Milton L. Mueller <firstname.lastname@example.org>
Different approaches to top-level domain naming embody three conflicting visions of Internet governance. One vision, which bases top-level domain names on ISO 3166 country codes, represents an attempt to force the Internet into the traditional governance structure of nation-states. An alternative vision bases top-level domain names on "generic," meaningful categories and features company or organization names at the second level. A third principle of domain naming puts top priority on the problem of reconciling domain names with company trademarks.
Using this analytical framework, the paper examines the results of the International Ad Hoc Committee (IAHC) decisions and makes two arguments:
(1) "Generic," content-oriented top-level domain naming is the principle best suited to the growth and development of the Internet. Such an approach maximizes competition and choice and minimizes the opportunity for political and legal conflicts inimical to the interests of the user public.
(2) The IAHC decisions of February 1997 issued mixed signals regarding which of the three principles of domain naming will be implemented. The paper argues that this lack of consistency may cause trouble in the future.
Domain naming and registration are crucial to ensuring that all users of the Internet are connected to each other. Because domain names must be unique, their creation and registration have been among the few centralized points of authority in the supposedly open, decentralized world of the Internet. Precisely for that reason, the process of domain naming has served as a lightning rod for legal and political controversies surrounding Internet governance and commercialization.
In early February 1997 a hastily assembled task force called the International Ad Hoc Committee (IAHC) issued a ruling making important changes in the way domain names are administered. Taking this ruling as its point of departure, this paper analyzes the political economy of domain naming and its implications for the future evolution of the Internet. The paper focuses on top-level domain naming in particular. It explains how different approaches to top-level domains embody three conflicting visions of Internet governance. One vision, which bases top-level domain names on ISO 3166 country codes, represents an attempt to force the Internet into the traditional governance structure of nation-states. This approach links top-level domains to national political boundaries, thereby encouraging an alignment of Internet administration and policy with traditional nation-state structures. An alternative vision bases top-level domain names on "generic" names that have some contextual meaning. Examples include ".com," ".gov," ".edu," or any other three-to-five letter string that suggests a distinct type of content or a specific function. These generic top-level domain names are completely disassociated from national political boundaries; registrations under ".com," for example, can be and are made by companies anywhere in the world. A third principle of domain naming puts top priority on the problem of reconciling domain names with company trademarks. This approach attempts to subordinate Internet domain names to prior legal claims for trademark status by finding a way to map domain names, which are internationally unique identifiers, onto trademark registrations, which vary by jurisdiction and by industry.
Using this analytical framework, the paper makes two arguments:
(1) "Generic," content-oriented top-level domain naming is the principle best suited to the growth and development of the Internet. Such an approach maximizes competition and choice and minimizes the opportunity for political and legal interference inimical to the interest of the user public.
(2) The IAHC decisions, while positive in several important respects, did not make a clear or coherent statement regarding which of the three principles of domain naming IAHC plans to pursue. Instead, the plan combined elements of all three approaches. The paper argues that a hybrid approach to domain naming principles is not viable. The inconsistency may cause serious trouble in the future, although it is possible that the problem will resolve itself over time through the choices of users.
Recent controversies over domain naming have produced an abundance of literature explaining what domain naming is and how it works. This paper provides a cursory description of the system, the evolution of top-level domain naming, and its descent into crisis in 1996. Readers who are familiar with the story may want to advance to the next section, Three warring principles of domain naming.
In order to be connected to the Internet, each host computer must have a unique Internet Protocol (IP) address. IP addresses are merely long strings of numbers. At their simplest, domain names map IP addresses onto names that users can remember and key into a computer more easily. More importantly, the domain naming system (DNS) divides administrative authority over these names into a hierarchy, starting with top-level domains (TLDs) and proceeding down to second-level domains (SLDs), third-level domains, and so on. Hierarchical organization is required to maintain universal connectivity in a network of expanding scale. A computer embedded within a DNS hierarchy need not know the address of every other user of the Internet to enable communication; it only needs to know a few names at the hierarchical levels above and below it, and of course the names of the computers at the same level in the hierarchy. Thus, domain naming neatly fulfills both supply-side and demand-side purposes. It makes addresses easier to use and it makes connectivity easier and more efficient to administer.
The DNS took its existing shape between 1981 and 1987. The impetus for its development was growth in the Internet. The first document to describe what is in effect the system of domain naming that existed until 1997, RFC 920 (October 1984), associated domains with organizations and conceived of domain registration as a hierarchy of delegation among organizations.
