[icann-eu] Domain Name Economics

[Thomas Roessler <roessler@does-not-exist.org>]

I'll try to paint the "big picture" of ICANN and related topics in 
this message.  Topics addressed include competition in the root 
market, competition in the TLD market, user and domain name holder 
representation.  Note that all this may just be silly nonsense - if 
so, please tell me about the flaws in my thinking.


The questions I'll be asking are mostly these: What does this or 
that player have to sell?  Why can the player sell that good?  What 
happens when multiple players try to sell the same good?


		 Registrars vs. Domain Name Holders


First, consider the relationship between registrars and domain name 
holders.  Obviously, registrars are selling sexy character strings. 
Further, they are selling the visibility of the domain to the 
Internet - including uniqueness.  It's obvious that it's close to 
impossible for the millions of domain name holders we have to go for 
visibility and uniqueness individually - in this case, they'd have 
to individually deal with all the DNS server operators on the world 
- a task which seems to be virtually impossible for most of the 
domain name holders.


		      The Natural Root Monopoly


Now, take this relationship one level up: Domain name holders become 
ICANN-accredited registries (and their customers, the registrars), 
and the registrar's role is taken by ICANN.  Still, what's sold is 
visibility, and uniqueness. However, there are only few players on 
this level, and - today - most of them are quite powerful.  Thus, it 
may be affordable - but expensive - for these players to directly 
buy visibility from DNS server operators (mostly ISPs).

It's interesting to ask what the situation would look like when 
competing root service operators would indeed offer pointers to 
conflicting versions of the same TLD.  To let things look more 
similar to today's situation, let's also assume that a certain set 
of old TLDs is unlikely to be subject to conflicts - for instance 
.com, .net, and some of the larger ccTLDs (.uk, .de, ...).

[The following sections are almost verbatim from a posting to
ga-roots.]

Assume that an established instance of, say, .web exists.  Assume 
that someone else creates a competing .web.  In this case, three 
things can happen:

- The second .web is perceived as nonfunctional for all the reasons 
   Kent Crispin's internet-draft lists, but does not strongly affect 
   the operations of the first one.  This would probably lead to a 
   shutdown of the entity which offers the competing .web, resolving 
   the issue.

- The situation is the other way around: The new .web is launched 
   with a large base of possible users, say, as a new TLD in the 
   ICANN root, or in an independent Tucows root.  In this case, it's 
   extremely likely that the earlier .web would be put out of 
   business - just like what's probably going to happen with Leah 
   Gallegos' .biz.

- Finally, both players may have similar strength in terms of 
   resolvers which access the respective root servers.  In this case, 
   all the problems Kent lists in his ID will occur in the worst and 
   strongest way possible.
   
   This effect would mean that both instances of the TLD are 
   preceived to be non-functional by end users and domain name 
   holders.  Given the alternative of other TLDs (such as the
   traditional gTLDs) which don't have these problems, registrants
   will most likely go for these TLDs (Kent neglects this argument
   when he talks about the Standards war - and indeed this is a
   critical point because it means that an extended standards war
   would be unattractive for players). The competing TLDs wouldn't be
   economically interesting any more for any of the parties.

   (Why go for .biz with all its problems when you can as well use
   .com?  I'd expect that this consideration would even override
   political preferences for, say, English and French .biz'es.)

To summarize, this means that the introduction of competing 
instances of the same TLD by weak players would ruin these players, 
that the introduction of competing TLDs by extremely strong players 
would ruin their competition, and that the introduction of competing 
TLDs by several players of similar strength will ruin all of them, 
by making the TLD unusable and by making all of the players unable 
to deliver on their promises of visibility, stability, and 
uniqueness.

In particular, we can conclude that plans of weak players to 
introduce new TLDs when they know that strong competition may come 
in the future (or even exists) avail to economical suicide, and can 
most probably be neglected.  (Sorry, Leah, but you don't have a 
chance.)

Thus, only the strongest players could reasonably offer "alternative 
TLDs".  However, even for them it would be extremely risky to 
introduce TLDs which compete with others' offers.

