Comment Close
Date
Statement
Name 

Status

Assignee(s) and
RALO(s)

Call for
Comments
Call for
Comments
Close 
Vote
Announcement 
Vote OpenVote
Reminder
Vote CloseDate of SubmissionStaff Contact and EmailStatement Number
n/aConfusingly Similar gTLDsAdopted
12Y, 1N, 0A 
Alan Greenberg (NARALO)13.09.201316.09.2013
12:00 UTC 
17.09.201317.09.201323.09.201324.09.201316.09.2013Cherine Chalaby
cherine.chalaby@icann.org 
AL-ALAC-ST-0913-04-00-EN

FINAL VERSION TO BE SUBMITTED IF RATIFIED

Please click here to download a copy of the PDF below.

 

 

FINAL DRAFT VERSION TO BE VOTED UPON BY THE ALAC

ALAC Advice to the ICANN Board on String Similarity

On 24 June 2013, as requested by the GAC, the Board New gTLD Program Committee (NGPC) considered the issue of singular and plural strings being confusingly similar and decided to let the original process as documented in the Applicant Guidebook stand (i.e. subject to individual objections).

Events and findings, which have occurred since then, indicate that the transparent, predictable and objective criteria called for by the GNSO New gTLD Policy Recommendations 1 and 9 have not been met, ultimately resulting in Internet end-user confusion.

As examples, recent string similarity objection decisions have ruled:

  • .sport is confusingly similar to .sports;
  • .tour is confusingly similar to .tours;
  • .car is not confusingly similar to .cars;
  • .hotel is not confusingly similar to .hotels;
  • .tv is not confusingly similar to .tvs;
  • .pet is not confusingly similar to .pets;
  • .shop is not confusingly similar to the Chinese string for .shop;
  • .shop is confusingly similar to the Chinese string for .onlineshopping; and
  • .com is both confusingly similar to .cam and not confusingly similar to .cam.

The ALAC is particularly concerned with the issue of singular versus plural strings. A central issue is that the "confusingly similar" test relies purely on visual similarity. Based on the initial evaluation adding an "s" makes it a recognizably different string. The recent NGPC decision (http://www.icann.org/en/groups/board/documents/minutes-new-gtld-25jun13-en.htm#2.d), re-affirms this position, although several NGPC members expressed regret that the wording of the Applicant Guidebook effectively forced this outcome.

The Applicant Guidebook section 2.2.1.1 describes the string similarity review:

This review involves a preliminary comparison of each applied-for gTLD string against existing TLDs, Reserved Names (see subsection 2.2.1.2), and other applied-for strings. The objective of this review is to prevent user confusion and loss of confidence in the DNS resulting from delegation of many similar strings.

Note: In the Applicant Guidebook, “similar” means strings so similar that they create a probability of user confusion if more than one of the strings is 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.

The ALAC disagrees with the NGPC decision. The problem is the belief that "visual similarity" relies purely on what would be called "pattern matching" in computer terminology. Pattern matching is certainly part of human perception, but it is not limited to that issue alone. At issue is whether two strings will be PERCEIVED as being equivalent, and perception is a far more complex (and less understood) issue.

Consider strings pairs such as:

  • hilton.hotel and hilton.hotels;
  • soccer.sport and soccer.sports;
  • poodle.pet and poodle.pets; and
  • taj-mahal.tour and taj-mahal.tours.

Will these be memorably equivalent or different to typical Internet users, individuals who have no knowledge of terms such as ICANN, domain name, TLD and second level? The ALAC strongly doubts that most Internet users will note the subtle differences.

If both singular and plural TLDs are delegated:

  •  Some registrants would register (and defend) their 2nd level names in both TLDs and map them to the same web and e-mail services. Doing so would reduce the impact on users, but would of course increase costs to registrants (including defensive registrations in both TLDs) who must double their registrations. Moreover, due to either policy differences between the two TLDs, or due to timing constraints, such duplication may not be possible.
  • Other registrants would use only one of the TLDs, resulting in quasi-random behavior from a user’s perspective. We cannot expect the typical Internet user to be able to differentiate between two such name spaces. The ALAC, which is responsible for representing the interests of Internet users in ICANN, believes that we have a genuine case of "confusingly similar".
The ALAC advises the Board to revisit the issue of new TLD strings, which are singular and plural versions of the same word, and ensure that ICANN does not delegate strings that are virtually certain to create confusion among Internet users and therefore result in loss of faith in the DNS.