It was natural, then, for RFC 920 to create a few top-level domains that symbolized certain categories of organizations. These categories reflected the range of organizations that the ARPA-Internet and DARPA research community of the early-to-mid-1980s had to deal with: military, commercial, governmental, and educational.
The authors of RFC 920 also proposed, essentially as an afterthought, the idea of using two-letter country codes as top-level domains for "countries." At the time this idea was proposed, no country domains had been established. According to Jon Postel,
...the country code TLDs were pretty much an afterthought to a difficult process of coming up with the original generic TLDs. A comment was made that some people might want to have names specific to their own country, and it was pointed out that a convenient list of country codes existed (ISO 3166).
The current TLD categorization scheme was thus a product of circumstances peculiar to the ARPA and defense-related research community in the U.S. in the 1980s. That categorization scheme was unprepared for, and unsuited to, the commercialization and the globalization of the Internet that took place in the mid-1990s.
The commercialization and global spread of the Internet from 1993 to 1996 put enormous pressure upon the DNS. The root of the problem was a collision between the old Internet policy of handing out domain name registrations for free on a first-come, first-served basis and the extraordinary growth of registrations in the ".com" domain. Monthly registrations in that domain grew from 4,000 in 1993 to 45,000 in 1996.
Growing consumption of the ".com" namespace had several fateful consequences:
Many of these controversies are aired in the "newdom" list. The archives can be accessed at http://www.ar.com/lists/newdom.
The domain naming issue thus precipitated a full-fledged crisis in Internet governance. Even though the primary issue had been the artificial scarcity of commercial TLDs, in the context of the global expansion of the Internet this problem quickly metastasized into a struggle for legal authority over the Internet.
The IAHC was formed in response to this crisis. It mission was to reform TLDs in a way that broadened the decision-making process to include more stakeholders while allowing ISOC and IANA to retain their preeminent position in Internet governance.
The IAHC was formally created in November 1996, though discussions of its formation date back at least to late August. Task force members were selected through informal consultations among the Internet Administration Board (IAB), the Internet Society (ISOC), IANA, the International Telecommunications Union (ITU), and two international organizations concerned with intellectual property, the INTA and the World Intellectual Property Organization (WIPO). Although IAHC does represent a broadening of the decision-making process, expansion is incremental and cautious. Of the 11 members of IAHC, six are resident in the U.S. The committee is chaired by the president and CEO of ISOC and includes the chair of the Federal Networking Council. Additional members are chosen primarily to represent trademark concerns, although the ITU also managed to win an important role for itself.
An accumulation of historical accidents over the course of the Internet's development has bequeathed to us a system of top-level domain naming that is based on three distinct principles. These are:
(1) the generic TLDs, which partition the name space at the top level into categories that denote a specific kind of content or function;
(2) the country-code TLDs, which divide top-level domains into categories based on national political boundaries; and
(3) the principle that second-level registrations within a TLD space constitute a form of intellectual property that must be legally consistent with, if not identical to, trademark registrations.
Obviously, the principles can coexist and will continue for some time. But how stable and functional will this hybrid TLD regime be? This section argues that each principle creates a distinct and often conflicting set of expectations among users and suppliers as to what a domain name is, how to use it, and what legal and political obligations go with it.
Generic TLDs (gTLDs) act as a rudimentary index to the type of content associated with a URL. As such, they have some mnemonic value for users. But they are not, and cannot be, a substitute for real indexes or catalogues with much deeper hierarchies and richer categories. In practice, gTLDs almost always involve SLDs that represent the name or brand of a particular organization. That is because commercial users (and many nonprofits as well) view their domain name as an extension of their organizational identity. Accordingly, they have a strong incentive to adopt URLs that are as simple as possible--and that means URLs that have as few levels as possible. A deeper, more content-oriented hierarchy would eliminate many conflicts over names. But such a hierarchy, while appropriate for Usenet newsgroups that organize topics of discussion, would not suit the needs of organizations that want their Internet addresses to be "advertisable" and easily identifiable. Thus, in a competitive, commercialized environment, gTLDs put a premium on a flatter name space. It is not inconceivable that in the future, intermediaries will gain control of a suite of interesting, meaningful, and easily remembered SLDs and parlay them into the ability to sell third-level domain names to businesses that have been crowded out of desirable second-level domain space. The point, however, is that gTLDs encourage namespace evolution that is driven by a mnemonic, brand-name-like dynamic.
gTLDs are also inherently transnational in orientation and administration. They are not tied to a particular political jurisdiction. Given the ease with which bits move around the world, an easily remembered, established brand identity in URL-space has much greater economic value if it is international than if it is purely local or national. This is true even if the content or function of a domain pertains to a specific country. A travel agent that specializes in tours of Africa would almost certainly prefer "Africa.tour" to "companyname.co.bw."
