Re: Public Resource vs Private Ownership

[Kent Crispin <kent@songbird.com>]

On Mon, Feb 16, 1998 at 03:35:38PM -0800, Patrick Greenwell wrote:
> On Fri, 13 Feb 1998, Kent Crispin wrote:
> 
> > > And Jay, like many others feels otherwise. Given the inability to
> > > reach any real consensus on this issue despite claims to the
> > > contrary, isn't the reasonable compromise in this situation to allow for
> > > both models? 
> > 
> > Not really -- for profit registries are fundamentally impractical:
> 
> In your opinion.

That goes without saying, of course.

> And judging by NSI's success in this area, an erroneous
> one.

Eh? NSI is currently being run as a shielded monopoly as a contractor
to the US Government. 

> Would you also address the "impracticality" of other registries such
> as .nu, .cc, .to, .tm all of which are operated on a for-profit basis?

First of all, none of them have been in business longer than a year, 
so they really convey no information about practicality or 
inpracticality. 

Second, they have the luxury of being controlled by sovereign
entities, and therefore have legal characteristics that are different 
from the commercial monopoly registries we are discussing.

Third, their names were delegated to them through a process 
that seems very unlikely to involve any lawsuits.

> I imagine the companies behind this registries would have a vastly
> differing opinion on the subject. 

Perhaps.  But Bill Semich, who is behind the .nu registry, is a
signatory to the MoU, as is Netnames, the company behind the .tm
registry.  Furthermore, I believe (though I could be wrong) that the
.nu registry is being run on an essentially cost-recovery basis. 

> Perhaps you would like to clarify your statement?

See below. 

> > It is quite clear that IANA's authority derives from community
> > consensus,and nothing else -- IANA has absolutely no way to force all
> > the nameservers in the world to point to the root zone it defines;
> > anyone else at all can set up root servers with minimal expense and
> > effort.  If other governments, for example, feel that IANA is playing
> > favorites with the USG, then they will indeed start their own root
> > servers.  
> 
> Which is why it is important to achieve a true consensus, something which
> despite claims to the contrary, the gTLD-MOU supporters have utterly
> failed in doing. If there was any sort of real consensus, there would have
> been no impetus for the USG to get involved.

As far as I can see the US government is involved ONLY to 1) protect
the interests of a very large defense contractor (NSI is a totally
controlled subsidiary of SAIC, a privately held multibillion company),
and 2) to deal with internal US political issues.  I am inferring
point 1 from the content of the Green Paper; I have been told that
number 2 has been stated by Magaziner.  There is no other reason for
the USG to get involved -- despite the fact that the MoU has been
controversial, it created a real, private sector developed solution
that was clearly about to work. 

And I disagree that the MoU utterly failed to gain consensus -- in 
fact it has gained far more support than any competing alternative, 
including the Green Paper.

[...]

> > Given the monetary value of an exclusive franchise on part of the 
> > DNS, there is absolutely no way that that IANA could give out 
> > such exclusive franchises without serious legal exposure -- unless...
> > ...the USG steps in and indemnifies IANA.  This would require special 
> > laws regarding IANA.
> > Such a move would rightly be seen by foreign governments as an attempt
> > to take control of the DNS.
> 
> Not if they were involved in the process.

It's pretty clear they were not.  In fact, it is pretty clear that 
the process was motivated purely by concerns internal to the US.

> This is an important
> constituent in the long term success of any organization charged with such
> a task whatever model is followed.  

Indeed.  I believe this will be clearly demonstrated by the critical
comments that various foreign governments and organizations will file.

> > Contrast this with a model where all TLD registries are either 1) 
> > delegated to a sovereign entity or 2) required to be run on a 
> > cost-recover non-profit basis.  
> > 
> > In this later case the possibility of any kind of lawsuit is
> > drastically reduced -- there is a uniform, consistent policy,
> > and registries are not high profit monopolies. 
> 
> Competition in the marketplace is a significant factor in determining
> price. At this time we are only beginning to see such competition as with
> the ISO-3166 registries such as .nu which is charging $25 a year for
> registration. I do not feel that you have the basis to substantiate such
> claims.
>
> While it would be correct to state that a company may have a monopoly on a
> TLD, they would *not* have a monopoly in the marketplace, which is far 
> more important in leveling such charges.  

