ICANN ICANN Email List Archives


<<< Chronological Index >>>    <<< Thread Index >>>

No more gTLDs until existing ones show greater usage, Ascension allocation method

  • To: new-gtlds-pdp-comments@xxxxxxxxx
  • Subject: No more gTLDs until existing ones show greater usage, Ascension allocation method
  • From: George Kirikos <gkirikos@xxxxxxxxx>
  • Date: Thu, 22 Dec 2005 14:33:14 -0800 (PST)


This email is in response to the invitation for comments on new gTLDs


I think we should be guided by the toll-free numbers allocation system.
Initially, there were 1-800 numbers. Then, as 1-800 filled up, 1-888
was created. Subsequently, 1-877 opened as 1-888 filled up. And so on
with 1-866. Similar city area codes were opened up as NYC or LA ran out
of phone numbers.

While it's not a perfect analogy (since there are a relatively small
finite number of toll-free numbers, as opposed to an almost infinite
number of 63-character domain names), if one considers the space of
"good" domain names (e.g. 5 characters or less, or single words, or
meaningful two-word combinations), the analogy becomes better.

Thus, the question becomes whether the .info and .biz gTLDs, the ones
added since the initial com/net/org TLDs, are "full".

I would argue that they are not. According to www.dailychanges.com,
there are approximately 2.6 million .info domains (many of which were
given away for free) and 1.3 million .biz domains. This compares to 46
million .com domains, 6.7 million .net and 4.1 million .org domains.

It would seem to me that .biz and .info need more time to fill up,
before considering adding more  gTLDs. They should at least start to
reach the levels of .net, before adding new ones.

The numbers above refer only to total registrations. Another "metric"
would be in how actively the domains that are registered are being
utilized, i.e. how much traffic they receive. A lot of the domains are
parked, and thus have greater potential to be actively developed in the
future. Such development should occur, before opening up new TLDs.
Ownership changes amongst existing domains over time would lead to the
best utilization of a domain name in a given TLD (thus the importance
of transparency and frictionless trading between existing domain
holders, to ensure that there are no institutional barriers to the best
use of a registered domain, that it find the owner who will use it

Otherwise, if one simply opened up new gTLDs using a haphazard
approach, without concern for the utilization of existing gTLDs, what
would happen is that the most "elite" character strings (i.e. the best
100,000 words, like "sex.tld", "music.tld", etc.) would be snapped up
immediately on a speculative basis, or an abusive basis (TM
infringement), and the development of those existing domains in gTLDs
like .biz and .info would be hindered. Essentially, one might end up
with ghost towns in .biz and .info, leading to a possible failure of
those gTLDs due to lack of development. It would be like what would
happen to real estate development in a suburb, if a government was
contemplating the idea of opening up competing areas for development.
Development interest in the suburb would be diverted to other lands,
due to the oversupply of land.

As I see it, we have 2 choices. One is to have a relatively small
number of very popular  gTLDs, e.g. 5 million+ domains actively
registered  and developed in each successful gTLD. The alternative is a
large number of niche gTLDs, e.g. 100 gTLDs with those 100,000 best
words registered in each one.

I tend towards the first choice, a small number of very popular gTLDs,

1) there would be less consumer confusion (e.g. if I see a TV ad for
product.info, and later have to recall the website, there's a lot
better chance of me remembering it if there are a small number of
possible gTLDs; if one is bombarded with .info, .shop, .store, .travel,
.music, .sex, .flowers, etc., one would be confused and probably
remember ".com" instead of the actual gTLD). 

2) I don't think we want to end up with "hobbyist" toy gTLDs that
pollute the purity of the namespace. For the most part, hobbyists who
are proposing their TLDs (to protect the guilty, I won't name them, but
I think we all know who they are!) can accomplish most, if not all, of
what they want with a sub-domain (3rd level domain) of an existing
gTLD, and don't need a gTLD. 

