RE: [registrars] Motion for a Vote on Grace Period Deletion Fee

["John Berryhill" <john@xxxxxxxxxxxxxxxxx>]

> I think we all clearly understand your example -- register 1000 names
> and pay ICANN for 10.  Paying the fee on 1% of transactions while the
> rest of the market pays on 99% (assuming a 1% add grace delete rate for
> mistakes and testing) sounds like a very attractive model to your
> client, but it's not equitable.

Apples and oranges.  You are comparing paying 25 cents for one year's worth
of registration, versus 25 cents for a couple of days' worth.  "Equitable"
given the time span would be about .34 cents per name, not 25 cents.  If you
want to pro-rate that fee over the term you are getting, that is a step in
the direction of reasonableness.  Also, if the concern is over ICANN
funding, at least one of the registrars involved has OVERpaid its fees to
ICANN and said, "Keep the change."  I don't know how many other registrars
have voluntarily done that.

It's interesting to see someone else comment that it is interfering with
their "legitimate" backorder service.  Verisign had a clearly articulated
position not long ago that the add storms incident to that service were an
abuse, and the only solution was WLS.  At that time, several of the
registrars here cooperated in a lawsuit against Verisign for the "right" to
continue abusing the registry.  Should we gather that registrar-based
backorder services have became "legitimate" because the right people are
doing it?

So, while we're at it, why don't we solve BOTH the add/drop problem AND the
add storm problem by charging 25 cents per add request.  Don't all jump at
once...

Verisign had some interesting changeable figures during that very similar
discussion:

---
http://www.dnso.org/clubpublic/ga-full/Arc09/msg00931.html
In a
recent 17-day period, VGRS estimated that nearly 1.5 billion
"transactions" (i.e., adds, modifies, checks, queries, etc.) occurred in
its systems in efforts by registrars to register just-deleted names.
That resulted in only 3,156 registrations, meaning that on average
465,399 transactions were made for every added domain name.)
---

----
http://www.dnso.org/clubpublic/ga-full/Arc10/msg03413.html

> During the development of these services, which
> involve various
> repetitive polling mechanisms, there was concern that these
> services were
> overloading registry resources during name drop periods.
> That problem has
> since been solved.

At extremely excessive expense to VGRS to support a business model that
results in one six dollar registration for as many as 500,000 transactions.
----

The numbers in that debate, from the same person, went from a measured 1:147
domain/transaction ratio to a fictitious 1:500,000 ratio without missing a
beat.  The numbers are marginally better here.

If WLS was not implemented, it was going to be the end of the world.  There
is a tremendous sense of deja vu at work here.

Dropcatching used to be an "illegitimate" practice, and in view of that
history you have to admit that it is comical to see another method of
traffic prospecting to be opposed by at least one registrar on the basis
that it is interfering with dropcatching, now called "backorder", of all
things.  During the WLS debate, registrars engaging in dropcatching were
supposed to have been worse than typhoid carriers, the registry was going to
melt down, internet users were going to be afflicted with boils, sores, and
locusts, etc.; and now the largest registrar provides a dropcatching
service.  Technically, this is the pretty much the same debate, and one can
almost foresee with certainty that, two years from now, yet another model is
going to be interfering with someone else's "legitimate try before you buy"
business.

The bottom line is that this process is a win-win-win-win .. It's a win for
the registry,  registrar, registrant and icann.

There are two parties who do not win in this.  1. Competing registrars that
do not have the well-backed clients to participate in this exercise  and 2.
Mark monitoring services who have to change their methodology to filter
names out of the zone file updates.

It is always possible of course to register and delete names for
traffic-testing purposes in the 'glue records'.  (inserting one new hostname
for every domain registered rather than stacking them all on the same
hostname as registrars are now).

Perhaps the assorted Mark monitoring services might like to consider what
methodologies they would utilise to filter for testing activities if
hundreds of thousands of domain names were registered each day,  each under
its own unique hostname. (i.e.: anynameyoucanthinkof.com registered under
the hostnames ns1 and ns2.anythingyoucanthinkof.com.)  How much more
difficult would it be?

If you want many names to stop showing up and deleting in the zones under
the same hostnames I would suggest you be careful what you wish for, you may
get what you want.

Ultimately, however, this form of prospecting will be self-limiting due to
diminishing returns.  There is good reason to believe the practice itself
will follow a Hubbert curve with a first inflection point in about two
months.