In this comment, I wish to not only answer the specific questions, but
propose a specific solution (with some areas left open).
Before I begin, I would like to mention some of the previous
submissions. Many of them have called for, directly or indirectly, a POC
that directly represents the Internet (or at least the MoU signatories)
on a one-person/one-vote basis, with little or no recognition of any
special constitutencies. I love this idea on its face; however, there is
already a body that serves this purpose--the PAB executive board.
I would like to see PAB and their executive board given substantially
more control under the MoU, with POC remaining a special assembly to
represent special interests. Since there has not yet been any RFC that
relates to this issue, I've left the specifics for a comment to be sent
separately.
If the MoU is otherwise unchanged, I would still like to see POC remain
a special constituency group, but with much more representation from the
end users--presumably by increasing the PAB representation to a 50%
level.
A. BACKGROUND
In general, the principles stated in the background are sound. However,
I would add two additional principles.
First, the terms of POC members should not be too long, to avoid both
inertia and the possibility of corruption.
Furthermore, many of the constituencies to be represented will not be
adequately informed of the MoU and related issues until after the system
has been operational for a short time, so the first terms should be
shorter than later terms. Obviously, the vast majority of most
constituency groups will probably not understand or care about the issue
for many years, if ever; that being said, there will probably be many
more people who want to be involved after the first few years than there
are right now.
The three year term as proposed in the MoU is a good compromise, but
different constituencies may need different specific terms.
Second, it is advisable to have terms staggered. Otherwise, there are
two possibilities after each election--either a massive changeover of
the entire POC at once, which could easily lead to chaos, or domination
of the proceedings by any re-elected members. This current MoU
anticipates this very well, and I have attempted to preserve the
staggering as it is implemented in the MoU in my comments.
B. QUESTIONS
1. How many people do you think can constitute a functioning Policy
Oversight Committee?
20-30. Actually, I believe that 25-35 is a better range; the proposed
solution below includes 30. While it is true that with smaller numbers
"everything runs more smoothly," it is also true that with with smaller
numbers, many important objections are glossed over. A group composed of
16 people is already too large to handle things quickly on a day-to-day
basis (it's already larger than the Los Angeles City Council), and
expanding it to 30 will not make much difference.
2. Which interest classes require representation in the gTLD-MoU Policy
Oversight Committee?
Note that some of these classes require advisory/observatory
representation only, not voting. Furthermore, some of these classes are
adequately covered by others. Please see the proposal below for more
specifics.
End-Users: Yes
Internet Service Providers: Yes
Telecommunications Companies: No
Public Interest Groups: Yes, but primarily those representing the
public's interest as it relates specifically to DNS or at least the
Internet
Business and Advertising Interests: Yes, but primarily those
participating in Internet sales, advertising, and other business
Content Providers: Yes
Internet Technical Infrastructure Interests: Yes
Software Suppliers: Yes
Trademark Community/Law: Yes
Intergovernmental Interests: Yes
Domain Name Registrars: Yes--both CORE registrars and others (especially
ISO TLD registrars, but also lower-level registrars like ml.org and many
ISPs, and NSI at the present time)
Other: Domain Name Holders
Other: Online Services (who could be treated as ISPs and/or content
providers)
3. Which of the listed interest classes above do you think are
underrepresented in the gTLD-MoU Policy Oversight Committee:
End-Users: Yes; although the 2 PAB votes are a start in the right
direction, this is not enough
Internet Service Providers: Yes
Telecommunications Companies: No, overrepresented
Public Interest Groups: Maybe, and possibly represented improperly
Business and Advertising Interests: Yes
Content Providers: Yes
Internet Technical Infrastructure Interests: Yes
Software Suppliers: Yes
Trademark/Legal: No, overrepresented
Intergovernmental Interests: Maybe
Domain Name Registrars: Yes, in that non-CORE registrars are not
represented
Other: Current Domain Name Holders
Other: Online Services, possibly
4. For each of the interests classes you have indicated in the gTLD-MoU
Policy Oversight Committee:
The answers to these four specific questions are better answered below,
in my specific proposal.
5. Please discuss below any other considerations vis-a-vis the structure
of the gTLD-MoU Policy Oversight Committee not covered in this
questionnaire.
I would like to propose, as a starting point for a conversation, the
following 30-member POC.
Note: If this table does not line up properly, try viewing it with a
fixed-width font.