Assuming that the strongest players are of similar strength (and 
they would be after a probably short time of chaos and destruction 
;-), you actually arrive at an argument which is quite close to the 
one which underlies the Mutually Assured Destruction doctrine (US 
and USSR each need the potential to destroy the other one, so none 
of them can afford to attack): Give all the strong players the 
opportunity to introduce competing instances of the same TLD, and 
they won't do this because they'd risk ruining themselves, or being 
ruined by the competition.

As a consequence, it seems highly unrealistic that competing 
instances would actually become reality in a way which would pose 
all the problems Kent lists.  Hence, it seems likely that, with an 
uncontrolled root market, players would quickly agree on a common 
ICANN-like organization.  Hence, it is likely that some kind of a 
natural monopoly would occur at some point of time on the root zone 
and TLD market.

[end ga-roots excerpt]


Also, it should be noted that only such a root monopoly would enable 
smaller players to participate in the TLD market: It protects these 
small players from larger players who may try to squash them using 
conflicting TLDs with - at least - a great destructive potential. 
It also makes global visibility cheap.


			New.net's White Paper


Let's compare this point of view to the New.net whitepaper on the 
TLD market, where they are proposing their "Market/Consensus 
approach", see <http://www.new.net/NewnetPaper.pdf>.

Basically, they are proposing that every player in the market should 
have to bear the cost of making his TLD visible to users:

| Similarly, companies with new TLD or "alternative naming" 
| strategies can use incentives to persuade ISPs to "turn them on" 
| and reach Internet users.  In addition, Internet users can ask 
| that their ISPs enable their domain name servers to recognize 
| New.net's or others' TLDs.  The ISPs will be able to choose 
| whether to do so based on the incentives in place and the demands 
| of their customers.  Yet unlike the cable situation, users can 
| switch ISPs fairly easily (admittedly with some potential 
| switching costs such as changing email addresses) to get access to 
| domain names that the users want to access.  Users also are 
| empowered by the availability of software that enables them to use 
| alternative domain names if they happen to use an ISP that chooses 
| not to provide the user with access to such domain names. 
| Accordingly, users are given a large amount of freedom of choice 
| and control over how they wish to use the Internet.

| The benefits of a market-based approach are clear.  By allowing 
| companies to develop new ways of working within the DNS technical 
| system, to raise capital, to market their products, and to do 
| everything in their power to serve their users, ICANN and Internet 
| users will benefit in two ways. First, ICANN will be able to 
| conserve its resources and focus on developing better technical 
| standards to enhance the DNS.  Second, if a company is trying to 
| serve customers without the shield of the virtual monopoly of 
| being an ICANN registry, it will by necessity be more responsive 
| to serving its users to gain acceptance.

In the discussion around these sections, the authors of the white 
paper are talking a lot about the cost of "splitting the root". The 
cost they aren't even mentioning is the greatly increased cost for 
making TLDs visible on a global scale.  In fact, new.net claims that 
this cost is healthy because it will make registries and registrars 
more customer-friendly.



			Breaking the Monopoly


Next, we ask what could break the monopoly over the root which we 
described as natural above.  

First, what situation would lead rational players to the conclusion 
that they should take the risk of buying together visibility 
themselves, and bypassing the root supplier - just as new.net does?

The recent events surrounding ICANN give some indication on what 
this may be:

- The monopoly could be unable to deliver its promises even to 
   strong market players - indeed, new.net is just one example. 
   Think about the Tucows representative who was talking about his 
   company being in a unique situation to launch an alternative root 
   system "as a defensive measure".

- The monopoly could be too slow, too reluctant, or too expensive in 
   adding new TLDs.

- The side effects of buying visibility from the monopoly could be 
   devastating.

Also, the monopoly could be broken or changed by external forces. 
In particular, the following players in the game come to mind:

- ISPs.  They could easily start to sell visibility themselves. 
   (Which is basically what new.net suggests.) This may have the net 
   effect of producing a different root zone monopoly, but mainly 
   under the control of the ISPs.

- Political and legal forces, which can change the rules according 
   to which the game is played - these are particularly powerful 
   since the United States Government still has a critical word to 
   say in the traditional root monopoly.

