We are submitting the present comment on behalf of CORE Association, Registry operator for three IDN TLDs and also Registry Service Provider for 20+ other TLDs. This submission is also made on behalf of COREhub, SRLU, a CORE subsidiary which is an ICANN Accredited Registrar (IANA 15). We should also mention that CORE is the Registry Service Provider for .quebec and has been working with .quebec promoters since 2006. 

We welcome the present Preliminary Issue Report and we would like to thank the staff’s work on this matter. In general, we support the proposal of a PDP for handling the issue of diacritics, but we would also like to make some remarks:

First, we still believe that the PDP is not strictly necessary to solve the issue that originated this debate: namely, the possibility of handling the existing .quebec with a new .québec as if they were TLD variants. In this regard, the Preliminary Issue Report inherits a misunderstanding that arose from the specific implementation of the 2012 Applicant Guidebook, which has been carried forward until now. The misunderstanding lies in pretending that the String Evaluation test is about “visual similarity” when in reality it is clearly stated that it is about “user confusion.”

Visual similarity is indeed the most relevant element for finding possible user confusion, but it is not the only or deciding factor. In fact, there are cases of user confusion in situations where the TLDs are less visually similar than others with higher visual similarity. And, most importantly, there may be cases with high visual similarity with zero possible user confusion. One example of this which has been developed in the corresponding PDPs is the use of variant TLDs: even if they are very similar in all respects (not only visually) they are delegated to the same Registry Operator, AND domain at the second level are managed by the same registrar, AND those similar domains are entitled to the same registrant. In this situation, we are making very similar things behave as the same thing, and therefore there is no opportunity for third party user confusion. 

The problem for the .quebec/.québec issue is that, on one hand, the LGR panel for latin script decided that “e” and “é” are not confusingly similar and, therefore, they are not variants. On the other hand, in the previous round, due to a broad interpretation, the implementation of string evaluation based only on visual similarity would have considered .quebec and .québec as “too similar”. This is a clear contradiction, but the solution is also obvious: applying, in this case voluntarily by the registry, the solution for TLD variants (that is, handling .quebec and .québec, which are not variants in ICANN policy terms, as if they were variants). 

We understand and support the position of the GNSO Council to not make exceptions, but here the point is not making an exception for .québec but rather to affirm the existing policy that the decisive factor in string evaluation is the “actual or potential user confusion” and not just the “visual similarity,” which is just one very relevant element (albeit not the only one) to assess user confusion.  

Therefore, we kindly request the Issue Report to be amended in two concrete points:

1) Clarify that the goal of of String Evaluation is always user confusion and not just assessing visual similarity as such;

2) Propose to the GNSO Council that stressing the point above could be a solution for some concrete cases such as .quebec/.québec. 

In any case, we reiterate our support to move ahead with the proposed PDP, since it would be able to handle different aspects that go beyond the concrete .quebec/.québec case. 

In general, we support the proposal of a PDP for handling the issue of diacritics, but we would also like to make some remarks:

First, we still believe that the PDP is not strictly necessary to solve the issue that originated this debate: namely, the possibility of handling the existing .quebec with a new .québec as if they were TLD variants. In this regard, the Preliminary Issue Report inherits a misunderstanding that arose from the specific implementation of the 2012 Applicant Guidebook. The misunderstanding lies in pretending that the String Evaluation test is about “visual similarity” when in reality it is clearly stated that it is about “user confusion.”

Visual similarity is the most relevant element for finding possible user confusion, but it is not the only or deciding factor. There are cases of user confusion in situations where the TLDs are less visually similar than others with higher visual similarity. Most importantly, there may be cases with high visual similarity with zero possible user confusion. One example of this which has been developed in the corresponding PDPs is the use of variant TLDs: even if they are very similar in all respects (not only visually) they are delegated to the same Registry Operator, AND domain at the second level are managed by the same registrar, AND those similar domains are entitled to the same registrant. In this situation, we are making very similar things behave as the same thing, and therefore there is no opportunity for third party user confusion. 

The problem for the .quebec/.québec issue is that, on one hand, the LGR panel for latin script decided that “e” and “é” are not confusingly similar and, therefore, they are not variants. On the other hand, in the previous round, due to a broad interpretation, the implementation of string evaluation based only on visual similarity would have considered .quebec and .québec as “too similar”. This is a clear contradiction, but the solution is also obvious: applying, in this c

CORE Association supports the proposal of a PDP for handling the diacritics issue, but we would also request the Issue Report to be amended to:

1) Clarify the goal of String Evaluation is always user confusion and not just assessing visual similarity as such;

2) Propose to the GNSO Council that stressing the point above could be a solution for concrete cases such as .quebec/.québec. 

Rationale:

We believe the PDP is not strictly necessary to solve the issue that originated this debate: i.e. the possibility of handling the existing .quebec with a new .québec as if they were TLD variants. The Preliminary Issue Report inherits a misunderstanding in pretending that the String Evaluation test is about “visual similarity” when it is clearly stated that it is about “user confusion.”

Visual similarity is the most relevant element for finding possible user confusion but not the only factor. There are cases of user confusion where the TLDs are less visually similar than others and there may be cases with high visual similarity with zero possible user confusion. One example of the latter is the use of variant TLDs: while they are very similar in all respects (not only visually), they are delegated to the same Registry Operator, AND domain at the second level are managed by the same registrar, AND those similar domains are entitled to the same registrant. Here there is no user confusion. 

The problem for the .quebec/.québec issue is that, on one hand, the LGR panel for latin script decided that “e”/“é” are not confusingly similar (and hence not variants). On the other hand, in the previous round, string evaluation based on visual similarity only would have considered .quebec/.québec as “too similar”. This obvious contradiction also has an obvious solution: handling .quebec/.québec, which are not variants in ICANN policy terms, as if they were. 

In any case, we reiterate our support to move ahead with the proposed PDP, since it would handle aspects beyond the .quebec/.québec