At best, gTLDs encourage the creation of a unique labeling semantic in which the different levels of the naming hierarchy work together to produce concise but descriptive and memorable labels. Internet-savvy firms have already begun to exploit the potential of the SLD, inventing punchy names such as "geocities," "media100," and "biomorph." The extension of this process to top-level and third-level domains adds value. In the future, networking-oriented organizations may build their corporate identity around domain names, rather than the other way around. Something similar happened to telephone numbers. As a way of navigating the telephone network, the original North American numbering plan had its own distinct syntax. 1-800-COLLECT or 1-800-CALL-ATT, however, represent the subordination of engineering semantics to marketing semantics. The entire string becomes a holistic label rather than an abstract syntax of unconnected hierarchical pointers. This process works best when the naming process is completely free of the constraints of naming structures external to the Internet.
Domain naming that starts with country codes at the top level is driven by a markedly different set of political and economic imperatives, and creates in users and providers alike profoundly different expectations than naming that starts with gTLDs. One of the chief advantages of country codes is that they provide a capacious, exhaustive, and mutually exclusive categorization of Internet top-level domains. It is a categorization principle, moreover, that people can readily grasp; that is, nearly all people understand that the planet is divided into nation-states and they know the names of many of them. This advantage is severely limited, however, by the ISO 3166 codes, which are often counterintuitive and confusing. Is ".au" Austria or Australia? Is ".il" Iceland, Ireland, or Israel? Exactly why is Switzerland ".ch" and why does the United Kingdom control both ".uk" and ".gb"? ISO-3166 codes are only marginally better than numbers--and the purpose of the DNS is to get away from numbers.
Putting aside the utility of the ISO-3166 codes for the time being, starting TLDs with country codes leads to a very different naming structure and expectation. Each country TLD registry has an incentive to recreate, within its domain, its own set of generic second-level and perhaps even third-level domains. Because it is usually governmental or academic/research organizations that administer these TLDs, these second-level generic domains tend to emphasize geographic and political categories. Another possibility is that they will mirror the categorization scheme of the existing gTLDs--a ".com" for commercial entities, ".edu" for educational organizations, ".gov" for government organizations, and so on. This has the advantage of standardization, making it easier for users to navigate. It is equally possible, however, that each country will devise its own unique set of SLDs. Although it is possible and even likely that these categorization schemes will be functionally as good as or better than the old IANA gTLDs, they will have to be decoded by users, or else treated as arbitrary data to be remembered. Users may have to contend with as many as 150 different second-level categorization schemes.
The problem here is that user communication patterns generally show little regard for national boundaries per se. Even when Internet traffic is primarily local or regional, the ultimate cause for the clustering has to do with content and culture, not sovereignty. It is hard to think of many cases in which a user's or content provider's primary interest is in the political sovereign of the site being accessed or published.
More importantly, ISO 3166 TLDs create an unavoidable tendency to align the administrative authority of nation-states with administrative authority over cyberspace. The country code TLD provides a powerful symbolic linkage between the "domains" of the Internet and the "domains" of political sovereignty. It is only natural that political authorities will close the loop. Proof of this is found in the IAHC decision, which explicitly recognizes "the national interests of sovereign nations in setting policies for...country code TLDs." In several developing countries, China being a notable example, there is internal contention over registrations under the ".cn" domain for precisely this reason. Country TLDs thus politicize control of registries and create the expectation that Internet policies can and should conform to national policies--in sharp contrast to the expectations associated with gTLDs.
The similarities and differences between trademarks and domain names have been noted by several writers. The IAHC clearly attached a lot of importance to reconciling domain naming with intellectual property protection.
