Re: [wg-c] Choosing the intial testbed

[John Charles Broomfield <jbroom@manta.outremer.com>]

(rehashing dead horses for the umpteenth time).

> > >    I personally don't see why AOL, with millions of customers, couldn't
> > > have .aol.  Or my AT&T couldn't use .att, if they wanted to.
> >
> > Because there is therefore no reasonable criteria by which you could
> > restrict *any* TM holder from getting a TLD.
> 
> Omigod! The name space might have to respond to user demand! The Horror!!!
> Kent, you have already proven to our satisfaction that adding additional
> TLDs to existing registries can be done at something close to zero marginal
> cost. So what's the problem with this scenario?

The problem is that this scenario (a flat namespace) is exactly what
prompted creation of DNS. The flat namespace was hosts.txt namespace.
If we look at the current namespace where ".com" has more names than
everything else combined (how many million last count?), as far as size
goes on a theoretical level, we can discount the existance of everything
else and say that operationally the current namespace consists of one single
TLD (ie ".com"). If you want to add everything at the root level, then
you've already got the example (just happens that it has a ".com" on the end
making it visually less sexy, but operationally it is the same) that your
asking for. Thus, you don't need ANY expansion at all.
Contradictory as it may seem, changing to an unlimited open root is very
close to not changing at all, so why bother? Just argue for no change.

> > >    Why not permit every ISP with a million or more customers to have a
> > > TLD of their choice?  This could be a simple rule.   I'm sure there
> > > would be complaints, but would it cause any harm?  It wouldn't hurt me.
> >
> > Wow.  "It wouldn't hurt me."  That's deep.  I'm really encouraged that
> > you are a consumer advocate.
> 
> Nice rhetoric, but you didn't answer his question. If you don't like the
> "million" threshold, lower it. The principle is the same. Why not?

So, the first 100.000 arrivals get a domain name (a TLD) and for the rest
it's tough luck. Imagine the outcry if you had tried to apply that argument
to any TLD (sorry, only 100.000 registrations allowed in ".com"). Limiting
in number but (for practical purposes) little else WOULD be considered
restraint of trade.

Also note that with an open and free for all root, suddenly the level at
which running a registry becomes interesting becomes the registry for the
root.

Today we can see 3 levels of registry:
-the root (run for the moment on a technical basis by NSI, but with
 modifications/additions/deletions coming from ICANN after getting
 authorization from DOC/NSF).
-under the root we have the TLDs (NSI for com/net/org, different operators
 runnning all the ccTLDs and lets not forget GOV/EDU/MIL/INT)
-under each TLD, each domain holder runs their own registry for their own
 domain (or outsources it to an ISP or whatever).

There isn't much business to be made by being the registrar for a domain
under a TLD (ml.org was a good proof), so it's widely ignored, and it comes
down to an outsourcing of operations (which is what companies that have
their own domain do to have it operational: they outsource the management of
the registry -the zone file- to an ISP, but the ISP doesn't decide what
host names he's going to add in there, and if the domain owner doesn't like
how the ISP is managing his zone-file he just moves off to another ISP).

There is a very limited universe at the TLD level, and limitation means
money to be made. For now it's too limited even for that, which is why we're
having all these discussion, but even at its current limited state there ARE
aprox 240 TLDs (even though -as I state above- 1 of these TLDs accounts for
the majority of the space).

Managing the root is not something (today) that has a great deal of
technical changes (though having a set of stable root-servers is not as easy
as some seem to think), so there are not lots of companies out there
shouting that they want to run the legacy root (then again it can only be
run by one single operator).

If you open up the root, then everything just jumps up a level, suddenly all
TLDs will be managed as SLDs are today, nobody would be interested in having
an SLD (just as very few are interested in having domains under ml.org), and
the operation of the root suddenly becomes exactly the same as the current
operation of ".com". Caching goes out the window (theres a great question
mark about its funcionality as such, but why let small details like that
bother you?). And I can imagine the outcry about the one company running the
registry for the root. The registrars still continue to do business of
course as entities will still want domains (but without messy bits behind.
After all, why would I want jbroom@outremer.com when I can have
jbroom@outremer ?)

Yours, John Broomfield.