Re: [ga] Draft Resolution

[Kent Crispin <kent@songbird.com>]

On Sun, Mar 18, 2001 at 10:53:09AM +1200, DPF wrote:
> On Sat, 17 Mar 2001 09:44:26 -0800, Kent Crispin wrote:
> >"huge"? I doubt that very seriously.  While people complain about the
> >lack of effectiveness of TLD vs TLD competition, the effectiveness of
> >serial competition is equally questionable, at least as things are 
> >currently constructed.  [+] (see note on "serial competition", below)
> 
> I've thought about this some more and I have to say I am more
> convinced than ever that "huge"is indeed appropriate.  By 2007 the
> registry side of *.com alone could well exceed US$1 billion - an
> amount which could see several quite large companies decide to put in
> exceedingly competitive bids for.

You mean in yearly income? Possibly, but there are many assumptions.  In
particular you are making the assumption that the other gTLD registries
will essentially have failed. 

> Verisign as an intelligent company
> is not going to ignore this possibility and rely on lawsuits to keep
> the registry (uncertain and expensive) but with no registrar business
> will be especially keen to keep the registry by performing to a
> standard (minimal complaints from registrars, lower prices etc) that
> is untouchable.

I don't think that is particularly realistic.  It doesn't matter what
standard they keep, other companies can still make good bids, so the
threat of a lawsuit is one of the major arrows in their quiver, and they
will keep it sharp.  Moreover, while the outcome of a lawsuit is
somewhat unpredictable, that is only part of the equation.  The other
part is how much money they can force ICANN to spend on legal bills.  It
is all very well to say that ICANN should ignore the cost of defending a
lawsuit in negotiations, but it would in fact simply be stupid and
irresponsible to ignore it.  You just mentioned that the value to
verisign of the registry could be $1 billion.  Therefore, verisign could
afford to spend $500 million on a lawsuit (and lobbying governments, and
bribing directors, and whatever), and still make half a billion.  
(This, incidentally, is the best argument I know for the proposition 
that registries should be non-profits.  If registries are huge profit 
centers, then there is no effective way for a tiny non-profit 
corporation to regulate them...)

[...]

> >Clearly incentives exist -- they always exist, but the
> >mechanism of organizational firewalls is deemed sufficient in many
> >anti-trust circumstances, and if you are going to claim that they are 
> >not sufficient in this one, you should document it.
> 
> On the contrary the onus is on Verisign to document that they are
> sufficient.  Plus at present they *may* be sufficient because the
> contract is for Verisign to only have a registrar business until 18
> May.  Once that is no longer the case then the incentives change and
> what was sufficient may be no longer.  I believe the increased risk is
> enough to want to avoid it.

This is all wild handwaving -- "...may be sufficient...", "...may be no 
longer...", "...I believe...".  

> Verisign may be the nicest guys in the
> world (yet managements change) but why put in place a regime which
> will place anti-competitive incentives in front of them,

If they succumb to the anti-competitive succubus, then they would be in 
material breach of their contract, and thus susceptible to losing their 
contract.  There will be more than 100 registrars just looking for the 
slightest breach. 

And believe me, I am not assuming that verisign are the nicest guys in 
the world.  I am assuming quite the opposite.  

> >> - Verisign would be in a position where it could purchase other
> >> Registrars thus once again gaining over 50% or even 75% of the
> >> registrar market in *.com, as well as the registry.
> >
> >Big "So what".  Any large company can do this; it is independent of the 
> >issue of the registrar/registry split.  See the below discussion on the 
> >"zero wholesale cost" argument.
> 
> The argument used by the ICANN staff is that they are now below 50% of
> the registrar market.  The point is they can easily change that
> overnight if they are allowed to by the proposed changes.  And having
> the registry also does give them an advantage.

Sorry, you missed the point.  IBM as a registrar could do exactly the
same thing -- sell registrations for way below cost, buy up other
registrars, etc.  This is simply a function of being big and having lots
of money.  This advantage accrues to any large, wealthy registrar with
other sources of income, whether or not they are somehow joined to
registry.  

Yes, there are some advantages that ownership of both a registry and
registrar could give to the owner, no doubt about it.  But the issue is 
a good deal more subtle and complex than you portray it.

> >> - The possibility of a change to the status of *.org registration
> >
> >1) There is a possibility of that, regardless.  In any case, the registry 
> >operator is bound to abide by "consensus policies".
> 
> Well according to ICANN staff/website we only have one - the UDRP.
> Absolutely everything else is a contractual matter it seems.

Please read section I.1 of the contract.  Though there are caveats, the
registry is obliged to adopt any new consensus policy.  The fact that
there is only one at present is immaterial. 

> >> - Granting the ability to Verisign to increase registry prices with
> >> only 30 days notice
> >
> >But there are price caps, as I recall.  Price caps are the only thing
> >that matters, in fact, because if some other company wins the rebid,
> >they win a monopoly for 4 years (or whatever the term might be) in any
> >case, and can immediately reset the prices. (* "serial competition", below)
> 
> What are the price caps?  The Registrars can't seem to find them.  

     ..."(ii) the prices shall not exceed those set forth in Appendix G."