In fact, the trademark issue is very easy to resolve--if there are no gTLDs. One could simply create a name system that reflects the jurisdictional and industrial categories of the established trademark registrations. One could also create a global trademark registry under its own TLD and require all trademark owners to register there, resolving naming conflicts across industries and jurisdictions by the assignment of random numbers or characters. It is in this sense that trademark protection represents a distinct principle of domain naming. If one wants to be truly rigorous and objective about trademark protection, one would have to reorganize the entire DNS around it.
The problem is that no one really supports this principle in its pure form. That is because a unique registration that serves as both a domain name and an unambiguous trademark claim produces ugly, complex, and hard-to-remember names. To wit:
"www.microsoft.software.computers.tm.us" or "www.united0h3x.tm.int"
No business in its right mind wants domain names like that; indeed, such names undermine one of the basic purposes of domain naming. What trademark owners really want is the best of both worlds: the legal protections against dilution, appropriation, or confusion associated with trademarks, and the simple, mnemonic names associated with gTLDs. It is this attempt to combine two distinct principles (trademark-based TLDs and gTLDs) that makes trademark protection within the DNS system such a hard problem. Neither gTLDs nor other naming structures map easily onto trademarks.
Once introduced into domain naming, the trademark principle creates its own unique set of incentives and problems. It seems clear that trademark owners have a strong incentive to establish protection claims in second-level domains that are as broad as possible and that extend across each and every gTLD. This could have some counterproductive effects. It is by no means clear that the Microsoft Corporation has a special claim to the domain name "www.microsoft.info," for example. That domain could legitimately be acquired by an independent company that specializes in providing consumers with information about Microsoft software or an investment newsletter that tracks the Microsoft Corporation's business. A user who wanted independent information about Microsoft might reasonably expect to find it at such an address, without necessarily confusing that site with one run by Microsoft itself. As long as registration under other gTLDs is available to the Microsoft Corporation, it is hard to see how its rights are violated if it does not control the domain.
As the number of gTLDs expands, trademark problems become worse in some ways but better in others. As gTLDs increase there is a greater chance that a user will not know specifically which domain a famous-name organization has registered in. A person searching for a Microsoft site cannot simply go to ".com" but may also look in ".firm" or ".inc," and so on. If a company is to preserve its exclusive control of all second-level domain names that resemble or reproduce its trademark, it must grab for registrations across almost all TLDs.
On the other hand, continued growth in the number of TLDs also undermines the trademark-like quality of a SLD registration. Indeed, the entire debate over trademark and domain names may have failed to realize how much the problem was simply a function of the artificial scarcity created by limiting most commercial registrations to the ".com" TLD. This has created unrealistic expectations regarding how much intellectual property is actually invested in second-level domain names. (More on this later.)
The IAHC plan was released 5 February 1997. The IAHC was faced with enormous political pressures and a crisis of legitimacy that required rapid action. It had to respond to specific demands from specific constituencies: users wanted more gTLDs; potential competitors were threatening to create new gTLDs; international interests wanted a seat at the decision-making table; trademark owners and the international organizations that represent them wanted to reconcile domain naming with intellectual property protection. It is not surprising, then, that the IAHC plan is a collection of discrete responses to these disparate concerns. In effect, the IAHC embraced all three principles of domain naming.
This paper will not discuss the detailed organizational arrangements of the IAHC plan. These can be obtained at http://www.iahc.org. In order to focus on TLD naming issues, the paper does not delve into several important questions, such as how many applicants, if any, will be willing to accept the IAHC's terms for registries, or the political and economic implications of the new governance structure.
The IAHC has created an additional seven generic TLDs. The newly defined gTLDs are:
In all cases but one (".web"), the IAHC appears to have deliberately avoided overlapping with the new TLDs being registered by AlterNIC and its allies. Instead of the highly desirable ".inc" or ".corp," for example, it went with the rather dull ".firm." 
The new generic TLDs will be administered and managed by multiple, competing registrars. Qualified registrars will be selected by lottery in a manner intended to assure an equitable geographic distribution of registrars. A Council of Registrars (CORE) established by a Memorandum of Understanding (CORE-MoU) signed by the registrars, will be responsible for providing common services among the registrars by operating a shared data base repository for the gTLD registries. CORE will operate as a Swiss nonprofit association.
