From: Tony Hill Sent: Tuesday, 27 March 2001 6:17 PM To: Jo Lim Subject: Comments on Stage 3 Report Hi Jo Following are ISOC-AU comments on the Competition Panel Stage 3 Report. A more detailed version should follow. regards, Tony Summary ======= The Internet Society of Australia (ISOC-AU) is currently preparing a response to the auDA Competition Panels first public consultation report. ISOC-AU has invited members of the Society to provide input and feedback into its response and will be forwarding the response to the Competition Panel in due course. This executive summary has been prepared in order that the Competition Panel may consider the issues likely to be addressed by ISOC-AU in its full submission in this round of public consultation. We note that the full response may show some variation with respect to this document. Competition Principles ====================== Which (if any) stages of domain name administration should be reserved to a single provider (non- contestable) and to what extent is it necessary to provide mechanisms to ensure non-discriminatory access to these stages? The registry should be non-contestable. In brief, these functions comprise: - management of the master database for.au and its subdomains - management and update of .au administrative and technical contact information (whois) - management of the corresponding public information service (whois) - provision of authoritative DNS nameserver - ensuring that the quality of data in the registry remains high (important to the long-term operation stability of the DNS) In order to ensure non-discriminatory access to these functions we believe that such functions must be the responsibility of a body that has the support of community, industry and government, has broad-based membership and has an appropriate constitution. We believe that auDA is the best organisation to meet these requirements. To what extent are industry-specific safeguards against anti-competitive practices required? We believe that in a model where the non-contestable functions are correctly identified and managed by auDA, few industry-specific safeguards against anti-competitive practice will be required. Management of the non-contestable elements should: - ensure that prices are based upon costs, and not on perceived value to consumers - be transparent with respect to costs and matters of operational importance - create an open access regime and aim to minimize downstream compliance and operational costs by ensuring consistent technical, commercial and operational standards for all .au subdomains - ensure that operational issues work in the best interests of consumers, such as ensuring a minimum of friction associated with the transfer of domains from one Internet Service Provider to another. Unless demonstrably in the public interest, management of the non-contestable elements should not creep to include functions that are contestable. We believe that the ownership structure of auDA (consumers, suppliers and industry associations) is sufficient to mitigate this risk. The Panel acknowledges that the application of the principles involves matters of opinion and judgement as to the relative costs and benefits of different models. This is particularly so when applying the principles to areas characterised by rapid change and unpredictability such as the Internet. In such circumstances, an approach of preserving options would be appropriate. We strongly support this view. auDA must be capable of responding to the challenges and opportunities that lie ahead. auDA is also a relatively immature organisation. For this reason auDA should in the short term avoid structures that limit a flexible response to future requirements. The critical issue with respect to auDA's ability to meet changing requirements is the scope of the non-contestable elements of domain name provision. It is the view of ISOC-AU that auDA should take a broad and generous view of the scope of the non-contestable functions. By adopting this approach, auDA retains the ability to devolve such functions to the market at little cost should circumstances require. If non-contestable functions the subject of for-profit market activity then auDA's ability to redress this balance may be either impossible, or may involve great expense and effort. Proposed Competition Model ========================== Resellers ========= We believe the Panel is creating undue distraction by identifying resellers of registrar services as an element of the competitive model. In doing so the Panel is anticipating the manner in which registrar services will be marketed. Such anticipation is needless. Section 4.5.1 is worthy of comment: 4.5.1 Domain name resellers operate at an intermediate level between registrar and registrant. Currently, com.au is the only 2LD with resellers. The types of entities acting as domain name resellers range from law firms to ISPs. Since the registry/registrar function is not currently separated for .com.au or net.au, it is misleading to describe .com.au as having resellers. Such "resellers" may more properly be thought of as registrars in a model which separates registry and registrar. Accordingly, Sections 4.5.2 and 4.5.3 which sets out "do's" and "don'ts" for resellers is inappropriate. It would be more appropriate to think of resellers as merely one element of the registrar operation. We recommend removing resellers from the competition model. To the extent that the Panel is concerned with potential marketing and operational practices of registrars, we recommend that the panel draft the charter for a panel to begin work on an industry code of practice relating to