[registrars] eNom WLS position; word document copy attached

[Paul Stahura <stahura@enom.com>]

Wait-list Service (WLS) 
Position Paper
eNom, Inc.
January 12, 2001

ENom, Inc. has extensive experience in the re-registration of deleted domain
names. Like SnapNames and Dotster, we are one of the largest users of the
current batch pool, utilizing our connections and performing the service for
other registrars that we host completely (perform all of their registrar
related technology functions) or for which we host just the "dropped"
service, for a total of 5 registrars.  We differentiate ourselves from
SnapNames and Dotster in that we employ a subscription model:  The users,
currently limited to 10, pay a monthly subscription fee to participate in
our "club drop".  Once they pay the fee, they can submit as many names as
they wish and we will attempt to register the names as soon as possible.
SnapName's model is a wait-list service and Dotster's NameWinner is an
auction model.  There are other competitors, too.

This document outlines eNom's predictions, opinions and recommendations
regarding the proposed wait-list-service (WLS). For reasons that are
explained below, we do not recommend that the WLS service be implemented.
But, if the WLS is implemented, eNom offers recommendations, which appear in
blue text below.   

1.0 If WLS...  The current system usage problem will not go away.
	1)	Batch pool usage will not decrease. First off, the argument
that WLS will take care of all the "valuable" names so that there will be
less valuable names that drop into the batch pool and therefore less demand
in the batch pool and therefore less usage, is total crap.  Since it costs
next to nothing to slam the batch pool and since competition for any names
is fierce, registrars will slam just as much for a list of $10 names as they
would have done for a list of $100 names.  This slightly counter-intuitive
conclusion also results from the fact that batch-pool users will try just as
hard for one name as they do for 100.  In fact, most try to limit the number
of names so that they can slam it harder for just a few names. We do not
enjoy slamming the batch pool, but that is the competitive system we have to
deal with.  If eNom wants to compete, we have to slam the batch pool.  With
WLS in place, slamming the batch pool will not decrease since the WLS
provides no disincentive to do so.  Reducing the number of names or their
"value" will not reduce the slamming.  There are thousands of extremely
"valuable" names that do not drop today, yet no registrar reduces their
slamming because these valuable names are not dropping.  If only one name
was dropping, and eNom wanted it for a customer, we'd still slam like crazy
to compete to get it.  Slamming occurs because that is the competitive
system that has been setup to allocate dropped names.  If WLS is
implemented, we therefore recommend the additional implementation of the
MIT/Tucows proposal, which utilizes much simpler, fairer, round-robin,
first-come-first-served submission queues, to completely solve the usage
problem of the batch pool, and at a minimum to implement the simple
modifications to the current system, such as better return codes, to reduce
the usage "problem" with the current batch-pool model.
	2)	Making WLS subscriptions expensive will not reduce the load
in the batch pool.  The conventional wisdom is that fewer names will use
WLS, due to its high cost, and therefore, more names will be "left over" and
end up in the batch pool, resulting in more names there for slamming.  Since
we already established that slamming has nearly no cost to registrars, and
all the slammers slam at maximum rates already, more names dropped will not
increase the slamming. We disagree with Chuck Gomes' statement: "...it is
important that the price be set high enough to avoid high volumes of abusive
speculative subscriptions.  Otherwise, we will end up with a similar problem
with the WLS as we now have with the deleted names issue.  That has become a
very high cost item for us in the regular registration business." because
the load on the WLS system (note: not the batch pool) will not be affected
by the price of the subscription; it will be affected by time.  The load on
the WLS system will dramatically increase the moment certain knowledge of
names' to-be-dropped statuses is known (see #3 below).  
	3)	WLS usage will spike for about a minute a day and be will
quiet for the rest of the day. This is because intelligent purchasers, who
