Re: Shared vs. Exclusive

[Kent Crispin <kent@songbird.com>]

Tony Rutkowski allegedly said:
> 
[...]
> The more registries, the more complex and bureaucratic
> the administration/oversight.  This suggests a low 
> value of n.

No.  How about:

"The more grocery stores, the more complex and bureaucratic the 
administration/oversight."

There isn't a simple relationship between "n" and the complexity of 
administration/oversight.   In fact, one of the goals of the IAHC 
should be to *set up* a simple regulatory structure, and let market 
forces work.  

In fact I think my example hints at a far more likely source of 
complexity in administration/oversight -- the monetary value of the 
items being overseen.  It is my belief that licenses for monopoly 
ownership of a desireable TLD (worth potentially many millions of 
dollars), with the unavoidable legal exposures that attend, will 
require far more complex administration than thousands of mom and pop 
shared registries would.

There are two competing models under discussion -- the "shared TLD"
model and the "monopoly TLD" model.  In either case some body (eg, the
IANA) "grants" the right to register SLDs in a TLD.  Let us suppose
that this "right" takes the form of a licensing contract -- that is,
in either case the registry "gets a license" to register SLDs in a
TLD.  The body that grants the license we can call the Licensing
Authority (LA).  (I quote several terms to indicate that they haven't
been defined yet -- things should become clear further down.)

(Also, I must mention that the scope of the "shared" vs "monopoly" 
dichotomy is not universal -- the ISO TLDs are out of scope, as are 
.MIL, .GOV, and others.  .COM, .ORG, and .NET are controversial -- I 
believe they should be shared.)

A good deal of the work of the IAHC, then, revolves around issues like
determining the nature of the licenses, the nature of the LA, what is
the source of the authority for the LA, the legal exposure of the LA, 
and so on.

The value of a license in the "monopoly" case is potentially very
high, making it a reasonable investment to engage in costly lawsuits
to get one.  And, since the LA is dealing in things of great value,
it's legal exposure is also very high.  Also, the level of authority 
needed by the LA is very high -- it would need strong legal authority 
that was international in scope.


OTOH, the value of a license in the "shared" case is rather low, and
lawsuits become much less likely.  The license, in this case, is more
like a business license, and though I am not a lawyer, it seems to me
that it can be structured to essentially eliminate the legal exposure
of the LA.  And the level of authority needed would be something at a 
much lower level, perhaps just the blessing of a recognized Internet 
Standards body.

I believe the legal tensions around the "monopoly" model are such that
a workable infrastructure for a LA supporting it is not an attainable
goal, at least not in any reasonable time frame.  However, I think 
it is quit possible that we could define an infrastructure for a 
shared LA fairly quickly.

To that end, here are some thoughts about how a LA for shared TLDs 
would work.

The LA would operate in a capacity very similar to the IANA.  It would
authorize new TLDs, and make them available to root nameservers.  As
is now the case, the operators of the root nameservers would be under
no contractual obligation to support a particular TLD, however, the
blessing of the LA would give a TLD the status of an Internet
Standard. 

A license would be a contract between the LA, the licensee, and all
other licensees for that TLD.  Coming up with the license agreement is
the crux of this whole approach, of course, and it won't be easy. 
However, it is certainly easier than coming up with the international
authority necessary to support a LA for monopoly TLDs. 

I am not a lawyer, but here are the characteristics that I would like
to see for this license agreement: When you get your registry license
you automatically enter a contract with other registries that serve
that same TLD, and the LA.  Included in the contract would be an
agreement to abide by a dispute resolution process.  Also included in
the contract would be a requirement that databases be shared [the
sharing wouild be mandated to follow internet standards].  Note that
the license is a contract between the registries, so one registry
would be able to sue another for lack of performance.  The license
would immunize the LA from suit, to the extent possible.  The license 
would further explicitly 1) relinquish all ownership claims to the 
name of the TLD, and 2) bear responsibility for defending against 
other ownership claims -- that is, the LA is the presumed owner, but 
in case of dispute, it is the responsibility of the registries to 
weather it.  [An alternative would be for the LA to get clear legal 
title to the name by trademarking it.]

The license would of course be engineered to, as much as possible,
keep things out of the courts, partly to avoid the expense, but also
to avoid the issues of which countries laws take precedence.  [As a
matter of practical fact, the license would undoubtedly be defined
within the context of US law, but a lot of work should be expended to
make resorting to the legal system extremely rare.]

The "sharing of databases" and the connection with the root servers 
are where the technical issues really come in.  These technical 
issues are currently outstanding.  Note that it wouldn't be necessary 
that there only be one standard -- a group of registries could, for 
example, produce an RFC for a different technique for sharing, and 
get it approved.

Note also that the standards don't need to exist yet.  For example, a
shared license could be given for .WEB, and no further licenses would
be granted until an RFC for a sharing protocol was approved.  Upon
approval of such an RFC, however, the LA would be free to license more
licenses for .WEB, and the the first licensee would be required to 
share data using the protocol defined in the RFC.

Anyway, I could go on, but as you can see, I think the shared TLD
model fits right in the current operation of the internet, without the
creation of a gigantic political and legal infrastructure. 

-- 
Kent Crispin				"No reason to get excited",
kent@songbird.com,kc@llnl.gov		the thief he kindly spoke...
PGP fingerprint:   B6 04 CC 30 9E DE CD FE  6A 04 90 BB 26 77 4A 5E