ISPCP Position on New gTLD Expansion ==================================== Introduction ------------ This paper reflects the position of the ISP and Connectivity Provider Constituency (ISPCP) within the Generic Names Supporting Organization on new gTLD Expansion. The Constituency has reflected on this issue in April of 2005 and published a joint paper with two other constituencies within the gNSO (the Commercial and Business Users’ Constituency and the Intellectual Property Interests Constituency). In December of 2005 and January of 2006 the ISPCP once again considered the issue. Fundamentally, the ISPCP believes that the mechanism for recognizing and implementing new, generic Top Level Domains is adequate. However, our constituency believes that there needs to be an informed discussion on how to decide when new gTLDs are needed. The ISPCP uses this occasion to reiterate its support of the existing strategy for new gTLD implementation and provide guidelines for establishing the need for new gTLDs. On New gTLDs ------------ The expansion of the gTLD namespace is not a self-evident requirement. Nothing in the ICANN mission, nor in the Memorandum of Understanding with the U.S. Department of Commerce requires that the gTLD namespace be expanded. Instead, the ICANN by-laws simply require the organization to use market-driven mechanisms to promote and sustain a competitive environment. The by-laws also suggest that the introduction and promotion of competition should be done in the public interest. As a result, the introduction of new gTLDs should never be solely done to create or recognize business opportunities as the by-laws indicate that new gTLDs should only be introduced because there is a strong public interest in the competitive benefits of different types of gTLDs. In addition to ICANN’s by laws, other foundation documents for our organization also provide support for a reasoned, deliberate approach to introduction of new gTLDs. The United States government, which is referenced in ICANN’s own strategy for the recognition of new gTLDs, does not require ICANN to recognize any gTLDs. In fact, while the MoU does require ICANN to have a strategy in place for the recognition of new gTLDs, it does not require the implementation of any specific number of gTLDs nor does it have a timetable for introduction of new gTLDs. Again this supports the adoption of a measured approach. Does Competition Exist? ----------------------- If the driving need in the gTLD namespace is to provide market-driven mechanisms for a competitive environment in the public interest, then how can one determine if the competition exists? Competition Amongst Registrars ------------------------------ The ISPCP believes that competition amongst registrars exists and flourishes. The model of having multiple registrars compete within given TLDs with an underlying single source registry has been an ICANN success. More than 500 accredited registrars now exist and competition on price, service and other differentiators flourishes. In fact, the competition in some parts of the name space is so effective that ICANN has had to make equitable treatment, contract compliance and inter registrar activity monitoring part of its strategic plan for the coming years. Competition Amongst Registries ------------------------------ The registry/registrar model gives each registry a natural monopoly. As a result, competition at the registry level is achieved when three conditions are met: 1. there are no unjustified barriers to entry to the market for registry services resulting in a level-playing field for market access leading to a plural supply base; 2. there is no market distorting supplier dominance which prevents registrants (users of the gTLDs) from having a fair share of any benefit; and, 3. market access brings new name types which add value in the eyes of registrants. The earliest attempt to introduce new gTLDs was a "proof of concept" round which included both open gTLDs (similar to .com) and restricted gTLDs (similar to .int). One goal of the "proof of concept" round was to introduce new gTLDs that would act as competition to open TLDs such as .com, .net and .org. The new, sponsored gTLDs were an attempt to introduce a new name type into the namespace. By any reasonable measure -- market share, number of names registered, percentage growth -- the introduction of open, unsponsored names failed to add any competition at the registry level. With five years of experience it seems to the ISP community that their failure was the result of being undifferentiated to existing market offerings -- a sort of "me-too" style of competition where the new names offered no unique value to registrants. The small number of sponsored gTLDs in the "proof of concept" faced a different problem. The early sponsored gTLDs were from tiny slices of the marketplace with minimal customer populations. In these small markets, the sponsored gTLDs have not made any significant impact in the area of competition. Evaluation of Competition ------------------------- In its April 2005 statement to the ICANN Board of Directors, the ISPCP listed five key results of the attempt to create registry-level competition. It may be confusing to consider the concept of registry level competition when discussing new top-level domains. First, a single vendor dominates the existing landscape for registry activity. Second, registries have a natural monopoly over their existing domain spaces. The ISPCP looks at registry competition as competition in the expansion of the name space. In this way, we assert that there must be competition at both the registrar and registry level. We continue to believe that these five points are crucial in understanding our constituency’s position that registry-levelcompetition does not yet exist. --> 80% of registrants were not new to the marketplace. 20% of the respondents in surveys were new to the DNS whereas 80% were not. The .name TLD had the highest proportion of new registrants at 44%. The .biz and .info TLDs had the lowest number of new registrants at 16% and 14% respectively according to the Summit Strategies Incorporated survey. --> Many registrations were "defensive" registrations to prevent bad faith use by others. Overall 41% of the organizations or individuals surveyed by Summit Strategies Incorporated indicated that their registration was defensive. This is further broken down by TLD: 52% in .biz 41% in .info and 34% in .name. --> Most of those who used the new names thought of them as a second choice. Very few people were using the new gTLDs unless their first choice was unavailable: 18% in .biz, 17% in .info and 8% in .name. --> "Me-too" competition is not working. There was a recognisable drop in .com, .net and .org registrations in the first twelve months that the new gTLDs were available. However, the market recovered very quickly and market dominance by .com, .org and ccTLDs continues to this day. --> ICANN’s "proof of concept" approach led to consumer