Dear Work Track members, Please find below the action items and discussion notes from the call on 26 September. These high-level notes are designed to help Work Track members navigate through the content of the call and are not a substitute for the chat transcript or the recording. The meeting recording and chat transcript are available at: https://community.icann.org/x/DBLfAw[community.icann.org]. The documents referenced on the call are attached and excerpts from the chat room are included below. Kind regards, Emily -------------------------------------------------------------------------- Action Items and Discussion Notes: 26 September 2017 1. Plenary Update: -- We reached consensus on appointing Cheryl-Langdon Orr as Co-Chair of the PDP WG, pending approval from the GNSO Council. -- Went through the four Work Track Sub Team reports.  No issues. -- Work Track Sub Team 5: At least 3 out of the 4 of having appointed a co-leader.  ccNSO has already appointed someone and GAC and ALAC are in their approval processes.  [ALAC chose Christopher Wilkinson.] GNSO choice discussions are ongoing. -- Discussion on overarching issues on categories, which is ongoing.

From the chat:

Alan Greenberg: We have. Christopher [for the Worktrack Subteam 5] Anne Aikman-Scalese (IPC): I believe the plenary said they would seek permission from the 3 GNSO Co-Chair nominees for WT5 to disclosse their names. 2. CC2 — String Similarity — Q 3.4.1 - 3.4.6: Singular/Plural Same Owner Analysis: -- Question from last meeting about how many singulars and plurals of strings ended up being owned by the same entity. -- Looked at currently delegated strings.  There are 21 pairs of singulars and plurals.  The specific question was where singular and plurals and the owners are the same.  There are only 4 pairs of those.  Car/Cars was merger and acquisition; fan/fans was acquisition; riese/riesen was acquisition; appliance/appliances applied with the same ower. CC2 Document: 3.4.1 -- There was a perception that consistency and predictability of the string similarity evaluation needs to be improved.  Do you have examples or evidence of issues?  If so do you have suggested changes to the policy recommendations or implementation that may lead to improvement?  For instance, should the standard of string confusion that the evaluation panel be updated or refined in any way? -- INTA, BC, RySG, BRG, Afilias, and ALAC provided suggestions with improving the string similarity evaluation with respect to singulars and plurals. -- INTA and NCSG recommended a longer period of time between the string similarity review, and the deadline to file a string confusion objection. -- NABP stated that where there is a verified TLD in a partiular industry sector, any string in the same industry sector should have the same or substantially similar restrictions to address possible confusion. -- RySG, BRG, Afilias, and Google recommended eliminating the SWORD tool. -- NCSG provided feedback that string evaluation should be consistent and effective at avoiding confusion and promoting confidence through fair resolution mechanisms. -- Additional suggestions for dealing with string similarities were suggested on the list. -- Reporting back on unified definition or framework to determine whether strings are similar.  Concluded that when you compare two generic terms that aren't tied to any trademark -- trying to develop a uniform framework, such as .hammers against a pre-existing TLD for .hammer is a very different analysis than comparing a pre-existing TLD consisting of a trademark against another TLD consisting of a trademark, but only one of which will be used as a brand -- there is no simple way to do that. -- The idea of tying synonyms together should be left outside of the string similarity issue.  Don't support the comment from NABP. -- One of the factors the NABP comments brought into consider is if there is the potential for confusion the potentially confusing string could be operated in a way that could be designated for a specific community. -- We are talking about balancing the public interest and consumer safeguards.  Not a given that we have to go in one direction or another.

From the chat:

Anne Aikman-Scalese (IPC): Agree with Paul Anne Aikman-Scalese (IPC): For example, what happens when we compare .hardrock with .hardrocks?  (I don't represent the Hard Rock or any rock bands or any rock climbers.) Roger Carney: +1 Michael, on synonyms Michael Flemming (GMO): Potentially confusing, maybe, but not restrictively. Unless the application itself outlined a plan to operate within the field that shared similarity with its synonym, then I could see that, but other than that it is too widestretched. Question: Should we get rid of the SWORD tool? -- Yes, it is worse than useless. -- The idea behind it was to provide an objective measure.  The fact that the scores differed from the similarity panel that was more of an indictment of the similarity panel, rather than the algorithym.  Shouldn't evaluate a tool that wasn't finished.  Should do a trial.

