Re: [wg-c] more on non-shaired gTLDs

[Kent Crispin <kent@songbird.com>]

On Mon, Mar 20, 2000 at 01:02:20PM -0800, Rick H Wesson wrote:
> 
> Eric,
> 
> why should new gTLDs be creted for potential registrant populations less
> than 10^5.

While this seems superficially reasonable, it is in my opinion going at
things in the wrong direction, and asking the wrong question.  Small
size of the potential registrant population is a secondary concern, and
in fact, in the other direction it is a truly bad measure -- we don't
want to only have TLDs that have potential registrations in the
multimillions -- that would just create repeats of the .com problem. 

We are far better off to have *many* TLDs, each with relatively small,
disjoint, populations, than we are to have 10 TLDs with huge registrant
populations.  By this standard, .int is an excellent TLD, even though it
is very small -- entities that register in .int, as a general rule,
don't also register in other TLDs.  .edu is also a "good" TLD, because, 
while some schools do register in .com, by and large .edu registrations 
don't overlap with other TLDs.

There will always be some overlap in registrant populations, especially
as long as there are huge TLD megalopolises like .com.  But I am 
suggesting that one of the primary criteria for goodness of a TLD is 
how well it captures a unique registrant population, or, in other 
terms, how well it partitions the total registrant population.

By this measure, a chartered TLD may or may not be a "good" TLD.  
.museum would be a "good" new TLD if the registrants didn't also register 
in .com; it would be a "bad" new TLD if the registrants all registered 
in .com as well.  Note that this factor works well with chartered TLDs, 
but is not restricted to them, because in some cases the name alone 
is a strong segregating factor.

By this measure, also, TLDs that appeal to ethnic or cultural groups are
also good, because entities that register in them will tend not to
register in other TLDs.  

Couple more points: 1) this "partition the registrant population" principle
would seem to somewhat enhance the use of DNS as a directory, but that is
an effect, not a goal.

2) This approach to name space management is also somewhat contrary to
monolithic commercial TLDs that seek to maximize registrations, because 
overlapping name/registrant spaces are good in that world.

Kent

-- 
Kent Crispin                               "Do good, and you'll be
kent@songbird.com                           lonesome." -- Mark Twain