will number most as the end-users of the WLS system, will only purchase a
WLS subscription (or switch subscriptions to a different name) if they know
for sure the name will be dropping, otherwise, the $40 fee will go for
nothing, and they will be unwilling to take the risk on that happening.
Therefore, as soon as the list of to-be-dropped names is known (via zone
file or other means), the WLS subscriptions for the most valuable of those
names will be grabbed immediately.  All the registrars who will be playing
in this game will slam the ftp site to try to get the zone file first (or,
even worse, continually perform hits on the root name servers to see if a
name is still there, or whatever), then, as quick as they can, they will
reconcile it with demand lists from their customers and try to be the very
first to get a WLS on a valuable name that for-sure will be dropping.
Unfortunately this inherent flaw to the WLS model:
			a.	Requires the WLS system to be sized much
larger in order to handle requests during this load spike while the system
remains nearly unused the rest of the day.  
			b.	Means that only sophisticated purchasers and
fast registrars, not the common consumer, will be able to grab names that
will definitely be dropping.  The mainstream consumer will only be able to
wait-list names that may drop.  The mainstream consumer will have to take
the risk that the name actually drops.  This risk will be huge because most
names that are not already expired will probably not drop during the WLS
period, and those that do, probably did so on error and should be returned
to the registrant not to the WLS subscriber.
			c.	Means that knowledge of which names will be
dropping (not to be confused with expiration date data) becomes extremely
valuable and the registrar with this information will have an advantage over
the others.  The complete list is information that only the registry knows
for sure, but a registrar who is performing deletes in or out of the 45-day
window, will know some of it.  Since Verisign performs by far the most
deletes, the Verisign registrar would have an advantage over the other
registrars because it will know most of the list before the others. If WLS
is implemented, we recommend that this information be freely, equally,
simultaneously, and openly available in advance (say 3 days) to every
registrar, and at a minimum not allowed to be used by Verisign to advantage
SnapNames or any 3rd party or registrar by disclosing it in advance to
anyone.  Even with implementation of this recommendation, the Verisign
registrar will have an insurmountable advantage in ordering WLS
subscriptions because they would know first which of their own names will
certainly be deleted and will therefore be able to order the WLS before
anyone else.   The Verisign-registrar deletes, by far, the most names, more
even than their 50% market share indicates.  WLS is a subtle yet excellent
(because it is subtle) method for Verisign-the-registrar to renew their
names to people, other than the current registrants, without letting the
rest of the registrars have equal chance at doing so.  Other registrars do
not have an equal chance because the risk is greater for the rest of the
registrars to make WLS subscriptions on them since we do not know which of
their names will for sure be dropping, but they do. This problem is
exacerbated by the fact that there are many names (a million or more?) that
are currently held by Verisign-the-registrar that are queued to drop.  The
Verisign registrar could give/sell this information to SnapNames (or anyone,
if they don't use it themselves) to grandfather SnapName's "SnapBack" list
so that those subscribers get the WLS subscription before any other
registrars have a chance to get the subscription with the same amount of
risk.  The other registrars could get the WLS subscriptions but without the
knowledge that the name would definitely drop, so therefore, they won't make
the subscriptions. We do not see a way around this problem even if the WLS
system implementation and operation is shopped-out to a third party
unrelated to the registry or to any registrar, or even if the WLS
subscription price is lowered to very low levels.  We consider this a
serious blow to WLS competition among registrars and, unintentional (benefit
of the doubt), yet inherent flaw in the WLS system, and recommend, for this
reason alone, that the WLS system not be implemented at all.  And if it is,
that minimally, the Verisign registrar queue of to-be-dropped names is
completely deleted before WLS begins. 