Community Representatives
--------- ---------------
End Users 9 (5/4) 2 years
Domain Name Holders 5 (2/2/1) 3 years
CORE Registrars 3 (1/1/1) 3 years
Other Registrars 2 (0/1/1) 3 years
IANA & Numbering Agencies 3 (1/1/1) 3 years
ISOC 1 (0/0/1) 3 years
IETF 1 (1/0/0) 3 years
ISPs/NSPs/Online Services 3 (1/1/1) 3 years
Content Providers 1 (1/0/0) 3 years
Trademark Interests 1 (0/1/0) 3 years
Internet Businesses 1 (0/1) 2 years
General Notes:
--------------
I do not believe in term limits; if people want to keep electing the
same representatives, I don't see any reason to prevent it. Generally,
the people who clamor for term limits are the same people who continue
to reelect the same representative until she is forced out by term
limits, such that term limits tend to have the exact opposite effect
they were intended to have: increased inertia, less direct control
exhibited by the voters, etc. However, I know that many people like the
idea of term limits. So, while I wouldn't take a strong stand on the
issue, I would recommend no term limits for any of the elected
positions. For some of the appointed positions, however, I do recommend
term limits, as stated below.
I do believe in recall proceedings. When properly handled, I believe
that recalls increase voter participation. However, recalls need more
than a 50% majority of the constituency, probably a two-thirds majority.
I would recommend recalls for all of the elected positions. For
appointed positions, I'm not sure.
Finally, there should be some non-voting representatives to act as
observers/advisors, representing ITU, WIPO, the World Court, the US
Government (for one term only), EFF, and other relevant governmental,
inter-governmental, and non-governmental organizations.
End Users: 9 representatives
----------------------------
The end users at large are the most important constituent community.
However, there are legitimate concerns that end users may not
necessarily select people with the proper technical or policy knowledge.
Based on the current leadership in PAB as compared to iPOC, this does
not appear to be a problem now, but it could very well be a problem in
the future. I believe that slightly less than one third of the votes in
POC is a good compromise.
As mentioned earlier, I would also like to see additional changes to the
MoU structure which might allow a reduction in PAB's representation in
POC, such as veto power for PAB over important decisions, and the power
to override vetos (by a larger vote) from POC or from IANA/ISOC (if they
retain veto power).
PAB can be used represent these members, but it would be better if a few
changes were made to PAB. For example, under the current rules, a single
person could own hundreds of organizations, sign hundreds of times, and
get hundreds of votes. Also, it would be best if the requirement to sign
the MoU were removed and replaced with a requirement to sign only a
less-specific statement of principles. If PAB is not used to represent
the end users, this should be reduced to 7 members elected by end users
and 2 elected by PAB.
Assuming that PAB is so used, it will be PAB's responsibility to
determine nomination and election procedures. I have two suggestions.
First: Anyone who receives nominations from .1% of PAB (or 2 members, if
PAB is under 2000) is placed on a ballot. The nominees are then put up
for a vote among all members. Second: A parliamentary election with
proportionate representation among slates may be a better or worse
option, depending on whether or not there appears to be a tendency
toward party or slate alignment within PAB. This is up to PAB to decide.
PAB representatives should be elected to 2-year terms. These terms
should be staggered by electing 5 of the first members to 1-year terms
(presumably those who come in 5th-9th in the voting) and 4 to 2-year
terms (presumably those come come in 1st-4th).
If PAB does not turn out to be a good way to represent the Internet
community at large, an electronic vote open to all people with email
addresses could be used instead, along the same lines as above but
without PAB involvement.
Domain Name Holders: 5 representatives
--------------------------------------
Domain name holders are probably the next most important constituent
group in a DNS system. The same caveats about technical considerations
apply here as in the end user representation, but I believe that this
can be alleviated by using the voting scheme below.
"CORE direct" domain names will contribute four votes, two to be cast by
the technical contact for the domain and two by the administrative
contact. "Direct" domain names are those which are assigned directly by
a registrar; "CORE direct" domain names are SLDs directly under a
CORE-gTLD. Since those participating directly in the system have the
most at stake, their votes should carry more weight.
Other "direct" domain names will contribute two votes, one to be cast by
the technical contact for the domain and one by the administrative
contact. A non-CORE direct domain name is one which is assigned directly
by a non-CORE registrar--that is, an ISO TLD registrar, NSI (at
present), or a lower-level ISO registrar (as appropriate). Examples of
non-CORE direct domain names include x.com, x.co.uk, x.nu, x.la.ca.us,
etc.
"Indirect" domain names will contribute one vote. An "indirect" domain
name is one which is assigned under another domain by the holder of that
domain. This mostly includes domain names under ISPs' direct domains,
but also includes certain special domains. Examples include x.ml.org,
x.thirdwave.net, etc. Since indirect domain name holders do not come
into direct contact with the DNS as a whole, their votes should carry
substantially less weight.
