Re: [ga] Re: ICANN before the US Senate...

[Karl Auerbach <karl@xxxxxxxxxxxx>]

On Wed, 6 Aug 2003, Stephane Bortzmeyer wrote:

> But let's stop that "Stick with technical matters" b...t.

I say "Let ICANN stick with technical matters."

As you say, some matters do have policy implications - in which case they 
are policy matters and better left to bodies that actually try to 
encompass the parties affected by the parties.  And I believe there is 
overwhelming consensus that ICANN doesn't fit anybody's definition of 
encompassing all the parties who are affected by its decisions.

This is why, in my senate submission -
http://www.cavebear.com/rw/senate-july-31-2003.htm - and in other
materials, I have suggested a guideline to decide when a matter is
primarily technical or when it has enough policy components that it ought
to be handled by a structure rather more synoptic than ICANN.

Let's look at the cases you mention:

> 1) Who will decide of the redelegation of a ccTLD? Especially in
> complicated cases like a fight between two local groups? This is not a
> clerical process, it is a political one, by essence.

I was, and continue to be concerned, that ICANN is being used by political
groups in countries as a means of fighting against other political groups
in those countries.  Recent examples where I felt concern in this regard
were when we (ICANN) redelegated the cc administration for Afghanistan and
Sudan.  I would say that it is the rare entity that is good at both 
writing down protocol parameters and deciding which among competing, and 
perhaps warring, groups deserves the nod to literally own a country's 
ccTLD.

ICANN's staff does like to travel around the world and act as the arbiters
of who is the best government for countries - some of the staff reports I
saw were gaggingly paternalistic in that regard.  I felt that the
imperialist days of Queen Victoria and King Leopold had returned and
ensconsed themselves in Marina del Rey.

> 2) Who will decide what gTLD to create?

Again, the skills used to allocate protocol numbers are not necessarily
the same when deciding how to regulate a complex economic system.  Many
analogies have been drawn between DNS TLDs and the radio frequency
spectrum allocation system.  Two points are most visible in that
comparision:  First is that spectrum, as opposed to TLDs is actually
allocated, TLD allocation has come to a standstill.  And second, there is
actually a technical component in frequency matters - there's not really
much of a technical question whether the net will go unstable, in the
sense that packets stop flowing and existing names stop resolving, if a
new TLD is added.

> 3) Who will decide to recognize (or not) Afrinic as a new RIR?

This was actually the subject of my last conversation with Jon Postel.  
The throught was that RIRs are logically representative of lumps of
connectivity - that, like Wirth's concept of modularity based on minimal
information flow between modules - RIRs are best defined by clumping
chunks of address space so that the richness of the routing infrastructure
between the chunks is minimized.

The bodies best suited to answer that question are the RIRs themselves - 
they know when, from the point of view of the net's routing 
infrastructure, when it is better to split a RIR or to coelesce.

		--karl--