[NCAP-Discuss] [Ext] Re: An Approach to Measuring Name Collisions Using Online Advertisement

[rubensk at nic.br]

> 
> 
> This is certainly one manner of thinking about the problem, but I do not agree with the idea that "we didn't hear about it in the way that we might have expected, so it must not have been a problem."  That dismissal is convenient but not data-based.  It is true that there are limits on what can be definitively learned, including memories of involved individuals, willing participants, available data, etc.  However, there are clear indicators--both in the ICANN reports and the survey data--that there were problems.  And the historical mapping and root server query data showed that name collisions potential issues are widespread.  My research thus far has generated as many questions as answers, and there is much more to uncover.

Just to add some uncertainty to the issue, there were voluntary measures both in prevention and mitigation that were also part of the 2012 roll-out. For instance, our 2012 TLDs reserved the top-N strings in the 2nd level listed in the JAS report and the New gTLD Applicant Group (NTAG) list of troublesome strings (including the infamous WPAD). Also, at least one TLD responded to a name collision issue by disabling controlled interruption for that TLD.


> 
> I worry about calling previous round "successful."  How is success defined?  Is it related to the number of name collisions reports?  Or is it related to the presence or absence of uproar in IETF or *NOG?  Or is it that the Internet didn't stop?  Some of these are very subjective, and some are fraught with bias, for various reasons (e.g., the text on the name collisions submission site itself). There are certainly advantages to the approach used in the previous round, but suggesting that it was successful--at least without any definitive metrics on which to base that subjective description--is counter-productive.

There is one success metric we can attach to the 2012 process, which is the 0 occurrences of life-threatening name collisions. Besides ICANN not knowing of such, this would be much harder to not trigger fall-out if such happened.


> 
> Finally, with regard to the use of "science experiment" as a description, I agree. However, that's exactly what controlled interruption has been!  I don't disagree with improvements over time, but I believe that the data needs to be analyzed.  That means that we can't simply try something and then call or successful--or alternatively call it unsuccessful and replace it--unless we have meaningful data to analyze.  We have *some* of that now, but I believe that there is more to be done to assess previous-round delegations, including controlled interruption.

I believe that description to only apply to the ad network part, but the OP might want to clarify that.



Rubens

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 529 bytes
Desc: Message signed with OpenPGP
URL: <https://mm.icann.org/pipermail/ncap-discuss/attachments/20220611/d8c7f2a4/signature.asc>