Blog: trademark abuse

Summary – A quantitative analysis of UDRP data for all open gTLDs concludes that the introduction of new gTLDs will result in approximately 316 new cases of cybersquatting, and that the resultant cost to trademark holders, overall, will be $870,000 per year – less than less than $.10 for each trademark registered worldwide, or about $.44 per trademark registered in the United States. The data show that cybersquatting correlates to registration volume across all open gTLDs, not to the number of gTLDs, but is more prevalent in .com.

A downloadable PDF of this study is available here.

A Quantitative Analysis of Trademark Infringement and Cost to Trademark Holders in New gTLDs

Will New gTLDs Increase Cybersquatting?

A vocal group of brand and trademark owners has lobbied ICANN, the US Department of Commerce, and the ICANN’s Government Advisory Committee (GAC) claiming that new gTLDs would unleash a tide of lawlessness that would cost brand holders a fortune to combat.

Owners of trademarks and brands have claimed that a new round of gTLDs would leave them facing an onslaught of cybersquatting and typosquatting, and that their policing and enforcement costs would be substantial. Here is a recent representative statement of that view:

It is possible that the new gTLD program could lead to hundreds, if not thousands of new gTLDs. This is likely to cause brand abuse, such as cyber squatting, to grow exponentially. As a result, the legal costs for brand owners associated with monitoring, registering, and enforcing domain names are likely to raise substantially. [Source: Leo Longauer, Head of Group Intellectual Property for UBS AG]

This campaign has been so effective that even intelligent observers like David Maher, Senior VP for Policy at Public Interest Registry, accept it at face value:

… there is a connection between the creation of large numbers of new gTLDs and the public interest in preventing a vast increase in cybersquatting and the spread of fraudulent practices. [Source: CircleID]

But is it really true that new gTLDs will bring a “vast increase in cybersquatting”? A wealth of relevant data allows an empirical test of this claim. Both the World Intellectual Property Organization (WIPO) and the National Arbitration Forum (NAF) keep excellent records of claims of trademark infringement brought through ICANN’s Universal Dispute Resolution Policy (UDRP).

The two main costs of domain names for trademark holders are:

1. the cost of pre-emptively registering brands in new gTLDs
2. the cost of monitoring and enforcing trademarks in new gTLDs

This study looks at enforcement, the second category. To understand the cost of enforcement, we studied the 10-year UDRP data on existing “open” gTLDs. We used the data from WIPO and the NAF, which together comprise the overwhelming majority of UDRP cases files. The sample is statistically relevant, comprising 8 new TLDs since 2001, with over 8 million domain names under management. (The study does not include Czech ADR, ADNDRC, or cases from former providers CPR or eResolution Cases. Inclusion of these cases, which deal almost exclusively with .com, .net, and .org domain names, would show an even greater incidence of infringement in legacy gTLDs compared to new gTLDs.)

This study is the first of several in which we will quantitatively examine the likely effects of new gTLDs on trademark holders.

Infringements Per Million (IPM)

A study of the UDRP case data for the last 10 years for both existing (pre-ICANN) open gTLDs (.com, .net, and .org) and for newer ICANN-created open gTLDs shows that the number of infringements within any open gTLD is quite predictable, depending primarily on the number of registrations within that TLD. (For the purposes of this study, we looked at those top-level domains introduced since 2001 that are either “open” — no restrictions on registrants — or whose restrictions are so easily circumvented, or so loosely enforced, that they are effectively “open.” Because trademark owners have not complained as loudly about restricted TLDs, we did not include them in our study. Not examined, therefore, were truly restricted gTLDs: .int, .gov, .mil, .edu, .museum, .coop, and .aero.)

To help understand the relationship, we introduce a new metric: IPM (infringements per million). The data show that among current gTLDs, IPM varies between about 15 and 40, with .com having by far the highest IPM at 41.71 infringements per million registrations.