The ALAC is also concerned with the lack of predictability and consistency in objection decisions. Allowing identical strings to be individually evaluated (based on different arguments and by different panelists) opens the door to inconsistent decisions. Moreover the clear lack of consistency among the evaluations of different panelists implies that overall, ICANN will either put some TLDs in contention sets where it is not justified, or delegate strings that will cause user confusion.

It is not clear how this can be fixed at this time, but proceeding with no change is either going to unfairly disadvantage some applicants and their target markets or cause guaranteed user confusion.

The ALAC advises the Board to:
  • review the objection decision system with multiple panels that leads to inconsistency and not only review the obvious case of .cam/.com where conflicting objection decisions have forced such review;
  • determine a viable way forward which will not create unwarranted contention sets nor delegate multiple TLDs destined to ensure user confusion and implicit loss of faith in the DNS.

FIRST DRAFT SUBMITTED

ALAC Advice to the ICANN Board

Revised 13 September 2013

On 24 June 2013, as requested by the GAC, the Board New gTLD Program Committee (NGPC) considered the issue of singular and plural stings being confusingly similar and decided to let the original process stand (subject to individual objections).

Events and findings which have occurred since then indicate that the transparent, predictable and objective criteria called for by the GNSO New gTLD Policy recommendations 1 and 9 have not been met, ultimately resulting in Internet end-user confusion.

As examples, recent string similarity objection decisions have ruled:

  • .sport is confusingly similar to .sports;
  • .tour is confusingly similar to .tours;
  • .car is not confusingly similar to .cars;
  • .hotel is not confusingly similar to .hotels;
  • .tv is not confusingly similar to .tvs;
  • .pet is not confusingly similar to .pets;
  • .com is confusingly similar to .cam;
  • .com is not confusingly similar to .cam;
  • .shop is not confusingly similar to the Chinese string for .shop; and
  • .shop is confusingly similar to the Chinese string for .onlineshopping.

The ALAC is particularly concerned with the issue of singular vs plural strings. A central issue is that the "confusingly similar" test relies purely on visual similarity. Based on the initial evaluation adding an "S" makes it a recognizably different string. The recent NGPC decision (http://www.icann.org/en/groups/board/documents/minutes-new-gtld-25jun13-en.htm#2.d), re-affirms this position, although several NGPC members expressed regret that the wording of the Applicant Guidebook effectively forced this outcome.

The Applicant Guidebook section 2.2.1.1 describes the string similarity review:

This review involves a preliminary comparison of each applied-for gTLD string against existing TLDs, Reserved Names (see subsection 2.2.1.2), and other applied-for strings. The objective of this review is to prevent user confusion and loss of confidence in the DNS resulting from delegation of many similar strings.

Note: In this Applicant Guidebook, “similar” means strings so similar that they create a probability of user confusion if more than one of the strings is 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.

The ALAC disagrees with the NGPC decision. The problem is the belief that "visual similarity" relies purely on what, in computer terminology, would be called "pattern matching". Pattern matching is certainly part of human perception, but it is not limited to that. At issue is whether two strings will be PERCEIVED as being equivalent, and perception is a far more complex (and less understood) issue.

Consider strings pairs such as:

  • hilton.hotel and hilton.hotels;
  • soccer.sport and soccer.sports;
  • poodle.pet and poodle.pets; and
  • taj-mahal.tour and taj-mahal.tours.

Will these be memorably equivalent or different to typical Internet users (the ones who have no knowledge of terms such as ICANN, domain name, TLD and second level)? The ALAC strongly doubts that most Internet users will note the subtle differences.