Since many indirect domain names are held by the same entities that hold
the immediately-superior domain name (the obvious case being www.xx.xx
under xx.xx), no entity, and no two related entities (a term which has
to be rigidly defined, but the idea is multiple ownership by the same
people), can have a vote for two indirect domain names under the same
domain name, or for an indirect domain name and any domain name under
which it resides.
I don't think there are any similar issues with direct domain names, but
if there are, the same rules can be used to resolve them.
We may want a maximum number of votes to be cast by any single entity
(e.g., a person with dozens of domain names, or a technical contact
working for dozens of different entities, each with their own domains).
This is something I'm not sure about.
Domain name holders' representatives would have a three-year term. For
the first term, of the five, two would be elected to a one-year term,
two to a two-year term, and one to a three-year term.
Registrars: 5 representatives, 3 CORE, 2 non-CORE
-------------------------------------------------
Registrars are the core of DNS administration (sorry for the pun), and
need to have substantial representation in the running of the system.
Since the CORE registrars have the closest connection to the system, and
the largest stake in it, their concerns should have more weight than
outside registrars. However, other registrars--especially, but not only,
the ISO TLD registrars--also have some relevant input.
CORE's 3 registrars should be elected by CORE. It's their responsibility
to come up with a procedure, but I would suggest a direct election, one
registrar/one vote, among all people with two or more nominations by
CORE members.
Each CORE representative would have a three-year term. The initial three
representatives would have one-year, two-year, and three-year terms.
The non-CORE registrars mostly consist of the ISO TLD registrars and
subsidiary registrars underneath the ISO TLDs (such as ca.us, la.ca.us,
etc.), but also include NSI, for as long as they manage their current
TLDs, and any other registrars who now or in the future manage special
TLDs like .gov, .mil, .int, etc. In addition, special registrars like
ml.org who provide domain name service under SLDs and 3LDs may be
included, as their service is technically similar in many ways to the
service provided by gTLD registrars, and they often conduct valuable
experiments (such as dynamic domain name service).
The issue of registrars from outside "alternative RSCs" is a tricky
issue, and needs to be discussed. At present, I do not believe that they
need representation. However, since I strongly suspect that the
situation of these alternative RSCs will be much different in a year
than it is now, it may be best to wait and see.
One representative would be elected by the non-CORE registrars on a one
registrar/one vote basis. However, this would provide a heavy bias
toward registrars under ISO domains with either significant delegation
or fragmentation (such as .us) and those with open shared registration
(such as .uk). To counterbalance this, the other representative would be
elected on a one TLD/one vote basis. It would be up to the registrars
for a TLD to decide among themselves how to cast their vote for this
representative. For example, Nominet may wish to have a vote among its
members to decide on how to cast its vote, while the .us TLD may wish to
have their single top-level registrar cast the vote for the entire
domain.
Both non-CORE registrar representatives would have three-year terms,
except that the initial per-TLD representative would have a two-year
term.
IANA & Numbering Agencies: 3 Representatives
--------------------------------------------
The agencies that handle IP number allocations have a function that is
closely related to DNS, and clearly have an important stake in DNS.
IANA is intended to be the "top level" of the system, working with ARIN,
RIPE and APNIC.. Furthermore, IANA is currently charged with the
responsibility of maintaining or delegating the DNS root functions.
However, because IANA already has special representation in the MoU
system, they don't need much additional consideration in POC.
To represent IANA and the numbering agencies, three votes should be
sufficient.
One representative would be appointed by IANA to a three year term. A
second would be selected based on a consensus (or, if worst comes to
worst, a vote, with IANA's vote used as a tie-breaker if necessary)
among IANA and the numbering agencies, with a three year term, shortened
to one year for the first term. A third would be selected by the
numbering agencies without IANA input, with a three year term, shortened
to to two years for the first term.
ISOC: 1 Representative
----------------------
ISOC needs representation in POC, as well as their special role outside
of POC. However, because of that special role, there is no need for more
than one representative. ISOC representatives would serve a three-year
term, including the initial representative.
IETF: 1 Representative
----------------------
Some representation is needed for the large set of people, many of whom
are dedicated volunteers, involved in the technical functions that make
the Internet work.
IETF is probably the closest thing to an organization of such people.
While IETF is an open organization, and some have argued that there is
no way to stop outsiders from "interfering" with their votes for their
representatives, for lack of any better way to represent these
interests, IETF will serve.
IETF's representatives would serve a three-year term, with the initial
representative serving a one-year term.
