Another thoughts on TRANSITIONAL

[Eric Brunner-Williams]

Andrew Sullivan wrote:
> On Sun, Dec 06, 2009 at 07:51:18AM -0500, Vint Cerf wrote:
> 
>> under IDNA2008 rules on a registry by registry basis (I mean registry  
>> in its
>> most general sense, not just TLD)
> 
> The overwhelming majority of DNS operators -- i.e. "registries" -- in
> the world are blissfully unaware of ...


It is also the case that the very large set of operators use only LDH 
in forming labels.

Either the assertion that the problem is scoped is true, or it is not.

If the problem has scope, that is, problematic values are not 
sprinkled at random across under some 300 distinct graphs sharing the 
property of being anchored in the IANA root -- .mil and the rest of 
the odd balls included, and are also not sprinkled at random across 
any other collection of distinct graphs having some other anchor, and 
even those irritating pseudo-graphs injected by amusing ISPs that 
appear to have anchors, yet do not (that's "stupid browser hacks" for 
Yoav and Subramanian), then the absence of signs of intelligent life 
in the larger universe of unaffected and uninterested and uninhabited 
(by problematic values) zones managed by DNS operators is not relevant.

Is our problem pervasive, or is it local to some graphs? Do we care 
equally, which is to say we don't care at all, about the graph that is 
anchored in .gr, and random nodes everywhere else?

I don't think our problem scope is well bounded by "at the
_top_ level we can co-ordinate something", or by "unbounded", that is, 
is application, rather than delegation, defined.

If the existing problem is scoped (and I do ignore the certainty that 
someone somewhere will create something intentionally in an arbitrary 
node in an arbitrary graph with an arbitrary anchor), then for central 
design choices to be revisited to allow an unscoped solution, is not a 
small choice.

Eric