Incompetance abounds at the InterNIC

Patrick Greenwell patrick at
Sun Jan 24 05:31:45 UTC 1999

On Fri, 22 Jan 1999, Dean Anderson wrote:

> At 07:32 PM 1/21/1999 -0800, Patrick Greenwell wrote:
> >> But Internic isn't very forthcoming in how they manage their systems. Even
> >> so, I think I could run an automated database service for very little cost
> >> per transaction. On the order of hundreths of a cent per transaction.
> >
> >You and half a million other people claim to be able to do this. Running
> >code please?
> Running code on any of the dozens of discount brokerages.  

We aren't talking about running a discount brokerage, we are talking about
running a registry. 

> Actually, their code is much, much more complex than a domain registry.
> And much less automated. So charging 14.95 (E-Trade's fee) is probably
> too much, even with some "free" speculation.

None of which is suitable for running a registry. 

> Stock brokerages are probably a bad example, because they are not
> completely analogous. 

I am glad we can agree on something. :-)

> They are way more complex, and much less automated, and probably less
> automatable. But it is significant that brokerages are LESS automated,
> MORE complex, and still charge LESS than Internic. [just
> want to insert some random upper case letters ;-)]

Agreed. NSI is making a killing because there is no effective competition.

> >> You also assume that domain speculators are the only ones who register more
> >> domains than they pay for.  They aren't.  
> >
> >Yes, they are.
> ?? You contradict this in your next sentence.

Apologies, didn't read closely...

> >> Further, speculators ultimately pay for the domains they sell.  
> >
> >No, the purchaser pays for the domains they sell. The speculator often
> >does not. 
> This doesn't make sense. Its like saying the customer of a hardware store
> pays for hardware, and hardware store does not.  While I suppose it could
> be literally true, it isn't evidence that the hardware store "didn't pay".

It makes a lot of sense. Currently a speculator can tie up a domain name
indefinitely without paying a cent for that domain. All they have to do is
submit a registration, for which they will receive free use of the domain
for a few months, then the domain goes on hold, and then when it is
deleted, they simply reregister the domain again, and the cycle continues.
Cost to the speculator for the domain: $0.00.

> Likewise, Internic got paid for the domain, because the speculator got
> someone to pay for it.
> >It stand up quite well. Having worked for a registrar that employed both
> >models (post-pay and pre-pay) I can state unequivocally that our empirical
> >data showed that because of speculation a post-pay model was undesireable
> >as the speculators on the system at the time were not paying for the vast
> >majority of domains they registered. 
> But they generate a net profit for the registry.

They don't do anything for the registry. 

> Why? They cause more domains to be paid for than would otherwise be paid
> for. The revenue they create for the registry is greater than the
> infinitesimal cost of the domains which are never registered.

But the domains *are* registered, yet never paid for. This does not help
the registry in the slightest, and it does not increase the net profit of
the registry. 

> >Demostratively false, that is why we are having this discussion. The
> >problems at NSI weren't caused from normal load that they expect. If you
> >want to argue that they have a shoddy, underpowered infrastructure, I
> >would tend to agree with you. 
> Demonstration please?  *CLAIMS* that they cause internic problems are why
> we have this discussion.  

Again I would point to empirical data, and NSI's own admission that they
are having problems. 

> However, those claims are false. Internic problems are due to poor
> management, and as you say "shoddy, underpowered infrastructure", not
> speculators.

Well, a question for you then: What peaks of volume should a registry be
able to weather? If you are happily doing 1X volume and tomorrow someone
slams you with 10X or 100X is it your belief that any competent company
should be able to handle the volume? 

Map this example to ISP bandwidth for the sake of argument. Does the
argument still hold? Do most ISP's maintain a reserve of 10X or 100X 
worth of "normal" bandwidth for peaks? Now I have no idea what volume of
requests that NSI received, it could have been 1,000 extra or 100,000 or
some other number. However I think that it is reasonable to expect things
to break at a certain volume. What's that magic number?

> >> Speculators that actively try to sell their domains quite likely cause more
> >> domains to be registered, and so they increase the Internic sales revenue.
> >> I haven't seen any evidence that the costs of their non-completed
> >> transactions are more than cost of their completed transactions.   
> >
> >Think about what you are saying. The cost of a completed transaction is
> >absored by the fees paid for that transaction. The cost of a non-completed
> >transaction is paid for by the fees *everyone else* pays for *their*
> >transactions. 
> This isn't a sensible argument when you look closely. 

Sure it is. :-)

> The total cost is subtracted from the total revenue.   But if you are going
> to split up the costs, you must do it fairly. 
> 	Speculator revenue   less   Speculator costs
> 	--------------------------------------------
> 							    = Net profit
> 	Normal revenue       less   normal costs.
> 	--------------------------------------------
> 							    = Net Profit	   
> Since the speculators are generating positive cash flow (the revenue they
> generate is greater than the costs they incur), "Normal" users aren't
> subsidizing speculators.

> What you are doing is this:
> 	Normal revenue 	less 	normal costs 
> 				less	speculator costs.
> 	---------------------------------------------
> 							     = Normal users subsidy
> You leave out speculator revenue, and draw an incorrect conclusion.

You proceed on a false premise which is why your model does not hold.
Speculators by and large do not generate *dime one* for the registry.
They go register 100's or 1000's of domains with no intention of paying
for any of them. When they are sold it is the *buyer* not the speculator
that is paying for the domain. Again, the speculator adds no value to the
transaction. The domains would have been sold anyways.  

> >> The only speculators that might be able to register thousands of domains
> >> with little effort are the ones who register on-hold domains.  But they are
> >> doing a favor for Internic, because they are collecting fees.  
> >
> >Where do you get this wild idea? The only way this occurs is if the domain
> >is sold which occurs in the minority of cases, and then it is usually the
> >end purchaser, not the speculator that is paying for the domain. The
> >speculator serves no useful purpose in the transaction chain for anyone
> >involved save themselves. 
> Speaking of wild ideas. I suppose that you think the marketing/sales
> department serves no useful purpose either.  After all, it is the customer
> (end purchaser) who buys services, not the sales person.  You are being
> most naive.

The domains would sell quite well without the speculator. The only reasons
we have the speculators is *because* domains are selling so well. If they
weren't there would be no motivation for the speculators to engage in this
activity. I

Coming to the ISPF-II?  The Forum for ISPs by ISPs

More information about the NANOG mailing list