Domain name growth across all TLDs has for the past ten years grown at a fairly steady 10 – 15% annually (Source: VeriSign Domain Name Brief). Past introductions of new gTLDs have not changed that overall growth. The new round of gTLDs, which foresees an increase of approximately 300 gTLDs (see page 6 of the 2009 ICANN Root Scaling Study, for instance), may, because of intensive marketing, increase this growth rate, but not by orders of magnitude. Based on historical data, the average IPM for the new open gTLDs listed in Table 3 below is 22.47, and 24.15 for .com, .net, and .org. Table 3 lists the IPM for each TLD.

The Data

Legacy open TLDs — .com, .net, and .org — account for the vast majority (94%) of all WIPO and NAF UDRP cases. This percentage has not changed significantly over the 10 years of data, as Table 1 below shows. (The anomalies here, in 2002 and 2007, are due to .info price promotions, where names were offered for free or near-free, which did increase infringements. Price of new TLD registrations – but not the existence of new TLDs – does increase speculative activities of all kinds.)

Table 1: Com, Net and Org as % of all UDRP Cases

Year	CNO	Other	Total	CNO %
Total	23595	1452	25047	94%
2001	2383	18	2401	99%
2002	1814	189	2003	91%
2003	1635	83	1718	95%
2004	2346	150	2496	94%
2005	3039	134	3173	96%
2006	2406	116	2522	95%
2007	2938	356	3294	89%
2008	3196	227	3423	93%
2009	3817	178	3995	96%
2010 (partial)	21	1	22	95%

 

Table 2, below, shows UDRP claims filed in new open gTLDs.

Table 2: Non .com, .net, or .org UDRP Cases by Year

Year	asia	biz	cat	info	mobi	name	pro	tel	Total
Grand Total	8	385	4	860	155	24	11	5	1452
2001	0	2	0	16	0	0	0	0	18
2002	0	74	0	115	0	0	0	0	189
2003	0	33	0	50	0	0	0	0	83
2004	0	90	0	59	0	1	0	0	150
2005	0	50	0	83	0	1	0	0	134
2006	0	27	0	76	9	4	0	0	116
2007	0	38	4	245	62	7	0	0	356
2008	2	47	0	120	52	5	1	0	227
2009	6	24	0	95	32	6	10	5	178
2010 (partial)	0	0	0	1	0	0	0	0	1

Finding 1: Infringement correlates closely to registration volume, but .com has the highest rate.

The vast majority of infringement occurs in pre-ICANN legacy TLDs, not in the newer TLDs. To understand what is likely to happen with the new round of gTLDs, we need to better understand these numbers.

The key metric this regard is “Infringements per Million,” or IPM. Table shows 3 the IPM across open gTLDs for 2009.

Table 3: 2009 Infringements Per Million (IPM) by TLD

TLD	UDRP Cases	Millions of Registrations	IPM
com	3502	83.97	41.71
net	192	12.63	15.12
org	124	7.93	15.64
info	95	5.5	17.27
biz	24	2.1	11.94
mobi	32	.935	34.22
asia	6	.215	27.91
tel	5	.238	21.1

Cases are calculated not by case number, but by number of domain names. The .cat and .pro TLDs are omitted from this table because for these purposes they are statistically insignificant.

We see that the number of UDRP cases is correlated with the number of registrations in the underlying TLD zone, varying from a low of 11.94 IPM (.biz) to a high of 41.71 (.com). To the extent that there is variation, the outlier is .com, with a higher IPM. Only 4% of all cases (178 out of 3817) occur in non-CNO (com, net, org) gTLDs. Only 1.5% (59 out of 3817) occurred in non-CNOIB (com, net, org, info, biz) gTLDs.

Intuitively, this makes sense: because most large brands and high volume websites operate in .com, one would expect a somewhat larger impact of typo-squatting and other infringement, even relative to the installed base. The next level of TLDs: .info, .biz, .net and .org, which cater to smaller websites, and are less viable as typo-squatting targets, have less than half the IPM ratio.

The newer group of TLDs — .mobi, .asia, .tel — fall within these broad parameters, with .mobi and .asia having a slightly higher IPM, perhaps because they were marketed to the domainer community. Nevertheless it is clear that across all TLDs the results broadly correlate to registration volume.