- Current "clients".  Just imagine Verisign and the ccTLDs team up
   to create a root service of their own, and don't pay money to 
   ICANN any more.  This is, in fact, the card the ccTLDs are 
   currently playing.

It's worth noting who's missing from this list:  End users and 
domain name holders.  I'll discuss these below.



		    Side effects of the Monopoly


What side effects does the root zone monopoly have on those at the 
end of the food chain, end users and domain name holders?

First, let's note that it may indeed be quite natural for an 
ICANN-like monopoly on the root zone market to impose certain 
conditions on its clients - such rules of the game can even be in 
the clients' best interests, in particular, when they help to defend 
the clients' interests against some of the possibly destructive 
forces listed above.  They can also be economically reasonable since 
they save clients from the costs connected with developping such 
conditions themselves.

One of the best examples for such regulations is, of course, the 
UDRP, which is being adopted by most of the new ICANN TLDs.

Such regulations in particular mean that domain name holders are 
seeing a regulatory monopoly: Without competing regulatory regimes, 
domain name holders don't have a choice, but to abstain completely 
 from registering a domain.  This choice is, of course, quite poor 
since the incentives for registering and using a domain are quite 
strong, and since most domain name holders are not directly affected 
by the regulatory regime (at least that's what they think).

The perception of the regulatory regime as benign by the large 
majority of domain name holders also means that they won't pay the 
costs it would take to buy services from a monopoly-breaking 
registrar which offers a better regulatory regime.

Thus, in order to actually break the monopoly, the regulatory regime 
would have to be extremely draconian, and in fact endanger the 
ability of registrars to deliver on the promise of stability in the 
perception of the domain name holders.

Finally, note that even with a new.net-like competitive model, model 
regulations may emerge which are adopted by players in order to save 
costs.  In particular, the kind of competition they suggest wouldn't 
necessarily lead to regulative competition!



       Representing Users' and Domain Name Holders' interests


 From the above, we can conclude that as long as domain name holders 
are sort of satisfied by at least some registrars' offers, it's 
unlikely that the root monopoly of which these registrars are part 
would break.

Thus, there are close to no market-driven feedback channels for 
users or domain name holders.  Users and domain name holders are, in 
particular, normally not among the possibly destructive players I 
listed above.


Thus, they can only exercise influence by allying with destructive 
players.  Now, let's go over the list of possibly destructive 
players:

- ISPs.  As long as supporting the monopoly doesn't mean you are 
   losing customers to competition which doesn't support the 
   monopoly, there is no incentive to attempt to break it.

- ccTLDs.  Since the more powerful ccTLDs can be assumed to be among 
   the TLDs which are unlikely to be attacked during a root split, 
   they are among those players on the market which are most probably 
   least interested in the monopoly, and possibly most interested in 
   destabilizing it.  

   Breaking the root zone monopoly and destabilizing the gTLDs would 
   in fact help the ccTLDs to squash a lot of competition.

   In particular, the ccTLDs don't seem to have a natural interest in 
   helping to better adopt the monopoly's regulations to users 
   wishes.  Just the opposite is, in fact, the case.

   (A similar argument could be applied to the operators of other 
   well-recognized TLDs.)

- Legal/political interests.  These are apparently the only ones who 
   are left.  And, in fact, the business and intellectual property 
   constituencies seem to be demonstrating such an alliance, 
   frequently raising similar issues.


			    Consequences


As a consequence, we should not expect most of the powerful players 
in the current ICANN process to support user and domain name holder 
participation.  In fact, a private-sector organization like ICANN is 
unlikely to be bottom-up for purely economical reasons.  If such an 
organization is expected to serve the public good, and to take end 
user and domain name holder interests into account, pressure is 
needed by those political parties who are in charge of "nuclear 
options".

As another consequence, the ccTLDs' move to become a supporting 
organization of their own looks extremely natural.  On the other 
hand, the bundling of TLD operators, domain name holders, 
intellectual property interests, and individual domain name holder 
interests in the DNSO looks extremely unnatural.

-- 
Thomas Roessler                        http://log.does-not-exist.org/