Registrations in Open ccTLDs

[ Overview - Results - Conclusions & Implications - Motivation ]

Abstract: The author analyzes domain registrations and usage in the "open ccTLDs" of .CC, .TV, and .WS in order to quantify the domains' size, usage, and registration patterns. Analysis first reports the number of open ccTLD web pages indexed by Google, finding open ccTLDs to be less than one one-hundredth as large as .COM when measured in this way. The author next considers registration of commonly-used dictionary nouns; while many such domains have been registered in the tested ccTLDs, more than 80% lead only to placeholders or to no web content at all. Finally, the author investigates open ccTLD registrations that use the same second-level strings as the primary .COM domains of major corporations worldwide (including the Fortune 1000 and Forbes International 500). Again, many such domains have been registered, and more than two thirds of such registrations are by registrants other than the registrant of the corresponding .COM. The web content available on sampled domains provides evidence both of substantial defensive registrations and of substantial cybersquatting.

Overview

Seeing the growth of COM, NET, and ORG, certain country-code top-level domains (ccTLDs) have decided to open their name spaces to all interested registrants, regardless of country. These domains are often referred to as "open ccTLDs" as distinguished from those "closed" ccTLDs that limit restriction to citizens or firms of their respective countries. Open ccTLDs include the Cocos Islands' .CC, Tuvalu's .TV, and Western Samoa's .WS; in their respective branding efforts, the .TV and .WS registries suggest that their ccTLD codes stand for television and web sites, respectively. With short, mnemonic TLD strings, these registries suggest that their ccTLDs make good alternatives to registration in the traditional global TLDs (gTLDs) of COM, NET, and ORG.

Recent ICANN efforts have established certain standards for gTLDs; among requirements are data escrow and disclosure, service level agreements, and dispute resolution systems. However, within ICANN's contractual framework, ccTLDs remain notably distinct from gTLDs; open ccTLDs thereby avoid many ICANN requirements even as they provide services generally similar to gTLDs. Various firms including Verisign and other gTLD registries have on multiple occasions called for ccTLDs to be subject to same regulations as .COM, .NET, and .ORG; see, for example comments in a November 1999 Public Forum and 2002 gTLD Constituency comments (PDF) to ICANN's Committee on Evolution and Reform. Recent ICANN policies give a nod to this suggestion, as in a budget framework that attempts to normalize contributions across gTLDs versus ccTLDs, subject of course to difficulties including ICANN's ongoing lack of contractual agreements with most ccTLD operators.

In current research, the author seeks to examine registrations in the open ccTLDs of .CC, .TV, and .WS. The author would ordinarily begin such analysis with a tabulation and random sampling of all domains registered in each open ccTLD; however, the author was unable to obtain such a listing (i.e. zone files) since the respective registries were unwilling to provide this data when it was requested by the author and since no ICANN or other policy obliges them to do so. Instead, analysis begins with an investigation of the total volume of .CC, .TV, and .WS content indexed by leading search engine Google; even the largest, .TV, is found to be less than one one-hundredth as large as .COM.

The author next examines registration and use of .CC, .TV, and .WS domains that take as their second-level strings certain commonly-used nouns, as reported by Princeton University's WordNet. Roughly one third of such domains are found to be registered in .CC, .TV, and .WS, while .COM, .NET, and .ORG registration rates among these common nouns reach as high as 98%. In addition, the rate of usage of these .CC, .TV, and .WS domains is substantially lower than in gTLDs, suggesting a higher rate of speculative registration within open ccTLDs than within gTLDs. A few registrants are found to register large numbers of common noun domains in open ccTLDs, further suggesting domain speculation and warehousing.

The author subsequently investigates .CC, .TV, and .WS domains that match famous .COM registrations by certain well-known firms and brands -- the Fortune 1000, Forbes International 500, and Interbrand 100. Results show substantial defensive registrations -- in which the firms holding famous .COMs also register the corresponding .CC, .TV, and/or .WS domains but fail to put these ccTLD registrations to active use. However, registrants other than .COM registrants have also registered many .CC, .TV, and .WS domains matching famous .COMs. While certain of these registrations seem to be held by registrants with legitimate interests in the names at issue, other registrants may not; some registrants are found to register many .CC, .TV, and/or .WS domains that match the brand and company names of many famous .COMs -- suggesting an attempt to resell or otherwise profit from the marks and brands of others.

