RE: [council] Progress on new gTLDs

["Bruce Tonkin" <Bruce.Tonkin@xxxxxxxxxxxxxxxxxx>]

Hello Avri,

> > - the evaluation of the first phase of the proof-of-concept was 
> > completed in July 2004
> 
> Was the pointer to this evaluation included in Olof's report. 

Yes - see section 10 of Olof's report:
>From the report by Miriam Shapiro (Summit Strategies International)
"Evaluation of the New gTLDs: Policy and Legal Issues", 10 July 2004 

http://icann.org/tlds/new-gtld-eval-31aug04.pdf 


> And who approved it?

The report was presented at the ICANN public forum in Kuala Lumpur.
http://www.icann.org/meetings/kualalumpur/captioning-public-forum-23jul0
4.htm

I don't see a record of the report being formally accepted or acted upon
in the Board minutes.

It has never been discussed within the GNSO.



> 
> 
> Other then the recommendation that is should be bottom up and 
> user driven, a good idea, has the council made done any work 
> on the structure of evolving namespace?

The conclusion of the work discussed in:
http://www.dnso.org/dnso/notes/20030612.gTLDs-committee-conclusions-v7.h
tml 
Was that the namespace should not be "structured".  Ie it was left to
third parties to propose appropriate names.

However I gather than in the process of reviewing the latest round of
"sponsored" TLDs, the choice of the new strings was the most sensitive
topic.   This essentially moves from technical coordination (e.g
ensuring DNS stability and security) to content issues with respect to
the content of the string.

E.g should <insert country name or region name> be allowed, and if so,
who should be allowed to run it?

E.g should <insert minority group name> be allowed, and if so, who
should be allowed to run it?

E.g should <insert sensitive moral topic name> be allowed?


So it may be that instead of asking whether the namespace should be
structured, we should ask whether any strings should not be allowed in
the root.     Unfortunately an answer to this question generally moves
us into content issues.

Perhaps we need a dispute resolution approach for when a new TLD is
proposed.

Note that the WIPO-II recommendation with respect to country names etc
is another example of moving into the content area.
From:
http://www.icann.org/correspondence/gurry-letter-to-cerf-lynn-21feb03.ht
m 

"You will recall that this Second WIPO Process concerned the
relationship between domain names and five types of identifier, namely,
International Nonproprietary Names for pharmaceutical substances (INNs),
the names and acronyms of international intergovernmental organizations
(IGOs), personal names, geographical identifiers and trade names. The
Member States of WIPO recommended that two such identifiers should be
protected against abusive registration as domain names, namely, the
names and acronyms of IGOs and country names (being one particular type
of geographical identifier). Details of the two recommendations of the
Member States are set out in the ensuing paragraphs."





> 
> I understand there are operational constraints on how many 
> could be introduced at a time.  But I don't have this 
> information.  Is this info avaialble, or does someone need to 
> research that operational constraint.
> 

This is an area where there have been a few statements from groups such
as the IAB, and the SSAC but there is not an easily accessible
description of the problem.  We can ask staff to collect this
information for us.

To give a high level answer. The issues are:
- how to manage adding new tlds to the root (ie the approval process)
- how to maintain the TLDs in the root (general IANA issue)
- the additional load placed on root servers from new TLDs

With respect to the last point, the load is not linear with the number
of new TLDs.  It is more related to the "use" of the new TLDs - ie how
many end user devices will be using a particular new TLD, and how
effective is the DNS resolver software at caching answers etc.   I
gather from earlier discussions with root server operators, that a
significant proportion of their traffic is a result of misconfigured
software on end user equipment.


> > If the outcome is
> > that only a limited number of new TLDs can be introduced at a time -
> 
> 
> Seems almost certain that it would be.  No matter how big the 
> limit, there are organizational scalability issues that would 
> prevent too many from happening, in a controlled and stable 
> manner, in any given unit of time.

Exactly.


> 
> > then we need to consider how they will be allocated (e.g ballot, 
> > auction, first-come first served).
> 
> If we define criteria for a successful application, then it 
> seems reasonable to take them on a first come, first served 
> as long as it meets criteria.  Using a ballot, leaves the 
> selection open to subjective and interest driven decisions.

Experience with operations within new gtlds is that first-come
first-served does not work when many valid applications are ready at the
same time.

E.g if 1000 people meet the criteria for .web, then they will slam the
registration system to be first.

> 
> As for an auction, this might be a good idea if the proceeds 
> were to be used for appropriate charitable purposes, such as 
> capacity building and education.  But it might prevent the 
> less well heeled from obtaining TLDs.  I am personally 
> uncomfortable with only granting new TLDs to the richest applicants.
> 

Yes those are some of the issues with auctions.  Ie what happens to the
funds, and how do you support those with less money.   It is possible to
use a combination of techniques  (e.g I understand the USA has a green
card lottery process in addition to its normal stringent processes for
getting permanent residency).

Regards,
Bruce