Comments and alternate proposals for application fees and non-profit TLDs

[Vittorio Bertola <vb@xxxxxxxxxx>]

Hello,

I am filing the letter I sent to the Board just after the Cairo meeting,
and the subsequent expansion on a request for more details (other
people's comments were anonymized and trimmed down to the minimum
necessary to understand). The letter regards mostly the issue of
application fees, but also the general philosophy and purposes of the
new gTLD program. I will try to develop a more comprehensive comment, if
possible, before the public comment period ends.


====
Dear Board of ICANN,

as I was standing in line yesterday morning in the Public Forum, but due
to prior commitments was not able to attend the "ad hoc" afternoon
session to express my views, I am sending them directly to the Board,
copying the Chairman, Vice-Chairman and ALAC Liaison so that at least
one of them can forward my message to the Board list, and I will publish
them somewhere for yesterday's audience.

Before I get to my point of substance... I guess that several people
already expressed their discomfort for what happened yesterday. However,
please let me reiterate that the Public Forum, where the community and
the Board discuss in plenary mode about the main topics of the moment,
is one of the most fundamental elements of ICANN's legitimacy and
accountability. Everyone knew since the beginning that at this meeting
the Public Forum would have been crowded and well attended, and the
decision to allot just one hour for it, then letting VIP speeches eat
even more into it, is a terrible mistake. I urge the Board to make sure
that there is ample time for Public Forums at every ICANN meeting -
given that this situation happens often, I see a need for clear
directions to staff by the Board.

Now - I would like to comment as a wannabe applicant for a gTLD
application which may or may not materialize, but that constitutes a
good proof for the remaining flaws in an otherwise well thought-out
draft RFP. Its main purpose is to save an ancient language and culture
which have been existing in my part of Italy for about a thousand years,
but which will disappear forever in twenty years or so, together with
the elderly people that still embrace them, unless we can succeed in
transitioning them to the Internet age.

A small group of volunteers has been working pro bono for years to
create online resources in this language - including, for example, a
Wikipedia edition. The existence of a gTLD specifically devoted to that
culture and language would make in our opinion a huge difference. It
would boost the sense of identity and community, and provide a visible
home to gather all efforts. However, this will clearly not be a business
opportunity - it is imaginable that initially the gTLD would have just a
few dozen registrations, which we would gladly give away for free
through a non-profit vehicle.

I think that what we would like to do is a deserving purpose, at least
as good as yet another dot com clone, and possibly better than the
abundant defensive registrations of any kind that we will see. To run a
TLD with such a few registrations, there is no need for big staff and
huge server farms - in fact, we are confident that we could get all the
time, skills and technical resources as volunteer work and in-kind
donations. However, even if we succeeded in this, we would still be
facing an impossible task to raise $185'000 now and $75'000 each year
just to pay ICANN fees, and we would likely score very badly against
operational and financial criteria designed for multimillionaire global
ventures.

Yet, if you think that what we are trying to do is obsolete, amateurish
or unimportant, please think again. This is the way all ccTLDs and gTLDs
started prior to the ICANN era, and most of them have become pretty
successful by now; actually, the only ones going for bankruptcy lie
among those picked by ICANN through its carefully drafted RFP processes.
This is actually the way almost every innovation happens over the
Internet, still today.

The Web? It wasn't invented by CERN, it was invented at CERN, by a
couple of individuals, in their spare time, as a byproduct of their real
job. Instant messaging? Peer to peer? Even innovations that overturned
billionaire industries were invented by one or a few individuals with no
money at all, or at most by small garage startups. What would happen to
innovation if the IETF required $185'000 to submit a new Internet draft?

I understand that there are costs attached to the establishment of a new
TLD, though $185'000 per application, even in an expensive country like
Italy, is enough to hire five or six people for one year for each
application, and one wonders why do you need all that work; and $75'000
per year to keep a TLD in the root, where the work required in the
absence of special events is literally zero, is plainly ridiculous.
However, if you want to extract money from rich applicants going for
remunerative global TLDs, or from big corporations with deep pockets
trying to protect their brand, that's fine; but please don't make other
uses impossible.