From the chat:

Alan Greenberg: Thanks Paul. Reminds me about an old joke that involves the line that particular information is "100% accurate and 100% meaningless" Paul McGrady: @Kurt - I promised that my description should be highly editorialized. Paul McGrady: @Kurt - how would the algorythm take into account trademark registrations? Anne Aikman-Scalese (IPC): Too bad the tool is o longer live.  We could run hypothetical strings against it to help find examples that might lead to a better tool. Paul McGrady: @Kurt - would the new Sword be for only comparing strings for string similarity and not for rights objection? Anne Aikman-Scalese (IPC): But should we incorporate any new tool into the panel decisions? 3.4.2 -- Should the approach for string similarity in gTLDs be harmonized with the way in which they are handled in ccTLDs? -- SSAC reference advice provided in SAC060: SSAC Comment on Examining the User Experience Implications of Active Variant TLDs Report (23 July 2013). -- Nominet and ALAC supported harmonization. Question: Could there be an appeals process for a TLD being consider in the ccTLD fastrack?  Response: On the notion of appeal.  That came out when the draft AGB was being discussed.  There was a strong request for appeals on the decisions of the review panel, but the notion was denied.  This group could suggest this in implementation.

From the chat:

Emily Barabas: ALAC response to 3.1.1: "The recommendation on string confusion is one that must be enhanced. Singular and plural versions of related strings proved to be problematic in the first round and must be addressed this time. Such provision should not be limited to just the addition of an S but should be more generalized as suggested in a recent Registry SG document. That being said, as discussed in relation the ccNSO Extended Process Similarity Review Panel (EPSRP) document, for strings that are inherently confusing in their own right, but for which STRONG irrevocable policies mitigating against confusion in full domain names, delegation could be considered." Anne Aikman-Scalese (IPC): It would also be pretty useless when there is a trademark involved in one string that has secondary meaning to the public - probably need to just scrap it. Michael Flemming (GMO): @Anne is that in regards to the harmonization with the ccTLD fast track harmonization? Michael Flemming (GMO): sorry, used by noun twice Michael Flemming (GMO): my* avri doria: appeals could apply to that as much as anything, i think. 3.4.3 -- The WG and the wider community have raised concerns specifically related to singles and plurals of the same word.  Do you have suggestions on how to develop guidance on singles and plurals that will lead to predictable outcomes?  Would providing for more predictability of outcomes unfairly prejudice the rights of applicants or others? -- RySG, BRG, Valideus, BC, Google, and vTLD Consortium supported putting singulars and plurals in the same language and script in a contention set. -- Proposal from the Registry based on language and script. -- So confusing for non-native English speakers.  Support the notion of singulars and plurals in the same contention set, but not their equivalents in another language in that same set. -- Based on the application -- if IDNs are thrown together with their equivalent in the same contention set, doesn't that suggest that only one will be delegated? -- Concerned about how we are doing this.  At some point are we going to discuss whether these comments have merit.  I don’t know how we go back to make decisions based on the comments.  -- The idea is that in this pass we are doing the due diligence.  In some cases you can see consensus on the way to go.  The staff is working on a compendium of where we are and where there may be something approaching consensus.  Next each Worktrack goes through its third and deciding pass and decide if we are going to change what exists with enough case, or will we stick with the status quo.  If we are going to change it in what way will we change it.  The next pass is where we have to come to a consensus point.

From the chat:

Anne Aikman-Scalese (IPC): And what do we think about the phonetic equivalent of a plural?  e.g. starz? Michael Flemming (GMO): Theoritically, though, the IDN equivalent of English is recognized Anne Aikman-Scalese (IPC): Another tough question.  What about .star compared to .starz?  And is that confusing with .etoile?completely differently. Kristina Rosette (Amazon Registry): Gg:  If I'm understanding you correctly, wouldn't that mean we would have to do the community evaluation at the beginning (and leaving string similarity and confusion to the end)? Michael Flemming (GMO): I don't want to pick at .shop, but .通販 and .shop are two different words and have different target communtiies. Michael Flemming (GMO): If they are thrown into the same contention set, don't we close ourselves off to allowing for the IDN to go live? Kristina Rosette (Amazon Registry): @Michael :  +1 :-) Gg Levine (NABP): @Kristina: perhaps; that would be one approach. Kristina Rosette (Amazon Registry): I can tell you how we got to the result.  Amazon was one of the applicants. Paul McGrady: I think we need to talk about both contention sets (2 new applications) and string preclusion (1 new application v an already delegated gTLD, which already delegated gTLD wins). Anne Aikman-Scalese (IPC): IN Track 4 we are talking about a registry getting the right to the equivalent idn when it is awarded the English word.  I dont really understand to what degree that is being proposed but it relates to string similarity and contention sets in my mind because the issue is not JUST foreign languages that use the script we use - it's also idns an idn equivalents.  Don't really know how this was treated in 2012 but Work Track 4 is now working on the idn issue.  Would appreciate if Rubens or Cheryl could clarify.