Finding 2: New TLDs will generate an estimated 316 new UDRP cases per year. Infringements will depend on total domain registrations, not the number of new TLDs.

Our first finding shows that the average IPM for open gTLDs created since 2000 is 22.47. What will the rate be going forward, what will be the total number of infringements, and what will be the corresponding enforcement cost to trademark holders?

Using the average 22.47 IPM for TLDs created since 2000, the new round of gTLDs would create 316 new infringements. This is calculated based on a rosy registration scenario for new TLDs; very likely, they will be less successful, and infringements will be fewer.

Last year, growth across all TLDs was 12%: this includes ccTLDs, which grew at 17% (Source: VeriSign Domain Name Brief). This is very much in line with historical growth of domain names, and we predict that the same growth trend will continue. For gTLDs, this will mean a growth from a combined total of 113 million today to 127 million in February 2011, or an additional 13.6 million names. If these additional names are distributed according to current market share, .com would go from 83.97 million names to 94 million names, .net from 12.63 million to 14.14 million, and so on.

In the past, the introduction of new TLDs has not significantly affected the growth of existing TLDs, and this dynamic is unlikely to change, at least in the short term.

Now, let us turn to the introduction of new TLDs. Suppose that with major marketing efforts the new gTLDs manage to double the growth rate of the overall market from 12% to 24%, and to capture 10% of the market in one year. (Again, this result is extremely optimistic for new TLDs.) The results would look like those presented in Table 4:

Table 4: Projected Registrations after Introduction of New gTLDs
(24% increase in market growth, but new gTLDs capture 10% of market)

Existing TLDs	Current Registrations (millions)	+1 Yr Total
com	83.97	93.71
net	12.63	14.10
org	7.93	8.85
info	5.50	6.14
biz	2.01	2.24
mobi	.935	1.04
asia	.215	.240
tel	.238	.270
New TLDs (combined)		14.07

Finding 3: The expected total annual enforcement costs for new gTLDs will be less than $870,000 per year, or less $.10 per trademark worldwide.

If all 316 new infringements were filed as UDRPs, at an average cost of $5000, the cost of enforcement to trademark holders would be $1.58M. There are 1.97 million active and pending trademarks in the U.S. Patent and Trademark Office, so on a per-trademark basis (for the U.S. only – clearly there are many more trademarks globally), the cost of new gTLDs would be $.80 per U.S. trademark, and if the 2.4 million registered trademarks in China and the 825,000 European Community trademarks are included, the cost of new gTLDs is $.30 per trademark.

But we can expect trademark holders to make use of the new Uniform Rapid Suspension (URS) process, which will have a cost of $500, not $5,000. What percentage of UDRP claims would be adjudicated through the URS process? We suspect that a majority of the cases that would have gone to the UDRP will now go through a URS proceeding. The number is hard to predict, but a reasonable estimate is that 50% of the claims that are now filed as UDRPs would be filed as URS proceedings. If so, the average cost of enforcing a trademark in the domain name arena will go from $5000 to $2750, or $869,000 – that’s $.17 per trademark registered in the U.S., Europe, and China. If all the world’s trademarks were included, the cost of new gTLDs would be under $.10 per trademark worldwide.

Conclusion

Trademark and brand owners will be faced with only minor enforcement costs from the introduction of new gTLDs. While the overall cost of UDRPs today is high ($19.5 million per year) – the culprit is .com – not the 10 new gTLDs that have been introduced over the last 10 years. (Cost of defensive registrations is not considered in this paper but will be covered in a later study.)

We estimate the total enforcement cost resulting from new gTLDs to be $869,000, or under $.10 per trademark registered worldwide.

Contrast this cost to the benefits of new gTLDs. The benefits of new gTLDs have been well rehearsed, but are worth repeating here.

• Ordinary web users (as well as brands) will not be forced to spend over $10,000,000 annually to purchase .com domains in the secondary market at inflated prices.
• Major cities such as .nyc, .paris, .berlin and .london want new TLDs. They see millions in revenue, increased tourism, increased efficiency in providing Internet services to their residents.
• Thousands of jobs will be created, because each new registry will need to employ 5 – 10 people at a minimum. (As a point of comparison, Afilias, which manages .info and provides registry services for .org, has over 200 employees.)
• Linguistic communities such as .gal (Galicia), .eus (Basque), and .bzh (Brittany) see huge cultural benefits.
• Vertical TLDs with strong user bases such as .eco expect to use proceeds from registrations to help solve problems such as global warming.