3) Many of those "toy" gTLDs are abusive, in the sense that they seek
to profiteer from defensive registrations by trademark holders during
"sunrise" periods. e.g. if Coke, Pepsi, and other Global 2000
corporations have to fork over $300 for each trademark, plus their
time, the operators of those hobbyist gTLDs reap windfall profits, at
no risk (i.e. those sunrise profits tend to be more than the startup
costs of their registries). 

I'd also like to remind folks of Tim Berners-Lee's (and the W3C
Technical Architecture Group's) opposition to new gTLDs:


In conclusion, I think unless there are truly COMPELLING and INNOVATIVE
reasons to introduce new gTLDs, there should be a long delay in
introducing new ones (perhaps 5 or 10 years, to let the existing ones
"fill up"). Metrics can be established in the meantime to measure how
"full" the existing ones are (e.g. using stats from Alexa, Google, or
other surveys on website utilization), both for websites and other uses
(e.g. domains can be used to point to other resources, beyond websites,
like ftp, email, phone numbers [enum], etc). 

This ensures the continuing stability of the internet. After all, the
Internet is no longer an "experiment", but is an established part of
everyday life.

If, despite the above, ICANN still feels new gTLDs should be
considered, I would like at this time to propose a gTLD allocation
method scheme, which I'll call the "ASCENSION ALLOCATION METHOD", that
I've not seen suggested before. In particular, it would be a
combination of market-based methods (due to the Coase Theorem, see
http://en.wikipedia.org/wiki/Coase_theorem ), and a "beauty contest".
In particular, a new gTLD applicant would have to meet the following

1) Demonstrate existing ownership of the proposed gTLD string in *ALL*
2nd level domains in com/net/org (and perhaps biz/info too). To
illustrate, if one wanted to start .EXAMPLE, one would already have to
be the owner of example.com, example.net, example.org (and possibly
example.info and example.biz too). While this may be a windfall for
.com owners, by the Coase Theorem it is a market-based approach, as any
prospective applicant would need to acquire the relevant domains
(perhaps at great cost) in order to be able to "promote them" from a
2nd level string into a top-level domain. Even a .com owner would have
to buy out the owners of .net/org/biz/info, potentially at great cost.
The market would decide whether its best usage is as a single gTLD, or
as a set of 2nd level domains. This promotion or conversion, from a set
of 2nd level domains into a gTLD explains the naming of the method as

2) One would have to give up all trademark rights to the gTLD string
(e.g. if Pepsi or Yahoo wanted to start .PEPSI or .YAHOO, the terms
"Pepsi" and "Yahoo" would have to be relinquished as worldwide
trademarks). This would ensure that the gTLDs are basically generic
(like .STORE, or .SHOP, etc.) or are newly created strings not
conflicting with existing TMs.

3) The owners of the gTLD could not, directly or indirectly via related
entities, own subdomains of the new gTLD. For example, if one owns
Hotels.com/net/org/etc., and wanted to form .HOTELS instead, to have
Virginia.Hotels, California.Hotels, etc., one can't continue to operate
Hotels.com (which would no longer exist, i.e. it would be surrendered
to IANA as a reserved name) or any of the .HOTELS subdomains. Those
would have to be owned by the public, via registrations. i.e. the
owners of the proposed gTLD have to decide, like VeriSign had to years
ago, whether they are in the registration business, or whether they are
in the content business, but not both.

Given the above 3 were market  based methods, there'd need to be a
further "beauty contest" component, to prevent namespace pollution, to
show there'd be a substantial public benefit, and to limit the overall
growth of new gTLDs (i.e. to ensure there aren't 5000 junk/toy/hobbyist

4) For "ascension" to occur, the owners of the proposed gTLD must
launch and operate it as a quasi-TLD for a period of at least 5 years,
using subdomains. For example, if the proposed gTLD is .SKYPE, to give
a realistic example (assuming they've met #1 through #3 already),
domains would be allocated as example.skype.com, example2.skype.com,
etc, could have emails as user@xxxxxxxxxxxxxxxxx, etc.) 