We maintain that there will be no differentiation or value-add amongst
registrars for selling WLS subscriptions on names that are not for-sure to
be deleted.  The only way to differentiate in that part of the WLS market is
on price.  If a registrar wants to compete for the names that for-sure will
be dropping, that registrar will, unfortunately (because of the load it will
generate), be compelled to build a "zone-file & demand-list reconciler and
immediate WLS grabber" system (or via some other method, such as a registrar
who knows which of their own names are being deleted).
2.0 If WLS...  the $40 price is much too high.
	1)	A "parallel" registry system, as the WLS has been described,
we believe, cannot cost more than the current registry system to operate,
and definitely not more than 6 times the current system, which a price of
$40 per name-year implies.  In fact, since this "parallel registry" will not
have to perform any day-to-day operations for the names it "registers", such
as zone file generation, or 100% uptime name server services, and will be
utilized by far less (5%) names, it should cost much less.
	2)	The fixed costs to build a WLS system should be less than
the $3 million Snap Names claims it cost them to build their system.  I know
our system, which competes very well with SnapName's system, probably cost
less than $100K to implement.  Why not shop-out the proposed development
effort to an independant third party to at least find out the reasonable
range for the fixed development costs?
	3)	Other solutions exist that do not require payment to
SnapNames or any 3rd parties for intellectual property.  If SnapNames has a
patent or other IP on a wait-list service and it is required for the
monopoly operation of the WLS, we'd like them to tell us the USPTO Patent
number so we can all evaluate the value of it openly.  It wouldn't matter if
the proposed service was not a monopoly service, but since it is, shouldn't
we all know what the intellectual property fees would be paid toward?
	4)	Whatever price SnapNames currently charges for their service
is irrelevant to setting the price of this monopoly service.  Since
SnapNames, for whatever reason, is currently the only company offering
Snapback's, their "market" data is not useful or meaningful.
	5)	Due to fierce registrar competition, the registrars will not
be able to charge more than $1-2 for a normal WLS subscription.  The price
registrars are able to charge is not related to the price that we buy them
for, or to the "value" of the name itself.
	6)	Higher prices will not limit the load in the current batch
pool (more on this below).
	7)	A $40 WLS-subscription gives the holder the option to
purchase the name at a later date for $6.  The holder can purchase the name
only if the name becomes available (is deleted) and only during a period for
one year.  The subscription is not tied to a particular name but can be
transferred from one name to another at anytime before expiration up to
three times.  In other words, any name.  By contract with ICANN/DOC any name
can be purchased for $6 now and for $6 in the future from the registry.
There is no volatility in the price of names from the registry.  Isn't there
is a formula to calculate today's value of such an option? And, although I
don't have a financial background, I'm confident it will work out to be much
less than $40.  I, for one, cannot believe a rational person will pay $40
for an option, to maybe, within a year, get the opportunity to buy a $6
product for $6.  I can believe that a rational person will pay $40 for an
option to buy a $46 product for $6 up to one year in the future.  Therefore,
what a $40 WLS price really means is that the price of the product is not $6
in the future, but more.  I am no attorney, but wouldn't this price increase
on future names be in violation of the ICANN/DOC agreement?
	8)	The worst problem about the price of WLS is that all of the
alternatives are cheaper.

We do not believe that WLS is the best system to allocate deleted names.  It
is far from it. But if Verisign succeeds in implementing it, it will
continue to cause the technical usage problems we are supposedly trying to
prevent, while perpetuating the Verisign registrar's advantage.  We believe,
that because this is a monopoly product (unlike the email forwarding, URL
redirection, DNS, and other services the Verisign registry is now selling),
that ICANN/DOC must limit the price offered by Verisign (for any monopoly
product: WLS or any other system), to cover the reasonable estimated
start-up costs, and the ongoing operation, with a reasonable profit to
Verisign.   Since the risk is small, for one, since the model they propose
to use is already in use at SnapNames (and we assume they've analyzed it
thoroughly), we believe the built-in profit should be small.  We believe the
cost-plus price to be much less than $40 per WLS subscription.

Most of the costs Chuck Gomes mentioned in his recent post are either
variable costs or fixed costs dealing with technical development and
build-out.  The first cost, Licensing, could be eliminated by implementing
another model.  Development, integration, equipment are all fixed.  Since
most registrars know how much their systems cost to build/purchase, which
are similar to a WLS system, so we all know about how much it would cost to
build a WLS system, and it isn't $40/name, unless only about 10,000 names
are projected to be sold.  Many of us have built registrar system, which
include website, whois, databases, billing and collections, etc., for less.
We all know nearly all of the rest: customer support, customer service, is
variable and depends on volume. Since Verisign is already performing these
services (very well in our opinion) for the current registry operations they
would have little incremental costs to add similar service for the WLS
product.  His list is as follows:
1. Licensing the technology (initial and ongoing costs)
2. Integration of the technology into Registry systems
3. Testing and quality assurance of the system
4. Development and support of OT&E versions of the service
5. Development and support of registrar agreements of the WLS
6. Customer service (initial and ongoing, 7x24x365)
7. Equipment
8. Operational support (development, implementation, ongoing maintenance,
etc.)
9. Policy development
10.Billing and collections (this will be a new service factor for us, very
different than regular registrations)
11.Legal liability
12.Easy to use tools for registrars etc.