registrars. Beginning work on the code of practice while the Work on the Development of the code of practice is probably outside the scope of work of the Competition Panel. Registry Function and Services ============================== We bring to attention of the panel the definition outlined in http://www.aunic.net/aunic.html: A "registry" in the context of the administrative functions associated with the operation of the DNS can be regarded as a public titles database. This database is used to publicly record an association between two types of entities: a person (or incorporated body or similar corporate body) and a particular DNS name. There can only be one authoritative definition of the contents of the registry. Likewise, the administrative processes supporting the creation, modification and deletion of records in the registry must be synchronised by a single provider. We note that is desirable to ensure that a only a single set of entity records (administrative and technical contacts) is needed to register and manage names in multiple .au domains. In other words, a registrant should be able to use the same set of handles while registering or updating their .com.au, .net.au, .asn.au names. Registrants to do wish to use create and manage separate entity records for each of .com.au, .net.au, .asn.au etc. This leads us to the more general point of data quality. The registry is responsible for data quality across all .au subdomains. The NSI registry also supports the association of handles with nameservers. This desirable situation simplifies the administrative burden of registrars who change nameservers and wish such changes to propagate across all .au subdomains. Consistent application of create/update/delete processes in respect of all .au subdomains is necessary in order to ensure that registrants need only create a single set of entity records and that the quality of such records is maintained over time. Cost and price ============== Cost and price are linked to transparency. The registry is a non-contestable function. Downstream users are entitled to transparency into the cost of providing this service. Ensuring that there is transparency into the costs of registry operation is consistent with auDA's constitutional obligations. Transparency into cost is also a means of fostering trust in the overall outcomes of the competition process. Technical standards, protocols and Service Levels ================================================= The critical point with standard interfaces is not whether they come from IETF or Network Solutions. It is that the same interfaces be used to manage names in all .au subdomains. Such standard should be set by auDA. Standards are not simply technical. The commercial and business interfaces should also be standard across all subdomains. Any other course simply increases costs to registrars - an outcome we wish to avoid. We believe that the Panel has significantly downplayed the risks associated with registries also operating as registrars. Such arrangements have met with significant industry concern in the Australian telecommunication space. Access to the Australia's fixed line telephone network is one example. Amongst other things, we are concerned with the potential that consumer perceptions arising from such an arrangement may distort the competitive landscape. We encourage the panel to elaborate the risks in detail if such a course is to be considered. Pros and Cons of Proposals 4.3A and 4.3B ======================================== 1.1 Cons of Proposal 4.3A ----------------------------- ISOC-AU believes that the "cons" of proposal 4.3A presented in the Competition Panel report are poorly articulated. 1.1.1 Registrants have to manage contact information for each 2LD. -------------------------------------------------------------------- Every domain name has two fundamental contact "handles", Administrative and Technical, to which all other contact information is cross-associated. If an entity wants to register in a different 2LD, it would want to re-use the same handles, otherwise the time, effort and subsequent cost to the entity increases, but separate registries would require separate handles. Worse, every time the information changes it must be updated, further multiplying the entity's costs, so it increases the risk that the contact database becomes stale. This has undesirable operational implications for the whole Internet. In order to be useful, a public-titles database must contain accurate and up-to-date information. 1.1.2 High consumer switching costs mean basis for competition is weak ------------------------------------------------------------------------ It is generally accepted that the costs to consumers of changing domain names is high. Some reasons in brief: - changing a domain name is like changing a brand name. It is difficult, expensive, and sometimes results in long-lasting, even permanent damage to a company's market position. - The internet may contains many references to a consumer's domain name, in search engines, newspaper articles and other documents. All there references become meaningless when the consumers changes domain name. Consumers exercise choice before they register names. Once a name is registered however, the name becomes more and more valuable to the user. Ever-increasing switching costs mean that the consumer is no longer sensitive to either price or service. In the early phase of domain name penetration, the number of new registrations vastly outnumbers renewals of existing names. As the Internet matures, renewals will come to dominate new registrations. Because of the aforementioned switching costs, registries will have little motivation to keep prices low or service levels high. 1.1.3 For-profit registries will seek to move to value-based pricing ---------------------------------------------------------------------- It is reasonable to expect for-profit registries to seek to migrate to value-based pricing and charge based on switching cost, rather than the cost of provision. 