An advantage of this approach would be that registrar competition could be implemented quickly using existing registries (where these are suitable for allowing access to competitive registrars), before exact technical standards, protocols etc for registries are finalised. (Unless the Panel believes it can make comprehensive and specific recommendations in these areas over the next few months.) At the international level, ICANN implemented registrar competition more than a year before proceeding to start any registry competition. This gTLD registrar competition is widely viewed as a success, while ICANN moves towards registry competition have been very controversial and have not yet borne any fruit on the ground. There would be advantages for stability in evolving one or more existing *.au registries rather than implementing untested registries, especially if other changes are happening in the .au space at the same time. With this approach it would be best to move zone file maintenance from registrars to registries (in cases like com.au and org.au) or to some independent party. If access to the "autodns" software owned by auDA (or similar automated software) can be arranged, then this should be cheap and easy, as demonstrated in gov.au and edu.au. One other factor which would need to be seriously considered is having the policy check function separated from the registry. For a registry like AUNIC which is designed to have the policy check done by an external entity (the registrar), this would not add any problems. In the case of AUNIC (with which I am familiar through doing volunteer work throughout last year) the costs of running the rest of the registry (even if it included zone file maintenance for other 2LDs with autodns, and even if it was operated on a professional basis) would still be significantly less than the cost of doing policy checks. Because of the high level of automation used, the AUNIC registry without policy checks could be run professionally for less than the Verisign gTLD registry cost per domain. The policy check function could easily cost several times this amount, depending on the policy automatability. Having the policy checks performed by a separate organisation would give a clear separation between policy setting and operations (policy implementation), and much more transparency on domain name pricing. There could easily be a separate policy check organisation for each 2LD. < The policy check organisation would not be required to have technical knowledge. They would just need to look at each domain application and decide whether to accept or reject it with any appropriate reasons. In analysing the competition impact of multiple registries, we need to consider what choices are made. Because each 2LD is tied to a particular registry, the important choice is the choice of 2LD by the registrant. The factors involved in this choice would be: The name policy of the 2LD. Whether their preferred registrar (eg: ISP) supports that 2LD. A preferred 2LD name for the registrant (for example, BHP is unlikely to want bhp.org.au or bhp.asn.au). The price charged by the registrar for that 2LD (which will include the 2LD registry price). Expected time delays for particular 2LDs (this would generally by dominated by the policy check time). The amount of advertising undertaken by the registry. Additional services provided by the registry. While it is possible for a registrant to make a decision based on factors like price and service of the various registries, it is likely that most registrants will make a decision on which 2LD they prefer, unless there are significant other factors.
This means that the price sensivity for registry services is likely to be small, and so there will be little competitive pressure on registries to contain costs. Without this kind of competitive price pressure on registries, it is hard to see how registry costs can be contained, other than having the price charged by registries to registrars fixed in the tender process. auDA could favour registries which propose a lower price in their tender (or perhaps a lower price cap, so registry fees can easily be reduced). In addition, once an organisation has chosen a domain name, the costs of switching this can be very high (depending on how widely the domain is used or known). For all but the very smallest organisations, this cost would be much higher than the cost of maintaining the same domain name (unless the cost of renewing got into thousands of dollars). So the only effective competition is at the time of registration, and there are many factors other than registry price and service on which a registry will be chosen (through the registrants choice of 2LD). This becomes even less significant if the registry is low cost and fully automated. This means that to a significant extent registries will be parallel monopolies (regardless of how many registries exist or whether 2LDs are created), rather than being in vigorous competition with each other. It is like the water supply market in Melbourne. While there are multiple water companies, each has a monopoly in a particular area, So if you wanted to change water companies, you would need to move house. Doubling or trebling the number of water supply companies (3) would not change the parallel monopoly nature of the industry. In the DNS space, it is much easier for registrants to choose registrars without having to consider other factors, such as choice of domain name. Because all registrants who want to use a particular 2LD must use that 2LD registry (they have no choice), it means that all those registrants must pay for additional services provided by the 2LD registry, regardless of whether they want them or not. If additional services (such as digital certificates) are made available at the registrar level, then registrants (end customers) will have more choice about whether they wish to pay for those additional services or not. (Similarly, end users must ultimately pay for registry advertising.) Given the number of registrants who write to AUNIC to say they have lost their AUNIC registry key and cannot recover it using automated systems, its not clear whether adding mandatory digital certificates at the registry level would be a good idea. It would require a careful analysis of what happens when a certificate is lost, or the person who has the certificate leaves the organisation, or in cases of name changes, mergers and organisational splits. Manual key recovery is one of the more substantial costs of operating the AUNIC registry at present (in terms of time spent), so these issues are worth putting thought into. On point 4.3.10, I would suggest there is a third way to ensure prices bear some relation to costs: Have the registry run by a suitable non-profit organisation, and operated in the public interest. I am not convinced that it is necessary for auDA to regulate service levels at the registrar level. If there is a competitive market, then it should be able to sort out service levels. As with any other competitive market, if registrars are not providing adequate service, then they will lose business. It should be possible for registrars to offer a "slow and cheap" option, as long as customers know what they are signing up for. So auDA's role could be just to ensure that registrars advise customers of the service levels provided by that registrar. There are some other issues I was hoping to see canvassed in this report. For example: How would expiry of domain names be done? Would the registry or registrar be responsible for expiry of domain names? What are all the things registrars are responsible for? (Mainly in an on-going sense, after registration time.) What happens to these responsibilities if a registrar ceases operations? Would there be common authentication tokens (like AUNIC registry keys) across the different registries and registrars?