Since open ccTLDs are in many respects similar to unrestricted gTLDs -- providing relatively low-cost domain registration to anyone interested, without substantive or geographic restrictions -- data about the usage of open ccTLDs may inform analysis of the likely usage of new unrestricted gTLDs. Consistent with the author's prior investigations of .BIZ, this research suggests that new unrestricted gTLDs are likely to include significant defensive registrations as well as significant speculative registrations. While analysis of the relatively new .BIZ and .INFO cannot well speak to the likely future usage of registrations in these registries, the open ccTLDs analyzed here are in each instance several years old; their rates of usage in providing web content suggest, again consistent with .BIZ research to date, that new unrestricted gTLDs may reach only low rates of usage (relative to .COM, .NET, and .ORG) in providing actual web content. Of course, should the number of unrestricted gTLDs change dramatically (i.e. if many unrestricted gTLDs are created), past experience may be of uncertain value in predicting future outcomes.

Results

Results are divided into the following four sections:

• Web Pages Indexed by Google - A Comparison Across TLDs
• Common Nouns - Domain Registrations and Use, and Top Registrants
• Famous Names - Domain Registrations and Use, and Top Registrants
• Change Over Time

Web Pages Indexed by Google - A Comparison Across TLDs

Certain Google queries* can well speak to the total number of pages in a TLD that are indexed by Google. If Google indexes most web content, these results would well speak to the amount of content in each TLD; if Google indexes roughly the same proportion of content within each TLD (i.e. indexes about half of .COM and about half of .CC), then comparison of Google page counts well speaks to the relative sizes of TLDs.

The table below summarizes Google indexes of .CC, .TV, and .WS. content. .COM, .NET, and .ORG index counts are provided for comparison.

* The method of obtaining such data from Google combines a restriction to a given TLD with a request for all pages that fail to include an arbitrary search string unlikely to be present on any web page. See Dan Tobias's explanation of this method.

** enic.cc and www.tv are the home pages for the .CC and .TV domains, respectively.

Common Nouns - Domain Registrations and Use, and Top Registrants

The author obtained a listing of commonly-used English nouns from a listing posted by Bhavin Turakhia (who had used this list to analyze sunrise registrations in .INFO). Mr. Turakhia reports that he had obtained the list from Princeton University's WordNet. The author excerpted from this list the 2910 nouns found by WordNet to be the most commonly used; the resulting list is subsequently referred to as the list of "common nouns."

The table below summarizes .CC, .TV, and .WS registrations of 2910 tested common English nouns. .COM, .NET, and .ORG registration proportions are provided for comparison. The top row of the table reports the proportion of sampled domains that are registered within each TLD. The bottom row reports, within each TLD, the proportion of registered domains found to provide actual web content. Provision of actual web content excludes those domains that provide only error messages, "under construction," "coming soon," or similar; a redirect; or an offer of sale of the specified domain name. Of course, some domain names are used only for purposes other than providing web content (i.e. domains used solely for email); however, to the extent that such domains exist equally across TLDs, this method nonetheless provides a valid comparison of rates of usage among TLDs.

* Provision of actual web content means providing, on the default web page in a domain, content other than an error message; "under construction," "coming soon," or similar; a redirect; or an offer of sale of the domain name at issue.

These results show two notable trends: First, roughly one third as many generic nouns are registered in open ccTLDs as in gTLDs; this is consistent with intuition that .COM is "full" while many "good" names remain available in open ccTLDs. Second, registrations in open ccTLDs are less often put to use in distributing web content than are registrations in gTLDs, again by a ratio of approximately one to three. This may reflect that open ccTLD registrations are speculative or defensive to a greater extent than registrations in gTLDs, as analyzed below in further detail.

Separate tables report top registrants of common nouns in .CC, .TV, and .WS. The top .CC registrant is found to have registered 42 distinct common noun .CC domains (3.8% of all common noun .CC domains examined); the top .TV registrant is found to have registered 68 distinct common noun .TV domains (7.5% of all common .TV domains examined); the top .WS registrant, 20 distinct common noun .WS domains (2.4%). (Of course, it is likely that these registrants hold many domains other than the common nouns analyzed in this section. Thus, their respective total registration counts are likely larger than 42, 68, and 20, respectively.) In addition, the overwhelming majority of registrations by top registrants are not currently providing actual web content. Accordingly, it is may be the case that some or all of these registrations reflect domain speculation or warehousing.