The data show that new gTLDs are less likely to be involved in UDRP claims than .com. An expansion of new gTLDs is not likely to significantly increase UDRP costs for trademark holders. If ICANN introduces the Uniform Rapid Suspension (URS) provisions currently under consideration, trademark enforcement costs for new gTLDs will sink even further.

I recently wrote a paper to which I gave the super-fun title Community Scoring in ICANN’s Draft Applicant Guidebook: How to Make Sure that New Top-Level Domains Are Meaningful and Useful.

I also submitted it as a comment to ICANN’s IRT (Implementation Recommendation Team).

Warning! This is very long. Worth studying every word, of course…

Summary: The new top-level domains (TLDs) from ICANN have the potential to usher in a much more useable Web, but ICANN needs to define a “community” TLD better so that existing communities of interest can create and manage their own TLDs. As it stands, these communities will be shut out and many new TLDs will become meaningless replications of .COM. We recommend tweaking the scoring in ICANN’s Draft Applicant Guidebook to make it easier to qualify as a community.
(more…)

Yesterday, I sent ICANN my comments about the draft recommendations from ICANN’s Implementation Recommendation Team (IRT), which has been tasked with coming up with a trademark protection scheme for new top-level domains. For the most part, I think they did an excellent job.

From the perspective of an Internet user (which we all are), having a clean namespace — no parked pages, no domains used for phishing, spam, malware, etc. etc. — is pretty important for having a good experience with your email and web browsing. Right now, .com and .net are cesspools, filled with garbage, and we all have to live in the filth, because more than half of all domain names are in the .com and .net zones. From the perspective of a business owner (which I am), having a clean namespace means that my domain names and associated businesses are more valuable. For everyone who is not a polluter, the less garbage, the better.

Cleaning up trademark abuse is part of cleaning up the neighborhood. I’m not supporting an erosion of fair use, or condoning the cynical reverse hijacking that some trademark owners engage in, or trying to expand trademark rights beyond what the law provides for. But everyone working in the domain name field has seen plenty of clear instances of someone registering someone else’s trademark as a domain name, then profiting from it, to the detriment of both the trademark owner and the Internet user who was fooled. There’s no reason these cases shouldn’t be treated separately from edge cases where the facts aren’t so clear. In fact, there’s a good reason to do so — it makes the neighborhood a nicer place to live.

I’m hopeful that new TLD registries will regulate themselves better than the current crop of gTLDs has done, and some of the trademark protection rules proposed by the IRT provide some tools to make that easier.

It’s not all good news: the IRT produced some ill-considered attempts to usurp some of ICANN’s authority, and I don’t agree with everything they’ve done. Nonetheless, I think they went a long way to removing what has been the Number One obstacle to the introduction of new TLDs.

My comments to ICANN are more specific:

Minds + Machines, a registry services operator with several clients participating in the new gTLD process (including, “.eco”, the environmental Top Level Domain application backed by Al Gore and the Sierra Club) thanks ICANN for the chance to comment on the new draft from the the Implementation Recommendation Team (IRT), and we commend the IRT on a well-thought-out and timely report.

In general, Minds + Machines supports new TLDs and namespaces that have semantic coherence and meaning, such that a TLD string informs an Internet user about the content likely to be found on websites within that TLD. We believe the restoration of semantic meaning to the top- level domain namespace will increase the usefulness and use of the Web and as a consequence make the Internet more valuable to both producers and consumers. We are concerned that the meaningless effluvium now proliferating within many current gTLDs harms the experience of Internet users and therefore reduces the value of Web sites within those TLDs. We expect that the new round of TLDs will be better regulated and more trusted than the current crop. Protecting the legitimate rights of intellectual property holders, while preserving the freedom to innovate with and within a new TLD, is an important step in that direction.