1.1.4 Inappropriate Property Claims ------------------------------------- A registry may seek to establish property rights over a subdomain that it manages. The .au namespace is a public resources. Allowing the establishment of property rights over a .au subdomain is an invitation to monopolistic abuse of economic power. 1.1.5 High cost of (ongoing) oversight ---------------------------------------- Monopolies are able to impose constraints upon users that are unrelated to pricing. Techniques for exploiting non price-related market characteristics are difficult to predict in advance, especially in the ever-changing world of the internet. Enforcement of changing obligations upon an group of registry operators may be difficult and expensive. 1.1.6 Reversing a multiple-registry course will be difficult or impossible ------------------------------------------------------------------------ ---- As mentioned earlier, the critical matter at hand is the decision about the scope of the non-contestable elements of domain name provision. If auDA errs on the side of making non-contestable functions the subject of for-profit market activity, subsequently rectifying this situation may be either impossible, or may involve great expense and effort. 1.1.7 Untested model for the provision of a single point of public access ------------------------------------------------------------------------ --- The proposed model suggests a mechanism whereby a variety of registries pool registrant information in order to offer a single point of entry for public access to the database (such as whois). ISOC-AU is unaware of such a scheme in operation anywhere in the world. In the absence of operational experience, such a model presents an unknown level of risk. 1.1.8 Multiple registry model is untested ------------------------------------------- ISOC-AU is unaware of any ccTLD managing its top level domain in multiple-registry model. Selected ccTLDs appear in the following table. Country TLD subdomains single/multiple registry Canada ca top level only single France fr top level only single Germany de top level only single Ghana gh com.gh, edu.gh, org.gh single Netherlands nl top level only single New Zealand nz .co.nz, .net.nz, .org.nz single Singapore sg com.sg, net.sg, org.sg single Thailand th co.th, net.th, or.th single UK uk co.uk, ac.uk etc single gTLDs com, net, org top level only single The single registry model is clearly regarded as appropriate for top-level domains around the world. 1.1.9 Transfer of domains --------------------------- A cost-effective and efficient mechanism supporting the transfer domains from one service provider to another is essential to minimize consumer-switching costs. The soon-to-be-former service provider has no economic incentive to co-operate. Consumers must be able to initiate the transfer process by being able to contact a single Registry. When consumers switch service providers, they typically wish to take all their domains with them. If consumers need to make contact with more than one registry, switching costs rise. 1.2 Pros of Proposal 4.3A ----------------------------- We offer an alternative view of the pros of Proposal 4.3A outlined in the report. In particular, innovation at the registry is important, but needs to be balanced with the need to ensure consistency of business and technical interfaces across all .au 2LDs, in order to minimize the cost of access to registrars. Registrars are the key to a flourishing competitive domain name environment in Australia. Each registry is motivated to encourage registrars to use their subdomain and not others. Innovation becomes a tool for lock-in of registrars and this flies in the face of the spirit of open access and universal connectivity that has helped the internet flourish. Different technical and commercial interfaces for each of .com.au, .net.au, .org.au, .asn.au, etc will vastly increase the cost of becoming and remaining a registrar. Technical and commercial complexity decreases the attractiveness of becoming a registrar and reduces competition. Much of the innovation we have seen in recent years around domain names has been in marketing and the bundling of related services such as trademark servicea and company registrations. ISOC-AU believes that registrars, who are closer to the registrants than the registry, are best placed to engage in this kind of innovation. Registrars will be rewarded more highly for their innovation if they are able to offer service across *all* .au subdomains. 1.3 Cons of Proposal 4.3B ----------------------------- 1.3.1 Innovation ------------------ The registry must be motivated to innovate. Under our model the registry operates on a cost-recovery basis and is therefore without a profit motive. Such a registry might conceivably lack the motivation to innovate and may have difficulty raising capital. We believe that while present, this risk is minimal for a number of reasons: - consumers and registrars will have strong transparency into the costs of the registry - consumers and registrars are the ultimate "owners" of auDA. Together with the mediating influence of the industry associations, the owners can be expected to stimulate innovation in order to better serve their own interests.