If both singular and plural TLDs are delegated:

  •  Some registrants would register (and defend) their 2nd level names in both TLDs and map them to the same web and e-mail services. This would reduce the impact on users, but would of course increase costs to registrants (including defensive registrations in both TLDs) who must double their registrations. Moreover, due to either policy differences between the two TLDs, or due to timing constraints, such duplication may not be possible.
  • Other registrants would use only one of the TLDs, resulting in quasi-random behavior from a user’s perspective. One cannot expect the typical Internet user to be able to differentiate between two such name spaces, and therefore the ALAC, which is responsible for representing the interests of Internet users in ICANN, believes that we have a genuine case of "confusingly similar."
The ALAC advises the Board to revisit the issue of new TLD strings which are singular and plural versions of the same word, and ensure that ICANN does not delegate strings that are virtually certain to create confusion among Internet users and therefore result in loss of faith in the DNS.

The ALAC is also concerned with the lack of predictability and consistency in objection decisions. Certainly allowing identical strings to be individually evaluated (based on different arguments and by different panelists) does not appear to have been a wise move. Moreover the clear lack of consistency among the evaluations of different panelists implies that overall, we will put either some TLDs in contention sets where it is not justified, or delegate strings that will cause user confusion.

It is not clear how this can be fixed at this time, but proceeding with no change is either going to unfairly disadvantage some applicants and their target markets or cause certain user confusion.

The ALAC advises the Board to review not only the obvious cases such as .cam/com where conflicting objection decision have forced such review, but the general case and determine a viable way forward which will not create contention sets where it is unwarranted, nor delegate multiple TLDs destined to ensure user confusion and implicit loss of faith in the DNS.
  • No labels

9 Comments

  1. Good stuff. A few suggestions


    Replace:

    Since that time, the situation has evolved, and we are now in a position where it is difficult to believe that we are using transparent, predictable and objective criteria for evaluating confusingly similar strings as called for in the original GNSO New gTLD Policy recommendations 1 and 9.

    with

    Events and findings which have occurred since then indicate to Internet end-users a confused, inconsistent, and un-transparent implementation of GNSO New gTLD Policy recommendations 1 and 9 that, in our opinion, does not appear to match the expected outcomes


    The ALAC disagrees with the NGPC decision. 

    A link to the decision, as well as a very brief analysis of that decision, needs to be added.


    Some registrants would register (and defend) their 2nd level names in both TLDs and map them to the same web and e-mail services.

     I would explicitly add the word "defensive" (as in "defensively register"). This concept is well understood as a reason why a registrant would have second level names in multiple TLDs even if they would only need one.


    The ALAC advises the Board to review not only the obvious cases such as .cam/com but the general case and determine a viable way forward which will not create contention sets where it is unwarranted, nor delegate multiple TLDs destined to ensure user confusion and implicit loss of faith in the DNS.

    To me this recommendation is vague and confusing. I'm not sure that cam/com is a source of confusion that we should protest; what is problematic is the inconsistency of multiple panels coming to different conclusions on the same contention set. We should be advocating that all instances of a specific contention be heard by a single process, perhaps involving a single multiple-person panel for large contention sets rather than multiple one-person panels.


    If we are speaking specifically about cam/com and singular/plural, we likely ought to address the issue of multiple script/language versions of the same word. Consistency (IMO) is more important than one way or the other, though I would suggest that on the whole ALAC recommend erring on the side of caution (ie, fewer approved nearly-similar strings) when the situation is closely contested. 


     

    1. Events and findings which have occurred since then indicate to Internet end-users a confused, inconsistent, and un-transparent implementation of GNSO New gTLD Policy recommendations 1 and 9 that, in our opinion, does not appear to match the expected outcomes

      I am happy with the sentence restructure, but it must makes explicit reference to the three descriptors in the policy: transparency, predictability and objectivity.

      A link to the decision, as well as a very brief analysis of that decision, needs to be added.

      Don't think they need that, but happy to add to make it more understandable to other.

      Defensive.

      Fine

      To me this recommendation is vague and confusing. I'm not sure that cam/com is a source of confusion that we should protest; what is problematic is the inconsistency of multiple panels coming to different conclusions on the same contention set. We should be advocating that all instances of a specific contention be heard by a single process, perhaps involving a single multiple-person panel for large contention sets rather than multiple one-person panels.