confusion. According to the Summit Strategies Incorporated survey: "There were reports of significant confusion among actual and potential registrants, registrars and the broader community following launch of the new gTLDs as a whole. Some confusion flowed from the nature of the "proof of concept" idea, which was to try different start-up mechanisms and see which of them worked best. Registrars and consumers therefore had to learn about not just one new mechanism, but several, and at nearly the same time." Five Crucial Principles ----------------------- The ISPCP believes that, if there are to be any expansions of the TLD namespace, any new gTLDs should create value for potential registrants. Registrants will perceive this value, if it exists and create user demand. We believe there are five principles that should govern and determine all future expansion: 1. Competition. Any new gTLD must create value-added competition amongst the registry community. In it not enough to simply provide interesting alternatives. New gTLDs should make the domain name system more useful and more accessible to broader communities of interest and a wider community of end users. A name which seems to simply supply "me-too" alternatives should be avoided. 2. Clear differentiation. Any new gTLD must provide clear differentiation from other gTLDs. When gTLD sponsors suggest a name whose differentiation is evident, that should be enough for the application. Success in achieving differentiation in the marketplace would be a function of the success of its business model. In addition the potential for cyber-squatting, fraud and abuse is also limited wuth sTLDs. 3. Good faith. Any new gTLD must avoid increasing opportunities for bad faith entities who which to defraud users. Any new gTLD proposal should avoid names that have the potential to confuse end-users because they are lexically similar to existing gTLDs. In addition, no new gTLD should be allowed that confuses the new gTLD with popular marketing terminology, brand names or trademarks. 4. Diversity. Any new gTLD must be able to serve both commercial and noncommercial users. 5. Certainty. A new gTLD must propose names that assist the Internet end-user in determining the relationship between the name and its purpose. Obscure names are not necessarily to be avoided: if the name has meaning and value to a delimited part of the marketplace, this relevance should be sufficient to meet the principle of gTLD certainty. The semantics of the name is just as important as the relationship between the name and its perceived purpose. New gTLDs -- How to Move Forward -------------------------------- The ISPCP believes that no new gTLDs should be introduced unless they can be shown to add value and competition while promoting the public interest of the name space. How should ICANN decide if a new gTLD proposal meets this requirement? - Any new gTLD proposal should be sponsored. - Any new gTLD must provide added value to the namespace -- not simply alternatives to existing, generic names. - Any new gTLD proposal should adhere to the principles of competition, clear differentiation, good faith, diversity and certainty. - Any new gTLD proposal must support Internationalized Domain Names at the econd and subsequent levels - consistent with the IDN architecture standardized by the IETF. In the standard Internationalized Domain Nname Architecture, the term internationalized domain name means specifically any domain name consisting only of labels to which the IDNA ToASCII algorithm can be successfully applied. ToASCII is based on the Punycode ASCII encoding of nameprepped (normalized) Unicode strings. ICANN issued guidelines for implementation and use of IDNA in July of 2003. - Any new gTLD must be shown to have support in the marketplace and provide a - demonstrable public benefit to the DNS namespace. ICANN should move forward cautiously: ensuring that no damage is done to the namespace and no problems are created for users of the namespace. ICANN should allow for proposals for sponsored names to come forward on an annual basis and provide a reliable process for judging the proposals against the criteria set forth above. Evaluators of the proposals should err on the side of caution: the bar for introduction of new gTLDs should be set high. Should an Auction Model be Used? -------------------------------- The ISPCP believes that an auction model is a very bad idea. An auction model for new gTLDs ends up selling the monopoly rights to a high bidder and then hopes that they have the wherewithal to provide good service. Unfortunately, it is not clear how the names to be auctioned would be determined. Instead of using the market to drive the introduction of names, an auction puts the determination of the names into the arbitrary hands of a few entities. Without the principles of differentiation, certainty and good faith, an auction model has no inherent ability to add value in the public interest. Finally, and perhaps most importantly, an auction model introduces the potential for market distortion by dominant or desperate businesses. As with bids for spectrum for 3G mobile applications, experienced organizations can sometimes grossly overbid on "potential" in an auction model. It would certainly be possible for a winner to overvalue a potential name and then never be able to economically bring it to the market. In addition, an auction tends to favor large, dominant players with significant capital backing. Given the current situation in the marketplace (84% marketshare by a single company) an auction could easily be dominated by existing players. Should the Marketplace Simply be Thrown Open? --------------------------------------------- The ISPCP believes that simply opening the name space to all and any new ideas is a very bad idea. There is often a proposal made that suggests allowing minimal interference in the introduction of top level domains. Many, many domains should be allowed to be introduced. The market will determine which will succeed and which will fail. Under such a proposal any applicant for a new gTLD would have to meet a very minimal set of criteria to establish a new gTLD under an existing registry. The implications of such a proposal are that many names would be introduced and that some would succeed. However, some would also fail. This is the core reason not to proceed with an open marketplace. The ISPCP community believes the expectation of significant registry failures, with no mechanism of safeguards is against the public interest and goes against ICANN’s mission and contrary to its core values. Any possible benefit that a laissez-faire approach might have is outweighed by the disproportionate cost due to the increased likelihood of multiple registry failures. -- Submitted on Behalf of the ISPCP Constituency Generic Names Supporting Organization Mark McFadden Secretariat, ISPCP Constituency Generic Names Supporting Organization Internet Corporation for Assigned Names and Numbers