We therefore welcome and support most of the recommendations from the IRC, with some caveats.

First, we are strongly in favor of a rapid takedown mechanism for clearly abusive domain names, whether used for phishing, spam, malware, trademark abuse or other illegal behavior. We thank the IRT for spelling out a mechanism to deal with these obvious cases and encourage ICANN to empower registries to act in similar fashion against abusive domain name use beyond the trademark arena. We note that the IRT’s recommendations in this regard are similar to policies used to great effect by CoCCA, Minds + Machines’ cousin in the ccTLD world. While we recognize that the system of watch lists and alerts may add some burden to registries and registry operators, we feel that this is small cost to pay to if it means a cleaner namespace — we note, however, that registries should be able to charge a reasonable price to cover costs.

Further, we strongly support the establishment of a centralized IP clearinghouse be set up as an outsourced agency under a long-term contact with ICANN. We thank Bart Lieben of Lada for championing this innovative solution. This solution will make new TLD launches far more economical for all parties and will significantly reduce errors and their associated expense.

We also support, albeit with some trepidation, the IRT’s findings that a list of Globally Protected Marks should be established. We understand that the compilation of such a list will be fraught with controversy, but we are hopeful that the limits of the list and its uses will soon be established by the courts. A list for exact matches will make it easy for registries and registrars to combat trademark abuse in a programmatic way. Registries cannot, however, be expected to police “confusingly similar” matches or other inspections that require intervention by a human with trademark expertise.

We strongly agree with the proposal that applicants be able to apply for more than one string in an application, without an additional application fee, providing such strings represent IDN variants of the same .BRAND in different alphabets. We furthermore recommend extending this concept to geographical TLD’s with multiple spellings and variants in different languages and IDNs (“Mumbai”, “Bombay” and the IDN variants of that city name for example).

Finally, we fully support the idea of “thick” whois. In our view, the “thin” whois is an artifact of a commercial ploy dating from the formation of ICANN and plays no useful role. A “thin” whois endangers the security of registrant data by spreading it across multiple registrars, some percentage of whom can be expected to run sloppy or shady operations, causing registrant data to go missing or to be held hostage in a registrar’s negotiations with ICANN. The “thin” whois is an idea whose time never was, and we would be pleased to see it piled on the scrapheap of rejected mistakes. We note, however, that the TLD registry must retain the right to create the policy regarding disclosure of contact data in order to ensure privacy protection. As a global operator, a registry should not be forced by ICANN rules to violate national privacy laws (in Europe, for instance) without complete indemnification.

Sadly, in one important area, we find that the IRT team has produced an unrevivable Frankenstein. Especially insofar as it relates to second-level domain names, the proposed “Post-Delegation Dispute Mechanism” mandates a scheme that undercuts ICANN’s authority, imposes impossible duties on registries, and, despite various safeguards proposed by the IRT, is an invitation to abuse. While the proposed mechanism may be viable (with amendment) for abuses relating to the TLD string itself, we feel that when applied to second-level domains names it will require registries to police the TLD namespace for potential infringements, which they are neither competent nor empowered to do, while leaving them without any enforcement mechanism. Instead, any diligence in regard to preventing trademark abuse will simply invite multiple punitive administrative actions by ICANN, in which the registry is set up as the “defendant.” Furthermore, the recommendation does not specify a date of registration of a trademark for a complainant, so that anyone could get a trademark after the fact for the purpose of filing complaints. Finally, we note that the language of the “ten point test” for this section is full weak wording: it is “probably” scalable; “we think it could”; “possibly”; and “may be workable.” This language indicates that the authors do not really believe this is the proper solution for ICANN enforcement of registry contracts, and neither do we. Minds + Machines strongly recommends striking this entire section; if ICANN is not doing its job properly, then the proper avenue is complaint to ICANN.

Despite this one area of disagreement, we congratulate the IRT team on their hard work and creativity, which has obviously yielded substantial results and promises to settle many long-standing arguments, and again thank ICANN for the chance to comment.

Sincerely,

Antony Van Couvering, CEO
Minds + Machines