As mentioned earlier, for most toy gTLDs, they can accomplish most of
what they want via subdomains, and don't need a gTLD. Operating as a
quasi-TLD, until it "proves itself worthy" of ascension to a gTLD,
allows ICANN and the public to separate the wheat from the chaff, the
serious proposals from the hobbyists, and thus avoids the pollution of
the global namespace with junk.

Quasi-TLDs would be subject to all the same ICANN standards (e.g. UDRP,
ICANN fees, equal access for registrars, WHOIS, etc.) as grown up

During this quasi-TLD period, the 2nd level domains would revert to
IANA as reserved domains, with the registry operating as a registrar
(e.g. if Sex.com was the candidate (and sex.net/org/biz/info were
acquired), since Sex.com could no longer be operated, via Requirement
#3 above, it would now be owned by ICANN/IANA).

5) After 5 years, a "beauty contest" would occur to ensure that the
proposed gTLD is "successful" and thus worthy to ascend to gTLD status.
Metrics to determine whether a quasi-TLD was successful would include
such things as the number of registrations (5 million+ would seem to be
reasonable), the percentage of internet traffic routed to that quasi
TLD (e.g. if *.skype.com was successful, with hundreds of millions of
registrations, and 2 or 5% of internet traffic, it could convert/ascend
into .skype and have emails like user@xxxxxxxxxxxxx instead of

Compliance with UDRP and other ICANN policies could be part of the
beauty contest. As percentage of internet traffic could be a
requirement, this would naturally put a hard limit on the number of new
gTLDs that could be added (e.g. if one required 2%, that puts an upper
bound of 50 new ones). Alternatively, if there's no traffic limit, one
could do an auction, or some other scheme (e.g. allocate to the quasi
gTLD with the highest number of registrations or greatest public

6) Public comment would be solicited, including those of existing
registries, including ccTLDs (this would help ensure .coom or .cmo
doesn't become a registry, as one could imagine *.coom and *.cmo would
get a lot of traffic, due to typos, but could meet some of the above

7) Other requirements based on input from the GNSO constituencies
(consider the above proposal a work in progress).

As one can see, the above requirements would set the bar very high for
a set of 2nd level domains to convert or ascend into the top level.
Money alone (via an auction) wouldn't suffice, as a gTLD would have to
PROVE itself as a successful quasi-TLD first. But, I think it would
satisfy many concerns. In particular, existing registrants of other
TLDs don't need to worry about dilution, as the proposed new TLD
registry operator has to first gain control of the relevant existing
gTLD names. e.g. an operator of .xxx would first need to own xxx.com,
xxx.net, xxx.org, etc., and operate it successful for games.xxx.com,
music.xxx.com, etc. for millions of registrants as a quasi gTLD, before
it became promoted/ascended into a .xxx. IP holders should be happy, as
the gTLD disclaims any TM rights, and was subject to UDRP, etc. for 5
years as a quasi gTLD, before it ascended to a real one. There'd be no
need for multiple defensive registrations on hundreds of toy TLDs, just
actual UDRPs on ones that ascended to true gTLD status. Existing domain
holders can already create apple.store.com or ibm.store.com, but are
not subject to UDRP for those subdomains....during a 5-year period, TM
holders could police the quasi TLDs to the extent they desire, either
through UDRP or through statutory/common law methods. 

There are some parallels to .kids.us. If that subdomain was actually
successful (as measured by registrations and traffic), and they owned
the kids.com/net/org, one could see it promoted into .kids. Of course
as we all know, .kids.us has been a debacle. It is better that we know
it is a debacle now while it operates as a subdomain, BEFORE it ever
becomes a true gTLD, an unsuccessful one that pollutes the namespace.

I'd appreciate any feedback or input on improving the above proposal. I
think folks are polarized between either the small number of TLDs vs
large number of TLDs models, and I hope this represents a method of
allocating small numbers of PROVEN TLDs (proven through the "quasi TLD"
operational period) that respects existing domain holders and IP


George Kirikos

<<< Chronological Index >>>    <<< Thread Index >>>

Privacy Policy | Terms of Service | Cookies Policy