      Was not trying to advocate on cam/com (although I personally find them confusing). It is a given that they must look at that one. I am suggesting that they review the more general issue of inconsistency at the same time. Not only in cases where there are multiple objections for the same set of strings, but for cases where it is panelists are reaching VERY different conclusions with SIMILAR type of pairs (such as the two Chinese ones mentioned). I will try to re-word. I am specifically not suggesting a solution, because I o not really understand exactly what they can do within the framework of the AG, particularly with objections that have already been heard. BUt they need to look at options.

       

       

      1. These are not defensive in the sense of preventing someone else from getting  the name, but directly to allay user confusion. But I have added a reference to defensive registrations.

        I have also re-worded the first change you suggested to ensure that the point we are making is that the Policy was NOT properly implemented.

    • .com is confusingly similar to .cam;
    • .com is not confusingly similar to .cam;

     

    Were there really 2 directly conflicting decisions?

     

     

    1. Three actually.

      There are three applicants for .cam. Verisign (.com) filed objections against all three. All objections are heard by a sole panelist (that is, a "panel" of one person). Two were assigned to the same person and he found that there was no confusion. The third was heard by a different panelist and he found that they were confusingly similar.

      Since the decision is supposed to be based on the strings alone and not on how the TLD will be used, it was (in my opinion) clearly a mistake to allow multiple objections against the same pair of strings to be heard individually, but that is what the AG allowed to happen.It would seem that the objections could have been consolidated by the dispute resolution service provider (AG section 3.4.2), but apparently they decided not to do that.

      If the AG rules are to be followed, the third applicant is now out of the running, and the first two are in a contention set where they either need to come to an agreement or go to auction.

      Here are pointers

      http://images.go.adr.org/Web/AmericanArbitrationAssociation/%7B2e0321f5-da2c-4b68-a4fe-732e45dd8380%7D_50_504_T_00224_13_determination.pdf


      http://images.go.adr.org/Web/AmericanArbitrationAssociation/%7B89cde8ce-0b0f-4511-8fa2-d34d3baf5295%7D_50_504_T_00226_13_determination.pdf


      http://images.go.adr.org/Web/AmericanArbitrationAssociation/%7Bccb221d7-8f2e-4b86-8dda-f76e69fabcd5%7D_50_504_T_00229_13_determination.pdf

      You can see the full set of string objections (and results to date) at http://tinyurl.com/StringObjections. The full address is  http://images.go.adr.org/Web/AmericanArbitrationAssociation/%7B3d7531bb-c0b8-462e-ab7a-2a6572300363%7D_ICANN_DRP_StringConfusion_Objections.pdf.

  2. I would make a special point about the directly contradictory decisions on .com and .cam  The issue there is not that they are confusingly similar; it is that they are identical and yet one decision says  confusingly similar and two decisons - on the exact words - say not so.

     

    I would take these examples out of the list and make a separate point: That on the exact same words, there are completely contradictory results.

    1. Holly, the situation with the various singular/plural is effectively the same thing. Multiple decisions, some of which say that a singular is confusingly similar to the plural, and other that it is not. The nouns differ, but it is effectively the same problem.

      The list is examples of problems. We then separate them out into to categories (singular/plural and other contradictions) in the following text.

      If you feel strongly about it, I can easily segregate the com/cam pair though.

  3.     The ALAC advises the Board to review not only the obvious cases such as .cam/com where conflicting         objection decision have forced such review, but the general case and determine a viable way forward         which will not create contention sets where it is unwarranted, nor delegate multiple TLDs destined to           ensure user confusion and implicit loss of faith in the DNS.

    I don't think that the board should address singular cases, but the objection decision system as a whole. I propose to modify this paragraph as follows:

    "The ALAC advises the Board to review the objection decision system with multiple panels that led to inconsistency and determine a viable way forward to avoid such contradiction.

    The ALAC wishes that ICANN will not delegate multiple TLDs that ensure user confusion and implicit loss of faith in the DNS."

    1. Tijani, at this stage we do not know if the NGPC will review the general case or not, but it IS clear that they must at the very least address the specific case of .cam/.com and rule on it.

      But I like your wording of the first part and will incorporate.