On the "Consumers Constituency" Question

[Eric Brunner-Williams <ebw@xxxxxxxxxxxxxxxxxxxx>]

Robert,

First, I want to state that I think the Board, and the RySG errored 
formally, and many in the GNSO Council errored informally, in deciding 
not to accept the application for municipalities, and quite possibly 
similar entities, for Constituency Status, subsequently Stakeholder 
Group Status.
The issues are several.

First, each municipality is served by a small number of access network 
operators, unlike the generic and sponsored TLDs, with the notable 
exception of .cat. Cities resemble countries in this fundamental 
aspect, and therefore city registry operators are just as capable of 
originating policy about access network operator behavior, e.g., 
monitizing NXDOMAIN return, as highly policied ccTLD registry 
operators. This means that unlike all but one of the current RySG 
members, cities have a potential for a policy based relationship with 
the ASO and its dependent Regional Internet Registries (RIRs) and 
Local Internet Registries (LIRs), which cannot be shared with the 
current RySG, less the sole exception of the .cat operator.
As a rule, the GNSO has no interest in what happens in the ASO, and no 
technical dependency upon the correct function of the RIRs and LIRs, 
except when we examine edge cases such as double-fast-flux hosting 
networks. This is not the case for Cities, for which each will be 
intimately familiar with the operation art, for good, and for bad, of 
the access network operators present in the City access network market.
Second, the 2000 applicants, NeuStar/NeuLevel, Afilias/TuCows, 
Register.Com, and Global Name Registry, and also SITA, COOP, and 
Museum, all had access to the policy making process prior to Louis 
finishing the contract process. This was both equitable, reflecting 
the investment of each in the process and its outcome, and also 
prudent in terms of then-present and latent competition policy issues, 
and Verisign/NetworkSolutions was formally, and substantively, 
accommodating to the technical and policy interests of the 2000 round 
applicants. I repeat for emphasis -- this was before the 2000 round 
applicants had acceded to "contracted party" status.
As a rule, "observer status", whatever that may be, is not equal to 
the status of Constituency, now Stakeholder Group member, and the 
issues on which the RySG comments profoundly affect the interests of 
the current round of registry applicants.
To restate, there is a substantive difference between locality defined 
registries and registries which are indifferent to locality, and there 
is a procedural difference between "observer" and "member", which has 
the potential at least of substantive consequences.
Before turning to the surviving "reform" Constituency application 
there is the question of how many Constituencies there are in fact.
At the Paris meeting the correlation between the votes of the entities 
which are currently afforded "Constituency Status" was published. The 
results were illuminating. It is counter-factual to claim that the 
Business Constituency, or the ISP Constituency, actually exist as 
policy originating elements of the GNSO Council. The facts available 
to all of us demonstrate that the IPC captured the BC and ISPC several 
years ago, as those of us who have been involved in the perenial WHOIS 
efforts have observed. The correlated data is incontrovertible, and 
the continued gift of "Constituency Status" to two shell 
"constituencies" is an error that can be corrected.
Finally, as to the Consumers Constituency Charter, as a continuation 
of Ira Magaziner's rather limited vision as to what constitutes the 
"stakeholders" in the DNS, the proposal to add a vehicle for the views 
of non-commerical "consumers" is hard to fault. However, we should not 
overlook the opportunity to review the wisdom of Ira's choices.
The DNS may be thought of as a wart on the transport layer (TCP, UDP, 
etc), just as ICMP, another protocl which has a redirection header 
allowing mapping, is a wart on the network layer (IP), and ARP, yet 
another mapping protocol, is a wart on the link layer (ENET).
The fundamental elements of domain policy are:
o label custodial, a function exercised reasonably well by the IETF 
through the periodic IDN Working Groups[1], to which the W3C and the 
UTC liaise[2][3],
o name space custodial, a function exercised reasonably well by each 
of the IANA and CNNIC, and their immediate delegees, aka "TLD 
registries",
o recursion custodians, a function exercised reasonably well by 
implementation vendors and their installed user base, some of which 
