RE: [gnso-idng] rethinking IDN gTLDs

["Gomes, Chuck" <cgomes@xxxxxxxxxxxx>]

The concepts of "intentional similarity" and "cooperating actors" seem
very useful in this regard.  I don't think that either could be
determinant in themselves, at least no "intentional similarity", but it
appears that they could be key factors in evaluating possible harm.

Chuck 

> -----Original Message-----
> From: Eric Brunner-Williams [mailto:ebw@xxxxxxxxxxxxxxxxxxxx] 
> Sent: Tuesday, December 01, 2009 11:22 AM
> To: Gomes, Chuck
> Cc: Avri Doria; gnso-idng@xxxxxxxxx
> Subject: Re: [gnso-idng] rethinking IDN gTLDs
> 
> Gomes, Chuck wrote:
> > Avri,
> > 
> > One of the main purposes of the restriction on confusingly similar 
> > strings was to avoid user confusion.  We talked about that 
> a lot.  If 
> > the chances of user confusion are null, why would the strings be a 
> > concern?
> > 
> > Chuck
> 
> 
> The property we (Edmond and I) were attempting to convey to Peter and 
> Paul circa the Delhi RyC meeting, was that the definition of 
> "confusingly similar" is incorrect where it finds any two SC/TC 
> equivalent sequences "confusingly similar". The correct definition is 
> that there are SC/TC equivalent sequences which are "non-confusingly 
> similar", to the SC/TC informed user, which the authors of 
> the current 
> definition are not.
> 
> More generally than to just the SC/TC equivalence issue, 
> dealt with at 
> the CDNC level by bundling, and present in the "variants" issue for 
> some IDN applications, now in DAGv3, is intentionally similarity, 
> where the process described in the DAGvX is incapable of being 
> informed that intentional similarity is present, and not 
> presumptively 
> harmful.
> 
> My point is that we're not simply discussing the chances of user 
> confusion being harmful to the user, and unlikely, but also 
> where user 
> confusion is beneficial to the user, and at present, presumed a harm.
> 
> There is "confusion" where the "color" and "colour" registries are 
> operated by non-cooperating actors (not the policy choice by the CDNC 
> members), there is not where the same two registries are run by 
> cooperating actors, and it is likely that if the actors are infact a 
> single operator, that harm does not result from the existence of a 
> "color" and "colour" label.
> 
> The Anglo vs American spelling was chosen as an illustrative example.
> 
> The absurd case is an applicant for "color" and "colour" going into 
> auction against itself to determine which member of this string 
> contention set shall be entered into the root, at the permanent 
> exclusion of all other variations of "colo[u]r".
> 
> Some knowledge, if available early in the evaluation process, 
> leads to 
> better outcomes than its concealment until after it becomes useful.
> 
> Eric
>