Stewardship of the gTLD space is assigned to the gTLD DNS Policy Oversight Committee (POC). The POC basically mirrors the composition of the IAHC itself; it is composed of members named by ISOC, IANA, IAB, ITU, INTA, and WIPO, but also includes representatives from CORE. A Policy Advisory Body (PAB) is formed to obtain public and private-sector consultation and review of POC and CORE activities. Changes to policy can be initiated by POC and enabled upon the agreement of ISOC and IANA, with the review of PAB and CORE.
The IAHC plan devotes a lot of attention to trademark problems. It acknowledges that second-level domain names "are capable of infringing trademark rights." It sets a policy that second-level domain names that are "identical or closely similar to names which are internationally known, and for which demonstrable intellectual property rights exist," should only be "held by, or with the authorization of, the intellectual property owners." In order to further protect the interests of property owners, the SLD application and renewal process will include detailed contact information including submission to jurisdiction, publication of all SLD applications on a publicly accessible site, a "voluntary" 60-day pre-assignment waiting period, and permanent identification of those applicants who chose to utilize the 60-day period. The IAHC had first proposed a mandatory 60-day waiting period for all registrations, but this met with such strenuous opposition from commentors that the requirement was relaxed.
Conflicts over SLD allocation on the basis of trademark will be adjudicated by "administrative domain name challenge panels" made up of "international experts in the fields of intellectual property and Internet domain names." The procedures for creating the panels and for bringing challenges before the panels will be administered by WIPO's Arbitration and Mediation Center in Geneva. Challenges will be heard online wherever possible. Challenges initiated within 60 days of the registration of the domain name in question will be concluded within 30 days of the initiation of the challenge. Both challenges and proposed decisions will be publicized on the Internet, and there will be a reasonable period for submission of comments by appropriate third parties (including relevant governmental or regional authorities) before a final decision is made.
An administrative domain name challenge panel would be able to decide that a challenged SLD should be excluded from the gTLD in which it was registered and, in appropriate and exceptional cases, from some or all other gTLDs that are parties to the gTLD-MOU. The panel has also been given the power to exclude SLD names that are "closely similar to the challenged domain name." Exclusions can be appealed, upon petition by a third party, to an administrative domain name challenge panel.
The IAHC plan also creates a special second-level domain ".tm" under the ".int" TLD for the registration of trademark-specific domain names, and recommends that each country-code TLD adopt the same practice.
Given the clear differences in the TLD principles described above, where exactly did the IAHC decision put us? In general, the plan is a partial victory for the gTLD principle, but it also contains a number of inconsistencies. Although it created seven new gTLDs, it also explicitly recognized "the national interests of sovereign nations in setting policies for...country code TLDs." Isn't it possible that nearly tripling the number of gTLDs and globally distributing the registries will undermine the incentives of users to register under country code TLDs? The IAHC does not seem to have explicitly considered this question. Even more confusing, the IAHC draft scolded users based in the United States for their "inadequate use of the .us TLD" and for the absence of functional SLDs under the ".us" TLD. It is obvious to anyone familiar with the Internet that the cause of both of these phenomena is the preference of American Internet users for gTLDs over country-code TLDs. In creating new gTLDs, the IAHC recognized and encouraged this demand; in all likelihood the effect will be to further diminish interest in the ".us" TLD. Why, then, did the IAHC document make such a big fuss about the ".us" TLD?
The IAHC's procedural recommendations went to great lengths to incorporate a trademark review process into the assignment of SLDs under the new gTLDs. Indeed, it has virtually institutionalized the international intellectual property establishment's input into domain name assignments. To be charitable, this is not necessarily another inconsistency. It may be that a rigorous trademark review process is required if the way is to be cleared for the expansion of gTLDs and their eventual emergence as the dominant DNS principle. But the inherent tension between gTLDs and trademark protection was noted above. The intrusion of the trademark principle could undermine the integrity and utility of gTLDs in various ways. It could, for example, set off a mad rush on the part of trademark owners to register any and every trademark-related SLD in every gTLD in order to establish property rights in domain names. It could encourage corporations with famous names to challenge any registrations that might possibly be confused with their trademarked names. If NSI is brought into the MoU it could mean that existing registrations under ".com" would have to be renewed annually under the new procedures. This could reopen a number of trademark issues that have been festering in that domain. It could also raise thorny jurisdictional issues, as companies registering SLDs are asked to submit to binding arbitration. Losers may attempt to mount challenges in national court systems.
