject to unregulated commercial monopoly control. However, the details of dealing with the difficulties created by the US government's handling of the NSI contract and NSI's charging, do not need to be tied to any of the other activities. Pursue them separately.
U.S. government, past and present
Initiative and funding
Both the initiative and the sustained direction and funding of Internet development, provided by the USG through ARPA and NSF, represent an extraordinary contribution to human knowledge and human utility. There are few examples in history of such global benefit achieved primarily through a single government.
In addition to the obvious kudos due for the original technical work, it is worth making special mention of a USG initiative which is not generally appreciated: Creation of NSFNet. Prior to the creation of the NSFNet backbone, the Internet had only one backbone (long-haul) network connecting the many user networks. That backbone was funded by ARPA and operated by Bolt Beranek and Newman. The software technology which allowed it to coordinate the backbone was a "routing protocol" which supported only the one backbone. Creation of a second, parallel backbone necessitated enhancing the routing protocol. This requirement moved the formal model of the Internet from a single, central authority to one of full distribution. In other words, creation of NSFNet enabled creation of the fully competitive, open Internet service environment which now exists. I would like to take special note of the vision and tenacity by then-program manager at NSF, Dr. Stephen Wolff, who was responsible for this initiative.
Global evolution, global authority
As important as it is to acknowledge history, it is also essential to acknowledge when things change. The Internet is no longer controlled by the United States government or United States organizations. The Internet has become truly global. A policeman in a remote part of Malaysia asked me whether the Internet had been invented in Malaysia. Billboards in Italy advertise local companies, with local business, using ".com" domain names. A retired dentist in Alsace-Lorraine, in France, communicates with long-lost family members in Houston and San Francisco using Internet mail.
It is past time for the United States government to recognize that the child has grown up and has developed its own authority. It has become a creature of the globe, not just the United States, although in truth there has always been a substantial contribution from the Internet community, dating back to the very beginning of the Internet.
This change makes the USG only one player among many. It is essential that the USG not abuse its physical power over certain key participants and organizations, in order to assert authority. If the USG is serious about the desire for Internet self-governance, then it needs to respect community-based efforts that are already underway and it needs to encourage, rather than direct, any necessary additional efforts.
Too much detail, too little time
The Green Paper states broad goals, but it also provides vastly too much detail and it provides it much too late to be useful. Rather it serves to cause delays in the evolution that was already underway. Much of the detail in the GP was copied from existing work by IANA and by the gTLD MoU, although it acknowledges neither. If these efforts are worthy enough to use their work, they are worthy enough to continue on their own. To the extent that specific modifications to their work are deemed appropriate by the USG, then the USG should openly and directly discuss the specific desired changes with the agencies already conducting the work and with the global community. That is quite different from claiming direct initiative and forcing a loss of momentum for existing efforts.
In spite of the controversy surrounding the gTLD MoU, the GP needs to recognize the breadth of effort which created it and the breadth of international support which exists for it. Opposition is vocal but comes only from some very specialized sectors of self-interest. In some cases, those self-interests can be adequately served by making incremental changes to the existing efforts, rather than re-starting the development effort, causing more years of delay.
Contradiction of existing decisions
The Green Paper forcefully asserts tha IANA's authority derives from the USG. Although I believe this position to be fundamentally incorrect, the position of the USG is nonetheless self-contradictory, using it's own claims.
IANA initiated the gTLD MoU effot and IANA provided an enabling signature to the MoU. Further, the USG supplied a participant to the development work of the IAHC which produced the gTLD MoU. A senior staff member from NSF, who was also part of the inter-agency Federal Networking Council (FNC), was a member of the IAHC. Hence there are two lines of USG "authority" which establish USG approval for the existing gTLD MoU work. Hence the Green Paper specification of a new gTLD effort, and one which contains fundamental differences from the principals and details in the gTLD MoU, represents a reversal of existing USG decisions.
Within the GP there is also an inconsistency between the administrative model for IP address model and that from gTLDs. The GP supports the top-level control over IP address assignment being done by a set of not-for-profit agencies. However for gTLDs it supports use of unregulated commercial agencies. The U.S. government needs to recognize the need for top-level administration of Internet administrative infrastructure resources to be done by not-for-profit entities operated in the public interest.
Although the Internet supports a high degree of supplier competition, there some ways in which users are inherently "locked in" and unable to participate in legitimate, free-market processes. The nature of Internet technology makes Internet service providers "locked in" to the group of Internet addresses delegated to them. A provider which must change that group must, in turn, force all its users to change their addresses. This is very detrimental to their retention of those users. Hence, top-level delegation of addresses must be conducted in a fashion which acknowledges the need for stable, fair assignment, done in a public interest style. The GP recognizes this requirement for IP addresses.
