Re: Tired of Waiting

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

Hi Roeland,

> MHSC is looking at the same business operations model as IOD (which Chris
> is not directly involved with anymore, he works for Microsoft now). I see
> no benefit in breaking out the registry. Most of the cost is in the
> registrar anyway. Especially if the hardware is already sunk-cost for other
> operations. In that case the registry is no cost, just a postgreSQL
> table-set, occupying disk space, and CPU, which other operations are
> already paying for. IOW, your argument carries no weight.

Yup. You're completely right. In a scenario where you own/control the
registry and you are the only one doing registrations, outsourcing the
registry has little or no advantages. One thing that just about everyone
agrees is the prevailing vision (even Chris Ambler I think, though he says
he doesn't like it), is that registry should be separate from registrar to
allow customers to be able to registrar under a given TLD through different
companies, and not be forced to go through one companies horrible procedures
when they'd prefer another companies simple procedures (or so that they can
get a no frills service as opposed to a hand-holding process, etc...)
One thing that you seem to agree with however is that the registry part,
once operating is nect to nothing as far as operating costs... (more on this
later).

> The trick is that no single-service ISP can do this. The same reason that
> we can cut Verisign's price structure out from under them, it's not our
> main source of revenue, it's gravy. We pay the rent from other, more
> profitable, services. However, we will never operate a service at a loss,
> but we don't have to. It's the incremental profits, from a full range of
> services, that make the next generation ISP profitable.

Agreed completely. Take a look at the UK, and tell me how many companies you
find that survive solely on ".uk" registrations and do no other services
(Purely out of coincidence, in the UK is where one of the largest
"registrars" for .everything is based, but that's another story). My guess
is that you'll find none. Count how many registrars there are for ".uk"
(doing it with nominet), and look at them. Most are ISPs that interface with
the registry to update/enter records for the customers that they have that
host webs with them, or are directly connected, or need email, or...
A domain name on its own is something pretty useless. You need a link
somewhere or a web server or a mail forwarder or a bunch of hosts or...
In just the same way that GENERALLY a customer doesn't get his IP addresses
from one place and then sets up somewhere else (I'm talking about general
cases), it would make sense that in general a customer would also get his
domain names through his upstream provider. Most ISPs would eventually
become registrars.

> > (maybe Chris can inform us of how much that part costs him in
> >systems, staff and personal time, but I'd be surprised if he was managing at
> >less than $10 per name. I think he has around $1000 names registered, so if
> >he wants to argue that he has spent less than $10.000 I would be very
> >surprised),
> 
> Then continue to be amazed. Given sufficient volume in other services, the
> TLD stuff drops below the financial "noise" level. I can run a TLD registry
> on a four-year old 486. Given three year depreciation rates, the box is
> zero-cost. Maint, once it is running, is not much higher than the cost of
> keeping a 150 watt light-buld lit 24x7. The real cost is the development of
> the registry/registrar software (about $600K) which, amortized over three
> years, comes out to $1.6K per month. At $35 per month, break-even is about
> 45 SLD's per month for that TLD. The interesting thing is that the second
> TLD uses the same software, but it's a sunk-cost. Better, the IRS will
> allow the write-off as a deductible loss., which recovers $0.28, per
> dollar-spent.

I said I would be surprised if he had spent less than $10K, but you say that
the development of the software costs $600K. That's exactly what I was
talking about... So if under your calculations he has spent $600K, and he
has taken back around $35K in registration fees, then he's heavily broke...
(Maybe that's why he so badly needs the monopoly over '.web')
You put forward a good case for having just one registry. The same
hardware/software setup to deal with one TLD should be able to handle 1000
TLDs, just spreding out the initial cost until it goes so low that its
ridiculous. That was the initial philosophy of CORE. However the consensus
seems to be that there should be more than one registry (I feel it's a
political imposition by people like Ira Magaziner or others who haven't
really looked as deep as they should, but with some impositions like this
its best to work with them than ignore them). In any case, thanks anyway for
defending the choice of a single registry.
Having covered initial cost, running costs are ridiculously low, and service
is relatively simple to give, so allowing a for-profit company to have a
strangle-hold on this (ie allowing him to charge whatever he wants for
example) would be bad practice.
Chris says "hey, then put a price cap". It's actually the other way around:
The conglomerate of registrars should actually say "we're willing to pay
THIS for the business of having a company run this registry for us. Who's
taking?" All the while, the ownership of the TLD stays in the hands of the
overall structure, but not owned by any one group (if you like it's nIANA
that owns all TLDs).

> > so from his $35/yr he has maybe $10 dedicated to the registry
> >business. CORE *TODAY* would charge him $15 per name, so he could "up" the
> >price by $5 to $40. Seeing that you're so convinced that the CORE bunch is
> >going to stay at ludicrously high prices, wow... he's going to make a
> >killing!
> 
> One man's killing is another man's poverty. Please define this killing.
> BTW, NSI has set the expectation for SLD prices. The upper bound, at least.
> Given that the minimal costs are "fixed" the amount of margin is
> volume-based. How many SLD registrations, per month,  are you assuming?

There might be 'expectation' for SLD prices, but nothing that causes people
to go elsewhere for a question of $30 more or less (in general).
When you setup your business, buy your furniture, get your internet access,
buy your computers, etc etc etc... the average Joe *DOESN'T* think about the
price involved when registering the domain name but rather thinks in terms
of what is best suited to his business, and then gets it. He will only back
down on his initial choice and get another one if the policy/difficuly/price
of his initial choice are WAY out of sync (ie a price of $10.000 to register
a domain or...)
I'm not just making this up! I've had experience before while living in a
European country where the policies really blocked getting registrations
done, but even so the companies would really fight for their names, and now
here in Guadeloupe companies are faced with 3 possible choices:
(Guadeloupe is an overseas department of France)
They can either get a ".gp", or a ".fr" or a ".com/.net/.org".
Costs and hassles:
For ".gp" if they are local, its FREE, with no yearly renewal or changing
fees. Hassle is minimal as far as proving their identity.
For ".fr" if the registration is done through a basic registrar, then the
cost of registration is aprox $200, each modification is aprox $50, and
anual renewal is around $18. Quite a bit of hassle in having to fax legal
documents to the AFNIC (why they call themselves AFNIC instead of FRNIC is
beyond me, but its their choice).
For ".com" it's $70 for registration, and $35 for renewal. Also minimal
hassle as far as identity checks go.

So:
.gp : free & minor hassle
.com: cash & minor hassle
.fr: lots of cash and more hassle

Do our customers first ask what the policies are and then ask for a domain
name (talking with my ISP hat on) ? Well, most of them are clueless, and
accept the recomendation of a ".gp" or ".mq". However, for those customers
who are already internet aware, when they get to us they already know what
domain name they want, and if it's a ".com" or a ".fr" they usually stick to
that choice. Have they heard of NSI, AFNIC, INRIA, IANA...? No way.

Therefore, any new TLD that comes in the market can have whatever prices it
wants, and it will be used. Even if its only to get registered by that bunch
of aprox 100 names that want worldwide visibility with "name.*" at whatever
price (names like intel, microsoft, netscape, chanel, etc...).
If you could get global visibility for a TLD with whatever uninteresting
combination such as ".asdasd" and charge $10K per registration, you'd
probably STILL be able to register at least 50 names... (and then there's the
renewals!)

Yours, John Broomfield.