So, if WLS is implemented, we recommend that the price be just enough to
cover the build-out and ongoing operation of the WLS system, which is much
less than $6 per name, and we estimate on the order of $1.00 to $2.50.  Many
registrars, such as CORE, do much more yet operate on about the same gross
margins.  Alternatively, for fixed costs, we suggest a competitive method:
	1)	Make a specification, and open the development to a bidding
process, or
	2)	Have a competition to find a completely separate third party
to build and operate the entire system at a reasonable cost.


3.0 If WLS.... Subscription purchasing will be time-based.
	1)	Before expiration: Unknowledgeable or unsophisticated
(foolish) users will be the only people making WLS subscriptions before the
name's expiration date because it is unlikely that the name will be deleted
before expiration, and is even possible to not be deleted after expiration,
therefore, most probably wasting their $40.  The further before the
expiration date, the higher the risk that their money will go for naught
will be.   But at anytime before expiration their risk will be large.
	2)	After expiration but before knowledge of "will definitely be
deleted".  Because there is still a significant possibility that a name that
has expired will not actually be deleted, not many WLS users will purchase
WLS during this period because to do so would mean taking a risk that the
name will actually drop.
	3)	After obtaining certain knowledge the name will be deleted,
but before the name has a WLS subscription.  Obviously if you are deleting a
name for which you know there is WLS demand, you will have an insurmountable
advantage and beat all others to the WLS subscription for it (our main
problem with the whole WLS model).  During the minute when the information
becomes public is when we believe all the frantic WLS purchasing and WLS
switching activity will happen for the rest of the registrars.  Since
switching (WLS proposal Section 2, a, vii) will be limited to 3 times, if
WLS is implemented, we recommend that the number of "switches" left on a
particular subscription be made available to the registrar that performed
the original WLS via the EPP protocol.  It is probably, in our opinion, an
easier implementation to allow unlimited switches, so at least we wouldn't
have to keep track of them.
	4)	At WLS start-up any name that is more than 45 days after
expiration, and is worth more than $40 to someone, will probably get a WLS
subscription due to the fact that those names should have been deleted
already and therefore are very likely to be deleted sometime soon.  Since
Verisign-the-registrar knows which names will definitely be deleted, they
have an advantage on these names because others will have to take a risk
that they will actually be deleted.  We recommend these "backlog" names,
that may number in the millions, which should have already been deleted and
which are already queued for batch deletion, be deleted before WLS begins.
Also, including them in any "test" will throw-off numerical market results,
unless you just want to know what names, if sure to be deleted, are worth
more than $40 to someone.


4.0 If WLS.... FUD will increase.  
There would be no reason to purchase a WLS on your own name because:
	1)	You could take the $40 and purchase 6 additional years, put
the name on the existing "registrar lock" so that it could not be deleted
without removal of "registrar lock" and have $4 change.
	2)	If your names drops by accident or mistake, we assume, you
could still get your name back, so there is no need to purchase WLS
insurance.
	3)	If your name does not drop by accident, then it means you
wanted it to be deleted or not renewed, therefore, there is no reason for
you to have a WLS subscription on it in the first place.

There is little reason to purchase a WLS on a name that is set to expire
after the subscription period ends, because:
	1)	The only reason to do so would be to catch the name if the
registrant submitted a "delete" command intentionally.    At eNom, and I
suspect we are representative of all registrars, we rarely, if ever, delete
names during the registration period, and especially not after the initial
"5-day" period.  I'd say we did it in about 1 or 2 cases, max, out of
650,000 names.  There is no incentive for registrants to do so.
	2)	Names that are deleted on error should be returned to the
original registrants, as occurs today, in most cases (due to registrars
cooperating in these cases), but as should occur nevertheless.