Famous Names - Domain Registrations and Use, and Top Registrants

The author had previously obtained from Fortune, Forbes, and Interbrand their respective Fortune 1000, Forbes International 100, and 100 Most Valuable Brands listings (subsequently: "famous names"). The author removed from these listings those companies and brands without a domain name or with a primary domain name other than a .COM. For each remaining domain, the author determined whether the corresponding .CC, .TV, and .WS domains were registered; when these domains were registered, the author determined whether the registrant was the Fortune, Forbes, or Interbrand firm at issue or was instead some separate entity.

The table below summarizes .CC, .TV, and .WS registrations pooled across a sample of 1271 distinct domain names associated with famous companies and brands. The second through fourth columns report the proportion of tested domains that have the characteristic described in the first column. The final column, "at least one ccTLD," reports the proportion of second-level strings that, in at least one of .CC, .TV, and .WS, have the characteristic described in the first column.

* Provision of actual web content means providing, on the default web page in a domain, content other than an error message; "under construction," "coming soon," or similar; a redirect; an offer of sale of the domain name at issue; or the same content available at the corresponding .COM.

** Since .WS does not provide a comprehensive public WHOIS service, it was not possible to determine whether the registrant of a given .WS domain was the same registrant who holds the corresponding .COM. The .WS WHOIS service provides only registrant name but not street address, email address, phone number, DNS configuration information, or other details.

A separate table details .CC, .TV, and .WS registrations that match the .COMs used by Fortune, Forbes, and Interbrand companies and brands. This separate table reports that Interbrand domains are registered at a higher rate than are Fortune and Forbes domains; 94.9% of Interbrand names are registered in at least one open ccTLD, while only 74.3% and 75.7% of Fortune and Forbes domains are registered in at least one open ccTLD. This suggests that top brand names are particularly frequently registered in open ccTLDs. Since the proportion of Interbrand names registered by a different entity is comparable to the corresponding proportion for Forbes and Fortune names, the increased Interbrand registrations in open ccTLDs seem to represent a combination both of increased levels of defensive registration (by the Interbrand companies) and of increased levels of registration of Interbrand names by other entities.

When .CC and .TV domains are registered to the firm that holds the corresponding .COM domain, the author found no instances of provision of actual original web content (as defined above) at the .CC or .TV domain. Instead, such domains typically lead to error messages, "under construction" messages, redirects to .COM sites, or content identical to that on the corresponding .COM site. Such .CC and .TV registrations are likely defensive; on this basis, the author therefore estimates that 27.0% of .CC registrations and 35.9% of .TV registrations are defensive. .WS registrations cannot be analyzed in this way because .WS does not provide a comprehensive public WHOIS service.

A total of 16 Fortune, Forbes, or Interbrand domains were found to include HTML or WHOIS text offering the purchase of their respective domains; these domains are airtours.tv, amica.tv, bausch.tv, depfa.tv, edison.tv, etrade.tv, goldenstate.ws, ipaper.tv, keybank.cc, leggmason.tv, mckesson.tv, national.ws, nordea.ws, owenscorning.tv, staples.cc, unitedauto.tv, wm.tv, and zurich.ws. In addition, one domain was found to offer sexually-explicit images; the availability of such images was not suggested by the site's domain name, nor did any warning page or disclaimer appear before the images; attempting to close the browser window yielded a popup with additional such images.

Tabulation of available WHOIS data for open ccTLD registrations showed a few registrants who have registered multiple .CC, .TV, and .WS domains associated with multiple of the Forbes, Fortune, and Interbrand firms. When a single registrant registers as many as 18 distinct such domain names -- each matching a distinct firm or brand name on a Fortune, Forbes, and/or Interbrand list -- one possible inference is that the registrant is not acting in good faith; few registrants have legitimate interests in the company names of multiple Fortune 1000 companies. Of course, there may be legitimate and legal reasons why a registrant holds such domains; for example, the registrant may be a lawyer or technical staff person holding these domains in trust for clients. Separate tables report top registrants in .CC, .TV, and .WS (with three or more registrations among the specified subset of domains), along with the number of domains and the specific domains registered by each. The top registrant in .CC registered five distinct .CC names that are, in .COM, used by Fortune, Forbes, or Interbrand firms; the top .TV registrant, 18; the top .WS registrant, 48.

