Re: [gnso-idng] reporting back to the council

[Avri Doria <avri@xxxxxxx>]

Hi,

I am afraid this is much longer then my normal email, and I apologize for that 
but I have to deal with several issues in this response.  Also I want to note 
that this is my own opinion and not the opinion of the NCSG.  I am submitting 
this to the NCSG as well as to this DT and am in the process of kicking off a 
work effort in the NCSG for us to further consider the issue and create a 
position paper on the subject.  Given the expertise in the NCSG on Trademark 
Law, Consumer Protection and Human Rights including Freedom of Expression, I am 
sure that whatever gets produced will be far more comprehensive then what I am 
submitting to this group.  I have, however, submitted my basic proposition to 
NCSG review and criticism already and must admit that some may think my 
position a bit too centrist.

In this contribution, I need to present my argument on the primacy of the 
visual aspect in issues in confusing similarity or likelihood to confuse that 
needs to be resolved by the application process. And I need to account for the 
other possible aspects related to confusing similarity and  likelihood to 
confuse that need to be resolved by the application process.  Finally I need to 
(re-)present my views on how I think these should be handled.

---

To start, the GNSO recommendation only mentions possible criteria by which a 
name can be considered confusing similar or likely to cause confusion but does 
not build a definitive list of those characteristics.  

So in a sense Chuck is correct when he claims that there is no specific 
endorsement of visual similarity as the only method to be used, and I believe 
that in a sense I am correct in stating that there is no specific endorsement 
of the use of translation as the basis for  confusing similarity or likelihood 
to confuse.  We are both right, and perhaps I should leave it there.  But I 
won't.

There is mention: that 'confusion may be visual, phonetic or conceptual' as 
well as many other mentions in the background descriptions of how various 
International statutes consider the issue.  And within those discussions 
different causes of confusion get different treatment.  

My argument is based on the differing treatment that the various possible 
causes of confusion receive in the examples of confusing similarity and 
likelihood to confuse.

1. Visually confusing serves as a base component of confusingly similarity in 
all discussions.  Whatever other views they may have about confusingly 
similarity all of the referenced documents accept that visual confusion is a 
reason for some string to be designated as confusingly similar to some other 
string (though of course there are discussions about what it means to be 
visually confusing - a difficult topic in its own right.)  Among the Trademark 
documents as I understand it, there is no argument against visually confusing, 
but there are many qualifications in trademark law as it applies to a wide 
variety of identifiers, not just domain names, for example that TLDs have no 
colors, are rarely "pronounced" as words or normal names are pronounced - so a 
lot of the TM confusing similarity stuff may not apply - but that is for review 
panels and law courts to decide - not this DT.

I believe that it is only visual confusion of the character patterns that is 
accepted as a reason for likelihood of confusion in all discussions.   

Does anyone argue that visual aspects should not be considered when determining 
whether a string is confusingly similar?  I think, for the most part, not, and 
the fact that everyone accepts visual confusion (at least to some degree) as a 
reason for confusing similarity and the likelihood of confusion,  gives it a 
special status as the one thing we all agree on - we need to be concerned about 
those strings that are visually confusing.

I believe this gives the visual aspect a primacy, and though this does not mean 
that they are the only considerations that need to be considered in the process 
it does mean that they are reasonably the only aspects that need to be 
considered for all new gTLD in the initial evaluation.

2. No other of type of confusingly similarity is agreed upon by everyone.  
Various are mentioned as things that have been considered and may considered 
again.  But none of these has the status of being agreed to by all of the 
International Legal Instruments.

E.g.

> For example, the Committee considered the World Trade Organisation's TRIPS 
> agreement, in particular Article 16 which discusses the rights which are 
> conferred to a trademark owner.[44] In particular, the Committee agreed upon 
> an expectation that strings must avoid increasing opportunities for entities 
> or individuals, who operate in bad faith and who wish to defraud consumers. 
> The Committee also considered the Universal Declaration of Human Rights[45] 
> and the International Covenant on Civil and Political Rights which address 
> the "freedom of expression" element of the Committee's deliberations.

> 1883 Paris Convention on the Protection of Industrial Property[48]. It 
> describes the notion of confusion and describes creating confusion as "to 
> create confusion by any means whatever" {Article 10bis (3) (1} and, further, 
> being "liable to mislead the public" {Article 10bis (3) (3)}. The treatment 
> of confusingly similar is also contained in European Union law (currently 
> covering twenty-seven countries) and is structured as follows. "...because of 
> its identity with or similarity to...there exists a likelihood of confusion 
> on the part of the public...; the likelihood of confusion includes the 
> likelihood of association..."