The IAHC's proposal to create SLDs devoted to trademark registration in the ".int" TLD seems even more confusing. In effect, the IAHC proposes to give trademark owners the opportunity to register a third-level domain in an obscure SLD under a little-utilized TLD category that few users know about or understand the meaning of. Faced with the alternative of doing this or of pursuing trademark rights in the SLDs of the more commercially appealing gTLDs, there is no doubt which one companies will choose. Furthermore, the IAHC makes it clear that whether trademark owners register in these specialized domains or not has virtually no legal or economic consequences:
there would be no obligation on the part of any trademark owner to register in any of the trademark-specific domain name spaces, or to be listed in any associated trademark-domain name directory. In addition, there would be no negative legal consequences to a trademark owner for not having a trademark-specific domain name or not being listed in a trademark-domain name directory. In particular, the existence of a trademark-specific domain name space does not imply that trademark rights in other top-level domain name spaces are negatively affected in any way.
Thus it is hard to understand what the purpose of this aspect of the plan is. In this writer's opinion the IAHC--both in the content of its plan and in its composition--gives far too much weight to established intellectual property claims in the domain administration process.
The intellectual property issues created by domain names appear to fall into one of the following three categories:
In this author's opinion, intellectual property law has no business intervening in conflicts of the first category. To do so would be to massively expand the scope of intellectual property protection and give multinational corporations an unfair advantage over smaller organizations in the establishment of an Internet presence. Small organizations and businesses have every right to extend their company names onto the Web, particularly when generic terms are involved. If they are nimbler than large corporations and register desirable names first, so be it. (Chances are they will not be faster, anyway.) Multinationals should have to pay the market price for excluding others from the domain name.
Cases of type 2 are not as clear-cut as the IAHC makes them out to be (it refers to name speculation as "extortion"). As long as (a) multiple TLDs are available in which to register, (b) fees are charged for registration and renewal, and (c) a "use-it-or-lose it" policy is in force, the damage done by such speculation is minimal. Name speculation was a serious problem only because of the unique expectations and artificial scarcity associated with the ".com" gTLD. Combating it does not warrant holding the entire DNS hostage to trademark concerns.
In both type 1 and 2 cases, the expansion of generic TLDs mitigates the problem somewhat. It does so not only by creating additional room for specific names, but also by reducing the expectations of users that a famous name in an SLD necessarily correlates with a particular company.
The only cases where intellectual property law clearly is appropriate are those of type 3, in which users are deceived or name holders are damaged by confusion or misappropriation of names. But IAHC's elaborate arbitration and review process is not necessary to prevent this. A general rule prohibiting it and imposing some kind of penalty on the perpetrators would probably suffice.
Unfortunately, the IAHC seems to have a much stronger form of intellectual property protection in mind during the allocation of SLDs. This may be an incorrect conclusion, but the report itself tells us very little about what criteria will be used to adjudicate challenges to second-level domain names. Thus we do not know yet which principle will trump the other.
The most important aspect of the IAHC plan is that it expands the number of gTLDs. A careful analysis of the TLD problem from the standpoint of an Internet user leaves little room for doubt that all TLDs should be gTLDs.
Domain naming is justified only insofar as it facilitates the ability of users to remember and key in Internet addresses easily, and the ability of system administrators to maintain connectivity efficiently. It follows that TLDs are most useful when they provide a relatively simple and intuitively meaningful first-level categorization of the Internet world. gTLDs are most in accord with the transnational character of the Internet. Furthermore, it is easier for a gTLD-based system to subsume other TLD principles than the other way around; that is, there can be a gTLD devoted to trademark registrations (".tm") or to information about national territories (".nat") or regions (".asia").
The author agrees with the arguments of Johnson and Post  that the Internet ought not to be subjected to national jurisdictions; cyberspace should evolve its own form of jurisdiction. Any attempt to impose territorial legal and political requirements upon a transnational medium is bound to be crippling and regressive.
Perusing the newdom list makes it clear that there is plenty of room for innovation in the area of TLDs. IAHC's choices of new gTLDs seem rather stilted in comparison to those being pioneered by AlterNIC and its allies. There are many entrepreneurs with excellent ideas for new TLDs. They may or may not meet a real market demand. But it is to be fervently hoped that the new administrative structure established by the IAHC will find a way to incorporate this entrepreneurial energy rather than shutting it out. The Internet's next step should be to find a suitable way to allow innovation and unrestricted entry in TLD creation and administration.