Change Over Time

This report reflects analysis conducted only once, and these results do not readily speak to change over time.

Information provided by the .CC, .TV, and .WS registries provides some guidance as to the rates of growth of the respective TLDs. The .CC registry reports 300,000 .CC registrations by August 2000; 350,000 by March 2001; and "close to 400,000" by June 2001. No additional data is available after June 2001. The .TV registry reports 100,000 .TV domains registered in the first five months and another 150,000 in the next three, by February 2001; by April 2001 .TV purported to have registered more than 350,000 .TV domains. No additional data is available after April 2001. Finally, the .WS registry reports more than 100,000 .WS registrations by November 2000, after nine months of global operation; by June 2001, .WS reported more than 150,000 registrations. According to ICANN's 2002-2003 Budget, .CC includes 200,000 registrations as of January 2001 and 500,000 as of March 2002; .TV 200,000 and 215,000 respectively; .WS 5,000 and 150,000.

WHOIS data from the .CC registry is notable in part because domains pending deletion are designated as such in .CC WHOIS responses. Of the famous name registrations in .CC, 7.9% were found to be pending deletion, while 1.7% of common noun .CC registrations were found to be pending deletion. The implication of this statistic depends in part on the duration of the "pending deletion" marking; the longer the typical duration of that marking, the more domains are likely to be so designated at any instant. The implication of this statistic further depends on the likely future state of domains deleted at present -- whether such domains will quickly be reregistered by a new registrant, or whether they will remain for some time available for general registration. Since the rate of deletion seems to be larger among famous names than among common nouns, it seems that registrants of famous names are allowing their domains to lapse more than are certain other .CC registrants; this may reflect that speculative .CC registrants are learning that the corresponding .COM registrants are unwilling to pay a premium for .CC registrations.

All testing and data collection took place between July 1 and 15, 2002.

Conclusions, Implications, and Future Work

These results support the following claims:

• Substantial registrations within the .CC, .TV, and .WS ccTLDs, but limited use of these domains. The hundreds of thousands of registrations per ccTLD (according to unverified registry claims) but low rate of web site availability (at least among tested domains) suggest that while many registrations have been made within .CC, .TV, and .WS, these TLDs remain as yet not widely used.
• Substantial defensive registrations. Approximately one third of famous name .CC and .TV domains are found to be held by the same entities that registered the corresponding .COMs. When such domains are not actively used, one likely inference is that the goal of each such registration was to prevent others from registering, using, or attempting to sell these domains; in other words, such registrations are likely to be defensive. Indeed, among sampled .CC, .TV, and .WS famous name domains registered to registrants of the corresponding .COM, not a single such ccTLD domain was put to active unique use in distributing content not otherwise available, and the majority were not used at all (showing only error messages or placeholders).
• Substantial cybersquatting. Of the registered .CC and .TV famous name domains, many are registered to entities other than the respective .COM firms. Some such registrants may have legitimate rights in the respective names. However, certain registrants have registered a large numbers of such domains, suggesting bad faith in registration; for example, one .WS registrant registered a total of 48 distinct domain names each used in .COM by a Fortune, Forbes, or Interbrand firm. Other open ccTLD domain registrations (including .COM famous names) include "for sale" or similar text in WHOIS data or on default web pages.
• Substantial speculative registration and warehousing. Of famous name and common noun registrations in ccTLDs, many report that they are reserved for future use. Top registrants of common nouns in .CC, .TV, and .WS have registered as many as 50 common nouns in a single open ccTLD.

The relatively low rate of usage of open ccTLDs, as compared with .COM, .NET, and .ORG, is consistent with the claim that providers of web content see little need for additional unrestricted gTLDs. If additional unrestricted gTLDs were of interest and benefit to content providers, one might expect open ccTLDs (that have in effect become unrestricted gTLDs) to be much used in offering large amounts of web content not otherwise available (i.e. in .COM). Accordingly, the lack of such usage of open ccTLDs might be thought to suggest a relative disinterest in additional unrestricted namespace.