> Australian Trade Marks Act 1995 Section 10 says that "...For the purposes of 
> this Act, a trade mark is taken to be deceptively similar to another trade 
> mark if it so nearly resembles that other trade mark that it is likely to 
> deceive or cause confusion"

> For example, the European Union Trade Mark Office provides guidance on how to 
> interpret confusion. "...confusion may be visual, phonetic or conceptual. A 
> mere aural similarity may create a likelihood of confusion. A mere visual 
> similarity may create a likelihood of confusion. Confusion is based on the 
> fact that the relevant public does not tend to analyse a word in detail but 
> pays more attention to the distinctive and dominant components. Similarities 
> are more significant than dissimilarities. The visual comparison is based on 
> an analysis of the number and sequence of the letters, the number of words 
> and the structure of the signs. Further particularities may be of relevance, 
> such as the existence of special letters or accents that may be perceived as 
> an indication of a specific language. For words, the visual comparison 
> coincides with the phonetic comparison unless in the relevant language the 
> word is not pronounced as it is written. It should be assumed that the 
> relevant public is either unfamiliar with that foreign language, or even if 
> it understands the meaning in that foreign language, will still tend to 
> pronounce it in accordance with the phonetic rules of their native language. 
> The length of a name may influence the effect of differences. The shorter a 
> name, the more easily the public is able to perceive all its single elements. 
> Thus, small differences may frequently lead in short words to a different 
> overall impression. In contrast, the public is less aware of differences 
> between long names. The overall phonetic impression is particularly 
> influenced by the number and sequence of syllables."

That quote mention many different possible ways of looking at and considering 
the issue in an international context, including issues of good or bad faith, 
freedom of expression, resemblance, phonetic or conceptual in addition to 
visual.  But at no time did the GNSO council make a definitive list - it 
referred the problem of determining what the appropriate causes for possible 
confusion  to the objection review team.

Finally in this discussion there has been a presumption that I was the only one 
who felt tension about the definitions.  This is most definitely not the case 
as expressed in the following.  The reason all of this was left ambiguous is 
because of this tension.

> xvii) There is tension between those on the Committee who are concerned about 
> the protection of existing TLD strings and those concerned with the 
> protection of trademark and other rights as compared to those who wish, as 
> far as possible, to preserve freedom of expression and creativity. The 
> Implementation Plan sets out a series of tests to apply the recommendation 
> during the application evaluation process.


3. The Question then becomes how did the GNSO Council propose dealing with the 
tension and the ambiguity.  As mentioned above, it did not do it by listing a 
set of possible conditions that would be tested against - except for those 
discussed in the Implementation Plan - which I cannot find, but remember as 
being restricted to Visual confusability as was DAGv1 at the time of the 
Board's approval for implementation.  What was provided was Recommendation 12:

> 12. Dispute resolution and challenge processes must be established prior to 
> the start of the process.

I note that this is not specific about challenges regarding confusing 
similarly, but I do believe there was an expectation that confusing similarity 
could be challenged.  DAGv3 backs this up with:

> 2.1.1.1 String Similarity Review
> This review involves a preliminary comparison of each
> applied-for gTLD string against existing TLDs and against
> other applied-for strings. The objective of this review is to
> prevent user confusion and loss of confidence in the DNS.
> The review is to determine whether the applied-for gTLD
> string is so similar to one of the others that it would create a
> probability of detrimental user confusion if it were to be
> delegated into the root zone. The visual similarity check
> that occurs during Initial Evaluation is intended to augment
> the objection and dispute resolution process (see Module
> 3, Dispute Resolution Procedures) that addresses all types
> of similarity.
> This similarity review will be conducted by an independent
> String Similarity Panel.

This shows that the attribute that will be tested for in the initial evaluation 
of new gTLDs is Visual confusion,  This corresponds to the special place given 
to Visual aspects of confusing similarity in all discussions of confusing 
similarity, i.e. as mentioned above, there is consensus about the need to test 
for visual confusion.  The consensus of the importance of visual confusion 
makes it unique among the possible causes of confusion. Whereas there is no 
consensus about other areas of possible likelihood of confusion except that 
they are grounds for objection. 


> 3.1.1 Grounds for Objection
> An objection may be filed on any one of the following four
> grounds:

> String Confusion Objection – The applied-for gTLD string is
> confusingly similar to an existing TLD or to another appliedfor
> gTLD string in the same round of applications.


and in

> 3.1.2 Standing to Object

> String confusion       Existing TLD operator or gTLD applicant in
> current round
> 

4.  This brings me to my conclusion: that while visual confusing similarity is 
included in the initial evaluation and every application is subject to that, 
any other form of possible likelihood to cause confusion must be declared in an 
objection and then will be adjudicated by the Review Panel according to 
relevant International law in relation to that particular gTLD.  So while it is 
ok for a Registry to consider whether it will make a objection based on 
translation or meaning, until such time as they make such a objection and it is 
approved by the Review Panel, it is not, properly speaking, a criteria for 
confusing similarity, it is just a possibility that may be considered by a 
Review Panel. And even after that it would still not be a criteria for any 
other gTLD until the relevant Registry had filed its own objection etc.