There are several pricing structures that could address this issue:
special prices for non-profit applicants, lower fees for TLDs that don't
reach a minimum number of registrations, or panels in cooperation with
appropriate organizations (say, UNESCO) to "bless" applications that
have specific cultural or technological value. Several people have
promised to submit practicable proposals in the next few weeks. But it
is paramount that ICANN doesn't sell out the domain name space without
putting in place features to address this issue.

In the end, while applicants will be judged by the RFP, ICANN will be
judged by the overall set of TLDs that it will add into the root. It may
get 500 or more of them, but if 90% of them will be private corporate
registrations, and the rest will be dot com clones with some kind of
vague specialization, ICANN will have failed.

But, looking also at other aspects, I am also afraid that the failure
might end up being much deeper. ICANN is becoming a well managed
business entity, through increased staffing and the introduction of
corporate best practices. However, ICANN is not just a business entity -
it is a strange beast with much more than that into it. What is optimal
for a business corporation might actually make parts of the community
feel not at home any more; and might make ICANN lose touch with its
roots, with the nature and spirit of the Internet. If this happens,
ICANN is doomed - all the governmental deals and business partnerships
won't be enough to preserve its prestige and credibility.

I see as one of the primary strategic roles of the Board that of
ensuring that the decentralized, flat and free nature of the Internet is
preserved, or at least not attacked, by the policies that ICANN adopts,
and even that these policies contribute to, or at least do not stifle,
the fulfillment of Millennium Development Goals and other worthy
objectives in terms of development and human rights. These are not just
high sounding words, they carry a meaning that must trickle down into
everything ICANN does when it comes to policies. When you are tasked
with a fundamental role in coordinating the Internet, there's more to
life than business as usual.
=====

> > would it make a huge difference to your application if it occured in
> > a first round or in a second round later?

IMHO, the main issues for which people won't be willing to wait are the
following:

1) the availability of strings; if this round is pretty big, then no
reasonable strings may be available in a second round;

2) the ability to sustain the momentum; as people are building momentum
to be ready in Q2/Q3 '09, having to delay for a couple of years might
kill the effort. If you're a group of volunteers, people will go do
something else; if you're a smaller business, you can't afford to keep
operations going;

3) not for my kind of application, but for others, the ability to
compete; you may have an innovative idea, but someone else could start
doing it before you, since you have to wait for the next round.

And of course, the history of ICANN is full of promises for a "prompt
next round" that have been regularly broken, so I guess that few people
will be available to trust ICANN on that; they will insist on getting it
this time.

I actually share the concern about not really being able to know what
set of gTLDs will in the overall be added to the root thanks to this
process.

However, I think that there is a fundamental error in the way ICANN
plans to define which set of applications gets first, which is a
selection by imposing artificially high fees. (I hope that we don't have
to argue on whether these fees are artificially high, but if we need to,
I'll be happy to oblige.)

This is why talking about certain applications as "applications that
need sponsoring" is IMHO wrong: they need sponsoring only because the
bars are willingly set too high.

There has been a decade of discussions on whether the addition of TLDs
should be "purpose-driven" or "purpose-neutral". "Purpose-driven" means
that there is somehow an evaluation of whether the proposed TLD is a
"useful" and "worthy" addition to the Internet; if it useful, you create
it, otherwise you don't. "Purpose-neutral" means that as long as you do
not harm others, repay ICANN of the cost (since ICANN is supposed to be
a non-profit, so it should not add profit margins to its costs...) and
show some minimal technical capability, you should be granted what you want.

What ICANN is trying to do, however, is neither of the two. In fact, it
is a purpose-driven allocation in which the purpose is considered worthy
as long as it can raise $185'000 + $75'000/year.

As a result, you are actually privileging purposes such as "global dot
com clone", "megacorporate defensive registration" and "vanity TLD for
billionaires" over purposes such as "hobbyist community TLD", "small
startup trying to do something new" and "non-profit campaign with
cultural objectives".