The same issues attend assignment of top-level domain names. Although users may have considerable flexibility in their original choice of domain name, they come to have no effective choice after starting to use the name. If a user is dependent upon a single, unregulated, commercial agency for its domain name, then that agency comes to have effectively complete control over the user. The cost of changing a domain name is much worse than the considerable cost of changing a telephone number, due to the fact that domain names are embedded in Web pages. For popular Web pages, it is not uncommon to have more than 100,000 links from other pages. If the owner of such a popular Web page must change their domain name, all of those links become useless. In many cases this will effectively destroy the company owning that popular Web page. Hence it is essential that domain names be administered under the same public interest model being applied to IP addresses.
Benefits of competition
True competition benefits consumers, by improving choice and reducing costs. It also benefits suppliers by ensuring on-going efficiency and alertness. Suppliers that have too much control over their customer relations tend to become complacent, creating long-term vulnerability in their business operation, as market forces change. Unfortunately "competition" can be difficult to achieve in some cases, due to the nature of technological or administrative requirements. In the case of domain names, it is possible to have considerable competition at the time a user first registers a name, but no real competition after that. In other words, users become "locked in" to their names and hence become entirely dependent upon the agency which administers the domain name registration data base (registry). It is possible to separate the sales and support functions among competing registrars, but operation of the registry data base must be conducted by a single agency, for each portion (top-level domain) of the DNS. It is certainly useful to structure DNS administration to encourage competition for initial registrations, but it is essential to recognize that there can be no meaningful competition among registries, at the time of registration renewal; the user is lock-in.
Choosing appropriate financial models
Ideally, competition should take place between unfettered organizations. Further there is long history indicating the benefits of having those organizations operating on a for-profit basis, since it improves the incentives for the employees and owners of the organizations. However, competition does not require a for-profit basis. In the case of top-level domain name administration it is both necessary and appropriate to have registration-time competition be between non-profit organizations. Given the excessive control over users after initial registration, an organization that has a for-profit basis will be inclined to charge excessive fees. There will be no way to prevent such actions and there will be every incentive for them.
Separation of efforts
Internet administration entails many, complicated requirements and processes. It is essential that the requirements be satisfied and the processes be successful. Long experience with the management of Internet technology and project has taught us to find ways to divide such complicated efforts into a series of independent efforts. The Green Paper does not provide an adequate basis for such separation of efforts.
The Green Paper asserts "experiments" with new gTLDs, but provides details which work against effectiveness in those experiments. It also permits irreversible actions, so that the term "experiment" is inappropriate. As with the plan being pursued by POC/CORE, the GP recommends a go-slow approach to the addition of new gTLDs. The set of names planned by POC/CORE was developed through extensive discussion, attending both to technical and end-user issues. The number is small, thereby ensuring no technical problems, but has a variety of names to explore the styles of names which will be popular with users. Inexplicably the GP chose a smaller number, reversing the careful discussions and agreement among the developers of the gTLD MoU. Further, the GP proposes to allocate one name per new data base registry organization.
This has two major problems. First it ensures absolutely no economies of scale in the operation of these new registries. Registry operation requires considerable investment. However the incremental cost of operation by a single registry for multiple names is tiny. If the new registries are to compete effectively against the well-entrenched operation of Network Solutions, they need to have an opportunity to achieve meaningful economies of cost, from the start. This requires assigning multiple names to any new registry.
The second problem is that there is no way to predict the likely popularity of any single, new name. To require a registry to depend entirely on the success of a single name is to ensure that some registries will fail.
The GP permits registries to be organized on a for-profit basis. Extensive discussion in formulating the gTLD MoU achieved a broad "rough" consensus that registries need to operate according to the public interest. As noted above, there is very strong indication that this requires both public oversight and not-for-profit (cost recovery) basis. If the GP is truly interested in go-slow, incremental and experimental changes to the Internet, it must divide the types of experiments and defer the ones which do not need to be conducted now. It is possible to authorize new, non-profit registries, and later consider permitting creation of for-profit registries. However it is not realistic to permit for-profit registries now and later discover that they are not beneficial for Internet users and decide to terminate them. For that matter, it will be possible to permit not-for-profit registries and later decide to terminate them, if necessary. The U.S. government needs to take proper cognizance of the types of steps which are incremental and reversible and those that are not.
The Green Paper represents an important step in the U.S. government's final divestiture of direct responsibilities for Internet funding and administration. The paper's proposal need to be modified to properly recognize existing, community-based efforts at self-governance, rather than having the USG directly provide fine-grained detail and direction. The Internet has become a global resource and it is no longer appropriate for the USG to claim ownership or control of Internet administration. Although the USG has considerable power over the current administrators of the Internet, use of that power is entirely inappropriate. The international community has already recognized the US-centric views in the Green Paper and considerable damage will be done if those views are allowed to persist. The most simple and most effective path for the USG to follow is to acknowledge that existing evolutionary efforts have adequate, international support, rather than to have th! e US G take a more active and controlling role. To the extent that existing efforts need refinement, the USG should purse changes as incremental, rather than starting entirely new projects as if no progress had been made by the existing, community efforts to evolve Internet self-governance.
David H. Crocker
675 Spruce Dr.
Sunnyvale, CA 94086 USA
TEL: +1 408 246 8253
FAX: +1 408 249 6205