With regard to point 3 ("Geographical names are not Generic in the usual semantic or, indeed, ICANN sense. Nearly all of those names are specific to places, cultures, regions, communities and their local economies. The fact that GNSO has taken the lead in the PDP WT5, does not diminish in any way the specific characteristics of geo-names."):
We should avoid getting drawn into a policy conflict disguised as a semantic argument. The term "generic" as used in "generic TLDs" has long since departed from any "semantic" meaning of generic. Even as far back as 2001, RFC 3071 recognized that "generic" was ambiguous when applied to gTLDs, whic could be "generic" as in "purpose neutral" or "unbranded and open for use in any way" or "generic" as in "purpose-specific" or "related to a particular genus of registrants." As a result of the 2012 New gTLD round, we now have hundreds of gTLDs that are .Brands. Brands are conceptually and semantically the opposite of generic. We have dozens of TLDs being used as "geo-names." We also have many that are "purpose specific" and many that are open to use by all without any type of "genus" implied or expressed. I'm sure there are other types and distinctions to be made, but ultimately these are all gTLDs.
The same word or string can have multiple meanings. In numerous cases, the same string could be delegated and used as a generic, purpose-specific, .Brand or geo-name gTLD, depending on the applicant's plans. We need to be conservative with the idea that a string or word is inherently one thing or another. For instance, "tours" could be a .brand, a geo-name, a purpose-specific gTLD or even a purpose-neutral gTLD.
The GNSO's role here is not one where it has "taken the lead" as a free-floating "fact" but one where it has that responsibility as a matter of ICANN policy.
In that vein, this is not "PDP WT5"; this is a "work track" within a GNSO Policy Development Process Working Group. The GNSO is a task-based entity, with that task being to manage the process of developing the policy recommendations for gTLDs. Anyone can participate in that process. It is irrelevant whether they are a member of (or a stakeholder represented by) any GNSO Stakeholder Group or Constituency.
That is not to say this is the only possible way in which ICANN could have been or could be constituted. Before there was a GNSO and a ccNSO, there was a DNSO, which "advise[d] the ICANN Board on policy issues relating to the domain name system (DNS) -- the system of names commonly used to identify Internet locations and resources." Some body could decide to re-imagine this structure yet -- reconsider what is a gTLD and what is a ccTLD (perhaps based on use and purpose than on the ISO 3166 two-letter list), and whether other categories should be recognized aside from these two. But this body is not that body. And this body has enough complexities and distractions to complicate and delay our work without getting into existential debates -- especially those that are firmly outside our remit.
It might be interesting to establish an unchartered discussion group to have those existential debates. However, it is the farthest thing from interesting (not to mention, productive) to turn this Work Track into that discussion group. I strongly encourage us not to get drawn down that rabbit-hole, which is in fact a rabbit-warren with a multitude of holes.
Greg