I haven't found Appendix G, either, but that language certainly means 
"price caps" to me.  So, while I don't know specifically what they are, 
I am confident they will exist.

[...]

> >If, on the other hand, you believe that competition between gTLDs is not
> >going to be effective, and that .com will forever be the only important
> >player, then it is very important to constrain the power of that
> >registry as much as possible.  Of course, it would be nice if we could
> >have both, but that option isn't available.
> 
> There will be some serial competition - especially amongst large
> businesses who will want all suffixes.

I believe that you misunderstand what I mean by "serial competition". 
"Serial competition" refers only to the competition that happens during the
periodic rebid process -- say, IBM bidding for .com, in competition with
Verisign.  It has nothing to do with the customer level.  I don't see
any relationship with "businesses who will want all suffices". 

>  But I think *.com will remain
> the king amongst princes so to speak for a long long time.  Not just
> the head start they have but also even the media talk about dot coms.

All bad, right now...

[...]

> I think under the current contract Verisign would be fools to rely on
> legal action to keep the registry in 2007.  They would rely on great
> performance and low prices which is what we want.

You must be joking.  From my perspective this is terminally naive,
especially given the history... 

[...]

> >[+] Note on "Serial competition": The basic problem is that, as things are
> >currently, a reasonable rebid cycle would be on the order of 4 years or
> >more (since the switching costs involved the change of registry operator
> >are sgnificant -- significant for the registry operator, significant for
> >the Internet comunity).  
> 
> How are they significant for the Internet community?  All you have to
> do is update Root Server A with the new delegation for *.com.  No need
> for the interface to change.

Bzzzt.  Nope.  DNS is only a small fraction of the change involved.  The
database of registrations would be transferred to a new company, new
hardware, new infrastructure, new operations staff for the registrars to
deal with, new whois service, new contracts.  The changes are very
significant; the lockin problems are obvious. 

[...]

> >[++] The "Zero wholesale cost" argument is bogus, because it neglects
> >the other side of the equation: "zero wholesale profit".  To see this,
> >suppose that the NSI registrar, in a fit of monopolistic greed for
> >market share, decided to leverage it's supposed "zero wholesale cost" to
> >the max, and to give away registrations for free.  In such a case, every
> >registration by the NSI registrar would be $6 lost in fees that would
> >have been collected from any other registrar.  This is a real loss for
> >the registry. 
> 
> I agree it is not economic sense for Verisign to have its registrar
> give away registrations for zero dollars.  However let's look at what
> happens if Verisign the Registrar decides to sell registrations for
> say $6.
> 
> First of all no other registrar would be able to sell at that price
> because it would give them a zero margin to pay their staff and costs.
> They would go bankrupt if they tried to see at zero dollars.

Nope.  Think of IBM as the registrar.  This is, once again, just a
matter of the wealth of the entity, and has nothing per se to do with
the registry/registrar split.

> Now Verisign the Registrar would lose money on the $6 sale fees but
> Verisign the Registry would make money ($6 less costs) and if the
> combined costs of Registry and Registrar are less than $6 then
> Verisign overall makes money, increases market share and bankrupts the
> competition.

But they *still lose lots of money*, and that is a critical factor --
just above you describing how they were sucking down $35/domain/year;
going down to $6 would be like an 80% drop in revenue.  To undertake
such a move they would have to be in a truly desperate situation. 

Think Verio the ISP as a registrar -- they could throw in domain 
registrations for $0, and as long as the combined costs of their ISP 
business and their registrar business were less than the fees they were 
charging for their ISP services, they would overall make money, increase 
their market share, and bankrupt the competition... 

Should we therefore disallow ISPs from owning registrars, because they
could subsidize the registrar costs with their other operations?
Obviously not.

You will probably argue "what about small registrars"? I think that in
any highly competitive commodity market, which is what domain name
registrations are becoming, small players are going to be at a serious
disadvantage, and this is largely independent of anything to do with the
registrar/registry split.  It is no coincidence that most registrars are
seeking other sources of revenue than pure domain registrations --
domain parking, web hosting, etc.  

James Seng pointed out that the current model doesn't deal with the
phenomenon of "resellers", but there are other possible linkages.  In 
general, the possibilities are only limited by the imaginations of the 
companies involved.  I am quite sure that the divestiture that NSI is 
contemplating for its registrar will abide by the strict letter of the 
contract, but I suspect that enormous room for creative deal making 
remains. 

> >It is perhaps unfortunate, but the fact is that *any* large company
> >enjoys precisely the same advantage that Verisign does, and the
> >inescapable conclusion is that small registrars are not going to make it
> >in the long run, regardless.  This is completely independent of any
> >issue of registry/registrar split.  
> 
> Disagree.  As shown above allowing any company to keep both the
> registry and a registrar will allow it to set retail prices at a level
> where they are the only company which can make money.

Yes, in an extremely simplified view that leaves out many critical real 
world factors, that might be the case.  I would rather try to look at 
what really happens, though.

-- 
Kent Crispin                               "Be good, and you will be
kent@songbird.com                           lonesome." -- Mark Twain
--
This message was passed to you via the ga@dnso.org list.
Send mail to majordomo@dnso.org to unsubscribe
("unsubscribe ga" in the body of the message).
Archives at http://www.dnso.org/archives.html