Analysis of common noun registration in ccTLDs shows first that far fewer common nouns are registered in open ccTLDs than in gTLDs. This result is consistent with intuition that .COM is "full" while many "good" names remain available in open ccTLDs. This may reflect that ccTLD registrations are, in general, somewhat more expensive than the competitive .COM, .NET, and .ORG space; alternatively, this may reflect that such names are thought to be less desirable by subsequent purchasers and are therefore less often purchased by speculators and warehousers. In addition, registrations in open ccTLDs are less often put to use in distributing web content than are registrations in gTLDs. This may reflect that open ccTLD registrations are speculative or defensive to a greater extent than registrations in gTLDs.

To the extent that .COM registrants failed to buy the corresponding .CC, .TV, and/or .WS domains, it may be of interest to consider reasons why they failed to do so. This outcome may reflect lack of knowledge within the relevant community; while most large corporate legal departments perhaps knew of .BIZ and .INFO before their respective rollouts in 2001, the availability of open ccTLDs was likely less well publicized and perhaps initially of uncertain importance. In addition, a lack of sunrise provisions within open ccTLDs may have failed to provide existing .COM registrants a head start over squatters, warehousers, or other companies with interests in their respective names. Finally, pricing policies in certaion open ccTLDs may influence firms' decisions not to register such domains; in particular, the .TV registry charges as much as $1 million (per year) for that registry's so-called "premium names."

To the extent that .COM registrants have failed to recover .CC, .TV, and/or .WS domains from other registrants who have registered them (especially registrants whose purpose is clearly speculative), it may be of interest to consider reasons why .COM registrants have failed to do so. Their inaction may reflect the perceived level of importance of the open gTLDs -- that, .COM registrants may think, few Internet users intensively use or default to ccTLDs, making pursuit of ccTLDs a poor use of scarce resources. Since .CC, .TV, and .WS each use the UDRP, the cost of challenging an allegedly-impermissible registration and use of a .CC, .TV, or .WS registration is unlikely to be significantly higher than the cost of a challenge of a gTLD name.

The large number of expiring registrations in .CC suggests a decrease in size of at least that ccTLD and perhaps open ccTLDs generally. One interpretation of this fact is that speculators, squatters, and warehousers may be learning that the open ccTLD market is not as profitable as they had hoped or expected; for example, they may find that .COM registrants are unwilling to pay large amounts to obtain the corresponding ccTLD domains. To investigate this effect, future work might attempt to quantify expirations more generally -- not just among the domains found to be "pending deletion" or near expiry in the current sample, but also among those allowed to lapse over a wider range of time. Such an inquiry might go beyond the Fortune, Forbes, and Interbrand domains studied here to investigate a broader sample of open ccTLD registrations.

Additional future work might attempt to investigate the following questions:

• The corresponding characteristics of additional open ccTLDs beyond .CC, .TV, and .WS.
• Trends in registration patterns by particular registrants across gTLDs and ccTLDs.
• The disposition of recently-expired open ccTLD registrations; whether domain warehousers tend to purchase such ccTLD domains as they become available.
• The extent to which service levels in open ccTLDs match or exceed ICANN's requirements for gTLDs.
• The extent to which open ccTLDs otherwise satisfy ICANN's additional requirements for gTLDs (escrow, insurance, public access to zone files, etc.).

Motivation

The purpose of this work is primarily academic -- to document the activity at issue for the benefit of those who seek to make policy decisions on related matters. For example, ICANN and the DNSO may in the future consider consensus policies that treat ccTLDs more or less like gTLDs; such policies might bind certain ccTLDs (i.e. all open ccTLDs) or all ccTLDs to the various standards adopted for gTLDs, or policies might preserve the status quo of granting relative autonomy to ccTLDs. The results documented here may be helpful in analyzing the relative merits of such policies -- evaluating the arguable need, or lack of need, of regulation by ICANN of open ccTLDs. ICANN and the DNSO may also in the future evaluate the creation of additional unrestricted gTLDs. Here too, the years of operation of open ccTLDs may prove helpful in predicting the likely futures of new unrestricted gTLDs; this research may therefore guide ICANN in creating or declining to create additional new unrestricted gTLDs.

Thanks to Martin Schwimmer for suggesting this project and providing guidance on helpful data sources.