Re: [gnso-idng] rethinking IDN gTLDs

[Eric Brunner-Williams <ebw@xxxxxxxxxxxxxxxxxxxx>]

Gomes, Chuck wrote:
> Eric,
>
> As I know you understand, even if there are no direct costs to
> applicants for the failure you cite, the bottom line is that the
> application fees could be lower if the evaluation process was less
> expensive.
>
> Chuck 

Each application portfolio manager is, in effect, conducting a kind of 
involuntary resource denial attack on ICANN, and all applicants (its 
own included), by making the cost of application preparation scale.
CORE's .cat has kittens. We have contract management, and document 
management, and ICANN application response management costs, but we 
have only one and a bunch of fractional costs of actual application 
preparation, and that's the biggie.
We'd like to signal to ICANN that "these e-reams of paper crudely 
nailed together" are one application needing the standard duty of care 
by the evaluator, and a bunch of edits off of that single application 
each of which has very limited scope, and for which no standard duty 
of care by the evaluator is warranted outside of these very limited 
scopes.
We like to signal that the consumption of standard complete duty of 
care units by the evaluator, a limited resources, by the N-1, can be 
released back to the general resource pool, to be expended on other 
applications, reducing the cost of the evaluation process.
Unfortunately, thus far in the DAGvX process, it is ICANN's volition 
that its evaluation (defense) cost doesn't scale proportional to my 
application (attack) cost. I can't tell you how frustrating its been 
explaining that "independent evaluation" of identical applications is 
only useful, as a question of method, in observing the behavior of the 
evaluator, that is, as a perversely wasteful form of quality control.
Which evaluators "pass" .cat's clone? Which "fail" the same 
application? What is the process utility in that?
This is why I followed up on the Edmon/Avri exchange. Useful 
information that would lead to lower evaluation cost if used early, 
exists in the case of every application brought by an existing 
registry operator, and more generally, by any multi-application 
author, where past applications as well as contemporanious 
applications matter, even applications brought by some, though not 
all, ccTLD registry operators ({.de, .se} yes, {.ir, .eg} no, as a 
completely arbitrary example where I know the details), and in 
particular, applications brought by an ASCII operator for a non-Latin 
label, or more generally, for any set of applications which meet some 
"contention" test based upon the assumption that the "contending" 
applications are independent, not correlated.
This list's discussion mentioned in passing an area where cost savings 
are significant, on the order of $1m, perhaps as much as $2m. It isn't 
unique to IDN application, it just happens to be possible with these 
applications as with other applications with some characteristics.
Writing this note has made me aware that some of the ccNSO members 
have a material interest in the issues before this GNSO centric list. 
I didn't realize that before this evening, beyond the IDN technical 
consistency set of issues.
Eric