5. What I have said about this during the course of this Drafting Team. 

Dec 11: -

>  i believe that we should just be stating the problem and not a solution. 

Dec 12: 

> Which says to me that it is a complex issue, but that visual similarity is 
> the predominant factor in determining confusion.  And only in cases where you 
> can expect the a person to know both languages would meaning enter into the 
> issue as a complicating factor.  Likewise for phonetic comparisons.
> 
> So yes, confusion is the key.
> 
> Most can be eliminated based on visual confusion.
> 
> And for the others where there really is a complicating factor that may cause 
> confusion, then and only then does phonetics and meaning come in - and it 
> comes in at the challenge level.

Dec 15: 

> My problem with your re-write is that it presupposes that translation is a 
> primary cause for 'confusingly similar'  and as I and others have argued, 
> this is not universally accepted (though I do accept meaning as a possible 
> complicating factor in an objection, I am not sure others even go that far).

> by lowest common denominator i meant the kind of similarity we can all agree 
> would be included in the category 'confusingly similar'.

(though i was berated for not understanding my maths well enough to know what 
lowest common denominator really meant)

... (many similar comments )...

16 April

>  Certainly various issues are discussed, but there is no statement of a 
> council decision for confusing similarity to be more the visual. 

As I said a lot was discussed but no decision was ever made to include more.  
The point is made in the discussion of recommendation 2 in the  GNSO 
Recommendation:  

> There is tension between those on the Committee who are concerned about the 
> protection of existing TLD strings and those concerned with the protection of 
> trademark and other rights as compared to those who wish, as far as possible, 
> to preserve freedom of expression and creativity. The Implementation Plan 
> sets out a series of tests to apply the recommendation during the application 
> evaluation process.

It is this tension that we are still expressing in this debate.

16 April

> I believe that it was never the intent of the GNSO Council to allow 'meaning' 
> within the category of 'confusingly similar'.  

This is strong, yet I stand by it.  While there as the intention of the council 
to allow those with standing to object on any ground for which they could find 
a basis, including meaning, there was never a specific GNSO council decision to 
specifically include meaning as a specific element that would be tested for.  
I.e I argue that it does not have the consensus status of visual likelihood of 
confusion.

6. How this related to the issue of the IDNG DT.

The issue we were discussing was whether a registry who had (or was applying 
for) a LDH ASCII gTLD would be able to apply for a name that would be 
considered confusing similar if others applied for it.  I agued that this is 
complicated by several factors.  Among those factors:

- Is this based on an initial evaluation visual likelihood to cause confusion 
or another objection based likelihood to cause confusion.

In the case of something failing the initial evaluation, I felt that this 
should be taken care of in extended evaluation.  I later realized that  in 
DAGv3 Extended evaluation is not available for issues related to String 
Similarity.  So this is a problem that needs to be dealt with if the council 
wishes to allow the option of the same registry controlling visually similar 
strings that could cause confusion. So I supported, and support, a minimal 
report to the Council  that in the case of visual confusing similarity perhaps 
it should be possible to request an extended evaluation were a Registry can 
request the visually similar IDN gTLD based on their ownership of the LDH 
ASCII.  

In the case of a name that might be considered confusingly similar based on 
something like aural evaluation (i.e. transliteration) or translation I did not 
see a problem.  Since these criteria are only taken into account when someone 
with standing objects on that basis and I would not expect the registry to 
object to its own application for being confusingly similar based on meaning.  
I figured the chance of that was rare - unless of course it was a question of a 
translation for 'commerce' or 'business' claimed by both com and biz - who 
would both have standing.

- In discussing two TLDs that are visually similar a likely to cause confusion, 
the council needs to determine what this means in terms of issues of confusion 
at the second level and the subject of TLD synchronicity.  While they may 
decide that this needs some policy considerations I am not trying to 
predetermine the result of that discussion - only saying that the the 
discussion has to happen before any decision is made on how to handle these 
situations.


7. My proposed solution to this disagreement in the DT is that I am willing to 
quit saying the Visual confusing similarity is the only possible criteria for 
determining likelihood of confusion if others will refrain from saying that 
reasons such as translation have been included in the GNSO's recommendations as 
more then possible reasons for objection.  I do believe we can write a report 
that steers carefully between the Scylla and Charybdis of this debate.   What I 
recommended is that we present the council with an adequate description of this 
complex problem in careful words that make neither of the parties in this 
debate or others in the group uncomfortable. And we make recommendations about 
further work the council may consider within its work prioritization model if 
we so wish.    


thanks
a.