There is little reason to purchase a WLS on a name that is set to expire
before the subscription period ends, because:
	1)	The purchaser does not know that the name will drop, or that
it will be renewed.  Since at least half of the names today are renewed,
their chance of actually getting the name is less than 50%.  Since most
"valuable" names are renewed, their chances of getting a valuable name are
even worse.

Therefore, we predict an increase in misunderstandings, and FUD (fear
uncertainty and doubt) in an attempt to get people to purchase WLS
subscriptions during these periods.  Registrars or resellers or others will
perpetrate fear of loosing your name in order to increase sales of WLS.  The
domain name industry could use less incentive for this kind of behavior that
is already giving our industry a "sleazy" image.  Best practices need to me
implemented, not just talked about.  We recommend:
	1)	Registrants are given the opt-out option so that they can be
told that they have nothing to fear, that no person, or worse, competitor,
could purchase a WLS subscription for their name and somehow get the name
without the registrant's purposeful deletion or non-renewal.   Opt-out helps
to remove FUD.
	2)	That if the WLS subscription has already been purchased, the
opt-out command would inform the WLS subscription holder of the registrant's
intentions to not delete the name by opting-out, and to show that if it does
get deleted, it was not intentional.
	3)	Regardless of the number of times a WLS subscription can
switch names, that a 5-day "buyer's remorse" period so that the registrar
will not have to pay the $40 if the subscription purchase was made
inadvertently, or if they buyer discovers his mistake in buying the WLS.
This is much like the 5-day period we have now with name registrations.
	4)	That it be mandated that the subscription buyer be told that
there is no "insurance" value (or other value) if he is purchasing the WLS
for his own name.
	5)	That a daily list be published that contained all the names
for which there was a wait-list subscription (does not include the
information of who purchased it) currently in effect so that any registrant
can go to any registrar (if their own registrar does not provide the "find
out if my name has a WLS" service, for example) and obtain the knowledge
that their name has a WLS-subscription on it.
	6)	That the registrar-of-record be told the "whois" information
for the waitlist subscriber for each of their names that has a subscription,
so that they could inform (potentially confidentially) their customer, that
their name has a wait-list subscription and who the person/entity is who
purchased the subscription.  Since the registry is not proposing to collect
this information, it must be made available via all registrars' whois
interfaces, adding to the complexity of this WLS "solution".
	7)	That WLS subscriptions be prohibited being sold for names
that already have their expiration beyond the WLS period. 
	8)	That if at any time the registration period for a name is
increased beyond the WLS period after the WLS subscription is sold, that a
message be sent to the WLS-holder informing him of the fact that the WLS
tied to that name is now worthless and should be switched (WLS proposal
section 2, a, vii) to another name whose registration expires before the
time remaining on the WLS subscription.

If WLS is implemented, and "opt-out" is not, in order to perform additional
protection services for it's registrant customers; eNom will be modifying
its registration agreement to reserve the right to not delete any name for
which there exists a WLS subscription, before or after expiration, and to
hold the name for the current registrant or change the registrant to any
third party after expiration.  Other registrars should be aware of this new
eNom policy before selling WLS subscriptions on names for which eNom is the
registrar, because those names will definitely not be deleted.

5.0 If WLS... Mainstream consumers will not be benefited.
Most assuredly the WLS system will not benefit mainstream users because:
	1)	Mainstream users who purchase a WLS service on a name
without the knowledge that it definitely will drop will probably, though
inadvertently, cause the name not to drop because the current registrant, or
anyone else, would know someone wanted their name and therefore would be
more likely to renew the name.
	2)	A mainstream user will have to know that the name they wish
to register currently has an expiration that is before the WLS subscription
period ends, or that the name they have a WLS subscription on just got
renewed past their WLS subscription period, which is information that
mainstream user may find hard to come by, especially if some of our
disclosure recommendations are not implemented.
	3)	Due to the timing of the order, only sophisticated users
will have the ability to order WLS subscriptions immediately after the
information that it definitely will drop is known.  By definition, only
insiders will know what names will definitely drop before the public.
	4)	There is no benefit to ordering a WLS subscription on your
own name, unless the policy is that a paid-up name, deleted on error, would
go to the WLS subscriber, a policy we recommend against.
	5)	Since WLS is first-come-first-served, the mainstream user
would have to beat a sophisticated user to the WLS subscription (or take the
risk that the name will actually drop), which would be difficult.  They'd
have a better chance in a deterministic (still FCFS), round robin
dropped-names allocation.


