Bruce, Could you provide the source for the conclusions you make below, or are they your own? Also, what is missing is putting it all in the context of demand (and even trying to figure that out), and in context of use. Regarding the latter, at some point .com may be less 20% but if a large portion of other gTLD registrations are only defensive and mostly point to .coms, is that a good sign of competition? Finally, I'm not sure it is a good idea for the regulator's own criteria and evaluation to be the only measure of the effectiveness of its regulations. So we should consider third party involvement as well. Tim

-------- Original Message -------- Subject: [council] Definitions and measures of competition , consumer trust, and consumer choice From: "Bruce Tonkin" <Bruce.Tonkin@melbourneit.com.au> Date: Thu, January 13, 2011 5:46 am To: <council@gnso.icann.org>

Hello All,

ICANN has committed to promoting competition, consumer trust and consumer choice in the Affirmation of Commitments.

ICANN has committed to organize a review that will examine the extent to which the introduction or expansion of gTLDs has promoted competition, consumer trust and consumer choice.

To be able to focus resources in the right places in the operating plan and strategic plan, as well as be able to measure the effectiveness of the expansion of gTLDs, it is important to agree some definitions, measures and targets for those measures in advance of the review.

With respect to measures - it is also useful if we can define some short term measures that are capable of measuring how the system is changing, and help guide improvements to that system in the short term. Longer term measures help determine whether we are meeting our strategic goals - some of these measures may take three years or more before they are can be effectively measured. In the discussion below I will suggest some short term and long term measures that could be considered by the GNSO.

Once we have definitions and measures, it would be appropriate to discuss what reasonable targets for those measures might be - which could be updated within 12 months after initial baseline measurements are made. Ideally the GNSO should focus on getting some good definitions and a few measures per definition to start with. The measures can be refined over time. Too many measures will mean that there is a lack of focus. Try to get about 5 key measures per area.

Here are some initial thoughts/examples to stimulate some discussion within the GNSO:

(1) Competition ===============

Competition can occur in various parts of the market.

(a) Competition can occur between operators of gTLDs that serve the same market - e.g a .shop and .store where the business model is to offer third party registrations in these gTLDs, or between two corporations that have competing products - e.g .brand1 versus .brand2.

A short term measure could be number of categories of TLDs and number of TLDs in each category. e.g you could use an industry standard for categories (e.g SIC codes http://en.wikipedia.org/wiki/Standard_Industrial_Classification) and classify TLDs into categories.

A longer term measure could be related more to market share of the competing parties in each category (often drawn as a pie chart) - this may not be just related to number of second level registrations (some gTLDs will have few second level names), but could be in the form of number of registered online users of that TLD (could be online customers for a product offered through that TLD). Ideally you don't want any one party with more than 20% of the market, and want more than three parties in a particular major segment.

(b) Competition can occur between domain name registry and DNS infrastructure providers - e.g between Neustar, Afilias and VeriSign that all support multiple gTLDs and ccTLDs. Many gTLD operators are likely to outsource to specialised providers.

A short term measure could be number of unique domain name registry infrastructure providers, and number of unique DNS infrastructure providers that support gTLDs. DNS infrastructure providers capable of meeting gTLD standards at the top level, will also usually offer their services to second level domain name registrants seeking a cost-effective and reliable DNS infrastructure.

A longer term measure could be related more to market share of the domain name registry and DNS infrastructure providers (often drawn as a pie chart), as well as assessing changes in price for services in the longer term. Ideally you don't want any one party with more than 20% of the market, and want more than three parties in a particular major segment. Marketshare may be in the form of number of TLDs supported by each infrastructure provider.

(c) Competition can occur between registrars that provide domain name registration services for registrants in a given market.

A short term measure could be number of unique (note many of the current registrars are owned by the same parent company) registrars in each country. Another measure could also track languages supported by the registrars - ie number of registrars for each language.

A longer term measure could be related more to market share of the domain name registrars (often drawn as a pie chart), as well as average pricing of services from that registrars for a standard bundle of services.. Ideally you don't want any one party with more than 20% of the market, and want more than three parties in a particular major segment. Marketshare may be in the form of number of second level gTLD registrations by each infrastructure provider.

(2) Consumer trust ===================

Much harder to define. I expect ALAC may be able to contribute a consumer perspective here on what consumer trust in domain names means to consumers.

(a) Consumer trust could be in the consistency of behaviour of top level domains

- ie a domain name is available nearly 100% of the time, and has consistent behaviour throughout the Internet

- could be measured in terms of uptime of TLDs, reachability of TLDs from different points on the Internet, and integrity (whether the resolution of the TLD can be tampered with). DNSSEC may assist with integrity. Not sure if there is a difference between short and long term.

(b) Consumer trust could be in the accountability of those operating TLDs or second level domains within those TLDs.

- could be measured in terms of whether TLDs are directly used for illegal purposes (e.g whether any successful legal actions taken against TLDs operators)

- could be measured in terms of amount of illegal activity carried out at the second level of new gTLDs - e.g number of successful UDRP disputes, number of successful prosecutions of second level registrants, Percentage of disputed second level domains both pre and post delegation, Percentage of malicious conduct complaints in new domains, Percentage of security breaches in new domains, Percentage of outages in new domains, Average response time to complaints in new domains

(c) Consumer trust could be in terms of whether TLD names and second level names have behaviours that are consistent with the semantic meaning of the TLD. e.g .bank relates to organisations that are official banks.

- most likely need to be measured via consumer surveys on whether consumers felt that domain names were representative of the content or purpose of the content of websites and emails associated with that name.

(3) Consumer choice

(a) Can be defined in terms of the perspective of a registrant as a subset of consumers, and whether the registrant has sufficient choice in choosing a second level domain name.

- in short term might measure in terms of number of names in new gTLDs that have a unique registrant (ie a registrant that doesn't have names in other TLDs with the same second level string) - this would show that a consumer has exercised choice in choosing their preferred domain name from range of names available. Other measures could be percentage of active use of second level domains (e.g nike.com), percentage of keyword advertising pages (e.g shoe.com), percentage of re-directs to domain names in other TLDs (e.g nike.de).

- in longer term could be measured in terms of traffic associated with second level names, and ensuring that traffic is more evenly distributed across the TLDs (ie of the top 10,000 websites by traffic - how many are associated with TLDs other than com/net/org and ccTLDs).

- in short term can also measure number of TLDs that are in new languages and new language scripts, and in the longer term measure the traffic associated with use of these names.

(b) Can be defined in terms of the choices a consumer has to use particular domain names to access Internet resources

- again this can be measured both in terms of number of languages and language scripts within TLDs in short term, as well as level of traffic on new names in the longer term.

- as per competition, a short term measure could also be number of categories of TLDs and number of TLDs in each category, and in the longer term the amount of traffic associated with these TLDs.

Regards, Bruce Tonkin