So you do limit the size of the first round, but, IMHO, you limit it in
a way that is non-neutral and potentially harmful to the future
development of the Internet.

It might now be too late to change the approach too much; however, I
think that you need to rebalance the criteria and the fees so to be more
purpose-neutral, or, if you're worried about the overall result, to edit
the process to make it more purpose-driven.

A way for the former approach would be to make fees proportional to the
intended size and revenues of the TLD. For example, non-profit
applicants for community based, non-competitive TLDs should get much
lower fees. Also, TLDs with less than, say, 50'000 registrations should
get lower fees as well - they would have to pay the rest of the
application fee if they ever go over 50'000 registrations (or you could
have more levels). And so on.

A way for the latter approach would be to part the process in two - a
first phase in which someone, for a cost-based fee, proposes a string
for a certain purpose and gets it evaluated and approved; a second phase
in which approved strings are published in search of applicants, with
different criteria and fees according to the intended purpose (or
better, 3-4 standard sets of criteria/fees according to types of
purpose). There is support for .web as a .com clone? Let's auction it.
Wales asks for .cym? Let's delegate it to the entity that better
represents the Welsh community through a beauty contest. Dot mobi tried
something like this - specifically tailored RFPs for specific second
level domains - and it was quite successful.

In any case, I think that at least for the first round you should do a
"dry run" and/or keep an "emergency shutdown" button for yourselves; I
mean, the Board should say from the beginning "since this is the first
try of this process and we're not sure we got it right, we reserve the
right to refuse any application at any point in the process and refund
the money". Yes this is discretional, but it's your only way to avoid
the risk of embarrassing outcomes, such as ".<blasphemy>" making it
successfully throughout the process, or Britney Spears spending the
money for ".britney". (Maybe not everyone thinks that these are
undesirable outcomes, but I do.)

In the end, as I already said, ICANN will be judged by the overall set
of additions. Everyone will complain if the set is too unbalanced
towards a specific type of applicant, or if they don't get anything of
what they want. Be sure to cater to everyone's needs, as much as humanly
possible.

=====
> > Oblige me...what do you mean by "artificial", in an edifice which is
> > entirely an artifice... ie constructed to serve a purpose?

I mean, they are artificial since they are higher than they need to be
to cover the costs of processing the application and keeping the TLD in
the root.

The GNSO recommendation says that the fees should in the overall cover
the cost, though they could be differentiated by type of applicant.
Since they are not differentiated, the fees should be equal to the total
cost divided by the number of applicants.

Let's start from the application fee. In the "cost consideration" paper,
ICANN says: we spent $1.8M to process ten applications in 2003, so the
processing cost was $180k/application. Then ICANN also estimated the
cost bottom-up. The paper says that ICANN already spent $12.8M, mostly
in salaries, on this program, which makes $26k/application. Then it
estimated the work-hours required by the processing itself, attaching a
probability to each possible processing path; here there are no
published numbers to support the conclusion, but it adds up to
$100k/application. Then there is a final figure, which is what in a
business plan you usually add as "misc" or "reserve" to cover for
unexpected expenses, which was estimated by Willis to be
$60k/application. The total, $185k/application, is actually very similar
to the 2003 cost.

Now, we have really no information to verify these evaluations - you
would have to go through the entire data set. Seen from the outside,
however, there are some things that are not convincing, and most people
I've talked to have questioned them.

The most frequent points are:
- how come that per-application costs for processing 500 applications in
a structured manner are the same than for processing 10 applications on
an experimental basis several years ago?
- how can you have a 'risk fee' of 1/3 of the total? isn't that
excessive, or just a way to inflate the fee?
- almost all the cost is made up by people's work; now, with $185k one
could hire 4-6 people for a year, per each application - or, with $80M
one could hire hundreds of people for a year; do you really need that
much work? what will those people be actually doing?
- even if this was the actual cost of processing these application,
isn't ICANN making this thing excessively complex? couldn't it just pick
a simpler process that would have a lower cost?