are nominally members of the previously observed dysfunctional ISPC, 
and by
o static address resource custodians, a function exercised reasonably 
well by hosting operators, generally absent from the GNSO Constituency 
model.
These all assume the existence, and policy neutrality of transit 
operators, peering points, and routing policy generally.
In addition to these fundamental custodial activities arising from the 
functional definition of the DNS -- a mapping mechanism between names 
and addresses, there are social policy elements:
o string custodial, a function arising from the social policy of 
applying trademark law[4] to domain names, a function exercised 
reasonably well by the Intellectual Property Constituency, now Stake 
Holder Group,
o string custodial, a function arising from the social policy of 
applying free speech law [5] to domain names, a function exercised 
reasonably well by the Non-Commercial Users Constituency, now Stake 
Holder Group, and
o access custodial, a function arising from the social policy of 
applying competition law [6] to domain names, a function exercised 
reasonably well by the Registrars Constituency, now Stake Holder Group.
Absent from Ira's model are the fundamental users of domain names, the 
applications which actualy attempt to locate resources using the DNS 
mapping function.
There is no place for stub resolvers and the very large class of 
applications which source resolution requests, in particular, using 
HTTP (browsers generally, but for reasons which arise from the 
abandonment of address policy, see ISPC remarks, above, has become a 
transport protocol in its own right) or SMTP. Absent also from this 
model are the fundamental means to authenticate and assure the 
instances of communication mapped through the DNS. It may be 
consequent to the government policy at the time which attempted to 
prevent "strong crypto" from being adopted in the civilian sector [7][8].
Browser and email user agent and transport agent technology and 
service vendors and end-to-end security technology and service vendors 
lack Constituency Status in Ira's model of a bottom-up public-private, 
multi-stakeholder organization. Consequently Ira's model has grown 
warts of its own -- an SSAC function and a recurring "maleware and 
anti-phishing" requirement, now an "Over Arching Issue". Abuse is 
pervasive, but no Constituency owns abuse as an issue.
There is a big gulf between connectivity and resolvability, which is 
determined by architecture and engineering, and access control, which 
is determined by administrative policy. ICANN lacks a Constituency, or 
a Stakeholder Group, or a Supporting Organization, which does general 
access control. The limited access control functions exercised by the 
IPC is simply too limited to address the more general areas of actual 
abuse.
The Constituency Charter proposes, at 11.2.3, to find funding "From 
major actors in the industry, including without limitation, 
registrars, registries, Internet service providers, and so forth" and 
at 11.2.4, "From the Non-Commercial Stakeholders Group (NCSG) and 
ICANN". Neither of these approaches could lead to an independent or 
sustainable Constituency. This weakness is not cured by 11.3, 
Conflicts of Interest.
The Constituency Charter anticipates a rather small number of members 
participating in the activities of the Constituency, as it devotes all 
of 10 to standards and discipline. If there are "users" the tools to 
manage their self-expression to their best ends is not a 
"Communications Officer" deleting links and comments. This simply 
can't scale, and seems irreconcilable with 9's statement on outreach 
"...to be as inclusive and representative of non-commercial Internet 
users as possible ..."
For the reasons I've mentioned I suggest that the Board thank the 
proposants for their efforts and decline the offer to create this 
particular Constituency.
I remind the Board that there is a "public interest" discussion, and 
that recovering the ISPC and the BC to their original purpose, as 
sources of agency distinct from the IPC is of paramount importance in 
the reform of the GNSO.
Please see also 
http://forum.icann.org/lists/affrev-draft-processes/msg00000.html and 
http://forum.icann.org/lists/affrev-draft-processes/msg00007.html for 
two points of view on the "public interest" which seem to me to be 
closely related to the question of a "consumers" source of agency 
within the GNSO.
Submitted as a personal capacity, though I am employed by CORE which 
has an interest in some issue outcomes, though probably not this one.
Eric Brunner-Williams