6.0 If WLS... differentiation will decrease.
The WLS model imposes the same model on all the registrars.  It gives the
registrars little room to differentiate except on price.  The current
"batch-pool" model allows the registrars to differentiate as evidenced by
the competing models that exist now: 
	1)	SnapName's WLS model
	2)	Name Winner's Auction model
	3)	eNom's Subscription model
	4)	Others
We believe the WLS model will eliminate this competition and
differentiation.  We believe the MIT/Tucow's proposal, implemented at the
registry instead of WLS, provides a method whereby the registrars can
continue to differentiate and compete because it allows for a variety of
models at the registrar level.  Differentiation may increase because it will
be easier, technically, for registrars to make different systems with
MIT/Tucows proposal in place.  This will benefit the "mainstream" user
because at least the mainstream user could go to a registrar whose model
best fits their needs and have a chance of getting their desired name.

7.0 If WLS... registrars margins will decrease.
The margins will decrease because registrars will lack differentiations and
innovations for which we can benefit consumers and charge money.  Due to
competition we will not be able to increase the price much over $40, no
matter how much the name may be "worth" because the potential WLS
subscription purchaser will go to the lowest price registrar since the
product is exactly the same registrar to registrar.  The price registrars
will be able to charge is not related to the "value" of the name as Chuck
Gomes suggests: "Depending on the value of the name, I would think there is
lots of room to play on your side.  There certainly seems to be plenty of
demand.  In fact, I would think that you as a registrar would have the
chance to make even better margins than we will."  We disagree with this
statement, although, as a profit-making entity, we wish he were correct.
We know that in the end, eNom will make money by innovating and by
efficiently providing products and services that have value.  No registrar
will be able to blindly mark-up WLS subscriptions 10, 30, 50, 100 (or more,
it is alleged!) percent and hope to make money on more than a few
subscriptions sold to a few confused consumers.
ENom, if we decide to offer the WLS service at all, will not offer WLS
subscriptions purchased before the expiration date on the name, for more
than cost.   We will not be selling WLS subscriptions linked to names that
are set to expire after the end of the subscription period because we
believe the name will not legitimately drop during the subscription period
and therefore, we believe, the subscriber will be spending money for
nothing.
	

In summary, we recommend that the WLS service be withdrawn by
Verisign-registry because it unfairly benefits the Verisign-registrar due to
that registrar's ability to obtain WLS subscriptions on 50% of the total
number of names with lower risk and therefore, with an anti-competitive
advantage, than the rest of the registrars.  If the WLS system were to be
implemented, we ask that our other recommendations outlined in this position
paper be implemented as well.
8.0 Questions: 
	*	It is proposed that subscriptions cannot be transferred to
another party.  What about between registrars?  Once a subscription "ripens"
can a registrar charge any price to actually register it?   
	*	Will SnapName's subscriptions be "grand fathered", or
presented to the registry at WLS start-up with priority or with certain
knowledge of which Verisign-registrar names will be dropping?  
	*	How many names are queued for deletion (how many expired
names are past the 45-day period)?  And will they be deleted before WLS
arrives?
	*	Since the registry proposes a one-year "test", would the
names that get WLS subscriptions toward the end of the test period get
one-year subscription periods (effective making the test 2 years from
go-live till those periods run out) or shorter periods, or would the
start-up period last, say one month, after which no more subscriptions would
be accepted so that the total test period would be 13 months?
	*	Will WLS subscriptions be sold for names that are within the
5-day "just registered and can be deleted without charge" period?

Attached is word document copy of the above text
 <<eNom WLS position.doc>>

eNom WLS position.doc