If domain names were simple commodities for sale, what you say would
be true.  But they are not.  There are actually several arguments 
-- I will present a couple.

First -- you actually have to analyze things in a little more depth --
you have to think long term, about what levels prices will tend to
equilibriate. 

There are actually two different modes where registries interact with 
customers -- first, in the initial "sale" of a domain name; and 
second, in the recurring "rental" of a domain name.  

Indeed we can expect that multiple for-profit registries *would*
compete for initial price, especially when they first start business
and are looking for market share.  Indeed, we could expect to see
"Special Prices!" and "Introductory Offers!" from these registries,
designed to get get customers as quickly as possible. 

However, the real revenue stream of a registry is not its new
customers, but rather its long-term customers -- its "rental" income.  
Here the forces affecting prices are totally different -- the point 
at which prices equilibriate is the point at which customers leave a 
registry for another registry.  In other words, the prices will RAISE 
to the point where customers will leave.  This is absolutely basic 
economic theory, and obvious psychologically, as well -- a registry 
has absolutely no incentive to keep prices lower than that point, and 
there is always incentive to raise prices.

Now, it is clear that the cost to customers to move domains is orders
of magnitude more than the cost of a registry to maintain the domain
-- amazon.com has a huge investment in it's domain; the name is loaded
in hundreds of search engines, and hundreds of thousands of bookmarks;
not to mention advertising, and the very large but intangible "good
will" that well-known names bring to a business.  NSI could easily
charge amazon $1000 a year for their domain without any fear of amazon
moving. 

Amazon is an extreme example, of course, but even to a very tiny 
company like mine (songbird) it would be a royal pain to move 
domains.

This kind of locked in situation is precisely the sign of a monopoly. 

You might argue that this is precisely the model for the rental real
estate market.  But in fact it is not -- in the real estate market
landlords are actually offering different levels of service, and
different goods.  But registries are *not* offering significantly
different levels of service, nor, I would argue, are they really
offering different goods: It is the registrARS that offer different
levels of service -- all registrIES are simply back-end database
operations that the customers never see.  And, at the registry level,
the "good" being offered is in all cases essentially the same -- a
string of letters.

I grant you that these are not *pure* monopolies.  But it certainly is
not the free competition you are claiming exists -- registries *are*
getting the benefit of monopoly pricing effects. 


Second argument: Competition requires that the goods be substitutable
-- salt doesn't compete with shoes in the marketplace.  Likewise, TLDs
are not "substitute goods" -- a name in .arts is not the same as a
name in .nom, for example.  And when you look at the suggested TLDs,
they are really quite different -- .com is not the same as .shop, and
neither is the same as .firm.  .arts, .nom, .info, .rec, .web, .net,
.org etc -- these are all really *quite* different.  People,
businesses, and organizations will pick names that suit their purpose,
and will *not* treat them as interchangable goods.  Thus, these 
registries don't compete in any meaningful way.

Furthermore, in order for competition to be meaningful there should be
many players.  The Green Paper, therefore, is simply incoherent --
even if competition were meaningful at the registry level (which it is
not), the Green Paper approach of restricting the number of registries
would operate against any such competition to begin with. 

Well, I've got to go now.  There are several other arguments against 
the proposition that registries compete.  Most people who argue for 
it are actually still thinking of a model where a registry is also a 
registrar, IMO.  But if you think through the issue and remember that 
registrars are the entities that actually sell to the public, then 
the situation is quite different.

And these arguments don't even get to the other big issue -- the 
regulatory complexity of for profit monopolies.

-- 
Kent Crispin, PAB Chair			"No reason to get excited",
kent@songbird.com			the thief he kindly spoke...
PGP fingerprint:   B1 8B 72 ED 55 21 5E 44  61 F4 58 0F 72 10 65 55
http://songbird.com/kent/pgp_key.html