(I disagree with the last one, but there's a line of thought going in
that direction, to the extreme of "why should it cost $185'000 to add
something into the root if it costs $7 to add something in .com". In the
end, perhaps the right question is: did ICANN do what it could to
streamline the process and create economies of scale?)

So, I respect your evaluations, but the supporting arguments weren't
really convincing, and I didn't find many people in Cairo thinking that
$185'000 actually matches the real cost. Most people really saw the
proposed two-round fee policy as a marketing exercise, as when you
market a new IT product and first keep your prices high to milk the
early adopters, then later you lower them to expand the size of your sales.

However, things get much worse when you move to the yearly registry fee.
If you don't do anything - you don't change name servers, you don't
redelegate, etc - the additional cost of keeping an existing TLD into
the root is technically zero. Sure, ICANN has to keep a name server
running, but this isn't really related to the fact that ICANN is
managing the root zone, and in any case ICANN would do it the same
without the new gTLDs. So there's really no way that I can think of, to
justify $75'000 per year.

In the end, I'm fine if ICANN wants to impose reasonable per-operation
fees on changes in the root zone. I'm also fine if ICANN wants to pass
onto these new registries a part of its general costs, exactly as it
does to the existing gTLD registries; but then, it should be done
through the same policy and fee scale for all gTLDs, old and new, and it
should be proportionate to the actual size and/or revenues of the gTLD.
I'm not sure why would ICANN have to deviate from the standard
$0.something fee per domain name that it already applies to gTLDs; if
you really want a yearly minimum even from smaller TLDs, IMHO it should
be somewhere in the range of the hundreds of dollars, not more.

> > Interesting - so, if we agree that the fees are appropriate, you
> > would re-classify this as purpose neutral?

Yes (if by appropriate you mean matching the cost). I think I would
still be pushing for a mechanism to sponsor certain "pro bono"
applications, but I think that it would reverse the onus of the proof -
it would be the applicant's task to prove that it really deserves
subsidization, not ICANN's task to prove that there really is the need
to impose such high fees.

> > However, I greatly fear that your mechanism, below, just imposes
> > more delay on applicants, ( more steps in the process) more
> > subjectivity as people judge levels of likely incomes and numbers of
registrations etc,

I don't see how you can be delayed by having to apply a per-size fees
table. Applicants could simply be billed as they go live, each time they
cross a threshold and move to a higher dimension (I assume that
registration data will have to be escrowed and that ICANN will have
access to them, so you don't even need to rely on the registry's
self-declaration of how many registrations they have). If you prefer,
you could ask the applicant to declare their initial intended size and
then staff vet it in some way, to prevent the delay in cashing the
higher fee on big applications; or you could restrict access to lower
fees to community applications only. But that's already an optimization.

> > The fundamental issue is fairness - this is "user pays" and
> > impecunious applicants are not supposed to have their business
> > dreams subsidized by the registrants of other TLDs.

You seem to assume that every impecunious applicant has a business
dream. Some just want to run a TLD  :-)

They want to make things happen in society, in culture, in technology.
Just by copying my letter to the At Large list, I discovered someone in
Argentina whose grand-grandparents came from Turin and who can't wait to
register in my TLD. That's already a reward, but it won't turn my idea
into a business or pay the fees for me.

Of course, no one would complain if a pro bono TLD application could
raise enough money to pay for the servers and for the time of the people
working on it. So be sure that ICANN gets a fair share of the money if
there is any money to be made. But if there's not, then ICANN should
just recover its costs stripped down to the bare essential.

And, there's many concepts of fairness... in Monday's workshop in Cairo,
Bertrand de la Chapelle made exactly the opposite point to yours -
fairness is letting each applicant pay according to its purpose and
size. He said, "I prefer something that could be gamed in the future to
something that is currently already unfair".

In the end, concepts like "fairness" depend on how we see the world, and
that's why this process is so difficult... and interesting  :-)

Regards,
-- 
vb.                   Vittorio Bertola - vb [a] bertola.eu   <--------
-------->  finally with a new website at http://bertola.eu/  <--------