 See the IAHC Web site, http://www.iahc.org/dns-refs/, and the Harvard conferences, http://ksgwww.harvard.edu/iip/caidraft.html for documents and papers on DNS.
 Without a hierarchy of domain names, the system would have to maintain a lookup table with the IP address of every individual host computer connected to the Internet and distribute this table constantly to each computer on the system. Such a table would be extremely burdensome to system memory and bandwidth and, at the Internet's current rate of growth (approximately 10,000 new hosts every day), impossible to keep current.
 "The administration of a domain requires controlling the assignment of names within that domain and providing access to the names and name related information (such as addresses) to users both inside and outside the domain." [RFC 920] J. Postel & J. Reynolds, "Domain Requirements," October 1984.
 [STD-13] P. Mockapetris, "Domain Names: Concepts and Facilities," ISI, November 1987. See also RFC 920.
 [RFC-799] D. Mills, "Internet Name Domains," RFC-799, COMSAT, September 1981.
 "[A domain] must be responsibly managed. There must be a responsible person to serve as an authoritative coordinator for domain related questions. There must be a robust domain name lookup service, it must be of at least a minimum size, and the domain must be registered with the central domain administrator (the Network Information Center (NIC) Domain Registrar)." [RFC 920]. Note that the DNS system still required a centralized administration point--an authoritative place where domains can be defined and registered. In the content of the early Internet, when the Net was unquestionably a U.S.-government-funded enterprise, the identity and legitimacy of this "central domain registrar" was not at all problematic.
 The document stated that "most of the top level names will be very general categories like 'government', 'education', or 'commercial'. The motivation is to provide an organization name that is free of undesirable semantics." [RFC 920], p. 2.
 Jon Postel, e-mail communication with author, 4 February 1997.
 Microsoft had to purchase the domain name for "Slate" for US$10,000, for example. See Shaw, 1996, for a discussion of these issues.
 See Robert Shaw, "Internet Domain Names--Whose Domain Is This?" [Shaw 1996] http://ksgwww.harvard.edu/iip/caidraft.html, for several examples.
 Carl Oppedahl, "Analysis and Suggestions regarding NSI domain name and trademark dispute policy." 1996. http://ksgwww.harvard.edu/iip/caidraft.html.
 http://www.alternic.net. See also http://www.mcs.net.
 Bill Frezza, "Top-level Domain by Any Other Name," Network Computing Online, 24 January 1997. http://www.techweb.cmp.com/nc/802/802colfrezza.html.
 Postel draft, February 1996. ISOC is to begin issuing up to 50 new registry licenses in early 1997 in exchange for fees and yearly payments. In turn, these new registries will manage up to 150 new international top-level domains (iTLDs) such as .www, .inc, and .ltd, charging users a fee to register within a specific domain.
 Robert Shaw, an adviser at the International Telecommunication Union, said of the Postel draft "Personally, I think this is totally without any legitimacy. ... The domain name server space is a global resource and neither IANA nor ISOC have the right to appoint themselves as the taxing authority for the root level to the tune of potentially millions of dollars a year." Quoted in Ken Hart, "Net plan masks power bid," CommunicationsWeek International, Issue 170 (9 September 1996). (Interestingly, no one asked Shaw about what gives ITU the right to charge thousands of dollars for publishing telecommunications standards.)
 Bill Frezza, "The 'Net governance cartel begins to crumble." CommunicationsWeek 635 (28 October 1996).
 E.g., Agmon, Halpern, and Harper at http://www.law.georgetown.edu/lc/internic/domain1.html.
 This aspect of the decision is hard to interpret. One would think that IAHC would avoid overlapping altogether, or overlap completely in order to drive AlterNIC out of business. Duplicate use of the same TLD name by competing nameservers threatens to fragment the connectivity of the Internet. However, this has happened before. For a detailed historical analysis of the period of competing, nonconnected telephone systems in the USA, see Milton Mueller, Universal Service: Competition, Interconnection, and Monopoly in the Making of the American Telephone System, MIT Press/AEI Series on Telecommunications Deregulation, 1997.
 Stanley Kaplan is a college test preparation service. In this case its chief rival, Princeton Review, registered the domain name "kaplan.com" and used the site to publish negative information about Kaplan's service.
 David R. Johnson and David G. Post, "And how shall the net be governed? A meditation on the relative virtues of decentralized, emergent law." Cyberspace Law Institute, http://www.cli.org/emdraft.html.