ISPs/NSPs/Online Services: 3 Representatives
--------------------------------------------
ISPs and NSPs are the backbone of the Internet (sorry, another pun), and
are also crucially linked to the DNS, whether or not they function as
registrars. Online services such as AOL and MSN are a special subset of
ISPs, and can hopefully be included under the same representation
(although they also obviously function as content providers, and will be
represented there). Some divisions and functions within large
corporations and educational institutions also fall into this category.
There is an important balance between small ISPs, who are often the
first to experiment with new services; large ISPs and NSPs, who provide
substantial bandwidth; and huge ISPs and online services, who provide
service for a large fraction of the Internet's members. Therefore, the
three representatives could be elected in different ways. One would be
selected in a simple one ISP/one vote election. Another would be
selected by bandwidth, perhaps in a one T1/one vote election. The third
would be selected in a one user/one vote election (where a user is
anyone who derives service from the provider, whether an end-user,
another ISP, or otherwise).
ISP representatives would serve a three-year term, except that the
initial per-user representative would serve a one-year term and the
initial by-bandwidth representative would serve a two-year term.
Content Providers: 1 Representative
------------------------------------
Web pages continue to become more important, and in most people's minds
a domain is associated directly with the web site or sites it hosts. The
content providers, especially web content providers, are obviously an
important part of this, and a major factor in our current need to expand
the DNS name space.
The question becomes, if we are going to represent content providers,
how do we do so? This it the one set of representatives that I'm most
ambivalent about, but there seems to be a major clamoring on their
behalf, so I'm inclined to believe in my initial instinct--that they
deserve representation.
First of all, should we distinguish between major content providers like
CNET and MSN, providers of company-specific content, informal content
providers (like those behind personal home pages and fan pages, which
make up a significant percentage of all web pages), and basic web page
publishers? If so, where is the line drawn?
Furthermore, who exactly are the content providers? In simple cases
(such as many fan pages and all personal pages), one person or one small
group does all of the work. But what about something like CNET? The
people who gather the information and pictures, the people who write and
edit the stories and edit the pictures, and the people who format and
maintain the web pages may be entirely different people.
The simplest way to resolve this issue seems at first glance to go to a
by-page basis: Each web page on the Internet gets one vote; the
webmaster who actually maintains the page or the person who hires that
webmaster uses the vote. However, the flaws in this solution are
obvious.
As I said before, I'm not sure these issues can be resolved and content
providers can be given representation in any reasonable way; hopefully
someone can come up with a good answer.
This representative would hold a three-year term, with a one-year
initial term.
Intellectual Property/Trademark Interests: 1 Representative
-----------------------------------------------------------
The intellectual property and trademark interests need to be directly
represented in POC--but not as strongly represented as they currently
are in the structured described by the MoU. I am unsure as to the best
way to handle this representation, so I suggest an INTA representative,
with a three-year term, with the initial term shortened to two years.
Internet Businesses: 1 Representative
-------------------------------------
While Internet business is an important constituency, it's also a hard
one to define. For lack of a better definition, I would classify any
company that spent any money advertising on the Internet as an Internet
business, and give them a vote per dollar spent on Internet advertising
toward selecting a single representative.
Because of the rapidly changing character of Internet business, this
representative would serve a two-year term.
At some point, it may be desirable to specifically represent other
classes of Internet-related business. As on-line shopping and other
forms of commerce become more prevalent, we may want to either add
another representative, or adjust the election of this representative,
to better represent internet shops, banks, credit agencies, etc.
However, it should be kept in mind that none of these functions has much
to do with DNS, while Internet advertising often does (certainly, nobody
else could have more useful input into what new TLDs would be most
marketable).
Final Distribution
------------------
To review the distribution of terms, I will reprint the table from
above.
Community Representatives
--------- ---------------
End Users 9 (5/4) 2 years
Domain Name Holders 5 (2/2/1) 3 years
CORE Registrars 3 (1/1/1) 3 years
Other Registrars 2 (0/1/1) 3 years
IANA & Numbering Agencies 3 (1/1/1) 3 years
ISOC 1 (0/0/1) 3 years
IETF 1 (1/0/0) 3 years
ISPs/NSPs/Online Services 3 (1/1/1) 3 years
Content Providers 1 (1/0/0) 3 years
Trademark Interests 1 (0/1/0) 3 years
Internet Businesses 1 (0/1) 2 years
There are 30 representatives, 20 with three-year terms, 10 with two-year
terms. Because of the staggering, there will be a total of 12 elections
every first and second year, 11 elections every third year.