TITLE: "Study of Over 11,000 .INFO Sunrise Registrations
Analyzes Violations of Trademark Submission Rules"AUTHOR: Robert A. Connor, Ph.D.
SOURCE: (c) 2001 DomeBase.com. May be reprinted or quoted with proper attribution.
Contact: webmaster@DomeBase.com
RELEASE: August 27, 2001, 9:30pm CST
INTRODUCTION:
.INFO is a new extension for internet domain names, like .COM and .NET, that has
been authorized by ICANN and is being implemented by Afilias. For the registration
start up, a "Sunrise Period" was designed to let trademark (TM) holders register
their trademarked names before the general public gets a chance to register names.
This helps to prevent "cybersquatters" -- people who register names trademarked by
someone else and then try to sell them to the trademark holder.
According to Afilias'
Sunrise Period rules, someone applying for a Sunrise Period registration warrants
that four conditions are met: (i) at the time of registration of the domain name,
a current (non-expired) trademark or service mark registration was issued in the
applicant's name; (ii) the domain name is identical to the textual or word elements
of the trademark or service mark registration on which the registration of the domain
name is based; (iii) the trademark or service mark registration on which the registration
of the domain name is based is of national effect; and (iv) the trademark or service
mark registration on which the registration of the domain name was based was issued
prior to October 2, 2000.
The Sunrise Period will be followed by a "Land
Rush Period" which is open to the general public. During the Land Rush, there
is a randomized selection process to pick a winner when more than one person requests
the same name. Thousands of people have paid to pre-register in the Land Rush period.
After the Land Rush period, registrations will be on a first-come, first-serve basis.
SUNRISE
REGISTRATIONS THAT VIOLATE THE RULES:
In addition to legitimate trademark holders
for whom the Sunrise Period was intended, "sunrise squatters" have registered non-trademarked
.INFO names with fake trademark names and numbers, post-cutoff trademark dates, and
other violations of the Sunrise Period rules. Unless action is taken, names
claimed by bogus Sunrise registrations will not be available for the Land Rush period,
depriving people who paid to pre-register those names in the Land Rush period of
their chance and investment. Discussion of this issue can be found at http://forum.icann.org/newtldagmts.
Some
inconsistent Sunrise Period registrations may be due to computer processing errors
or misunderstanding, not bad intent. Some people have been surprised to see
their registrations in the Sunrise Period and have asked to have them canceled.
However, other Sunrise registrations may be intentionally fraudulent.
Sunrise
Period registrations can be challenged. Anyone can challenge a Sunrise Period
registration by paying $295 to WIPO ($75 net cost after refund if successful).
However, a successful challenger does not get a non-trademarked name even if the
original Sunrise registration is proven fraudulent. Thus, there is little incentive
for individuals to challenge sunrise squatters for non-trademarked names.
To help
address this situation, Afilias announced on August 15th that they will review and
challenge questionable Sunrise Period registrations after the challenge period is
over. However, such challenges would not be in time to release names
for the Land Rush period, so people who pre-registered in the Land Rush period would
still be cut out. This situation could be corrected by the "DomeBase Proposal"
outlined at http://www.DomeBase.com or the "Simple Solution" outlined at http://www.TheInternetChallenge.
PURPOSE OF THIS STUDY
This large-scale study of over 11,000 Sunrise registrations
is important for identifying the size of the sunrise squatting problem, classifying
the types of violations, and developing ways to correct the problem. This study
is based on .INFO WHOIS records that are publicly-available at the Afilias website.
Anyone can look up information on a registered .INFO name by going to http://www.Afilias.com
and using the WHOIS search. A list of all 11,000+ names used in this study
is available at DomeBase.com.
This study was conducted by Robert A. Connor, Ph.D.,
webmaster of DomeBase.com and Associate Professor at the Carlson School of Management
at the University of Minnesota. He alone is responsible for its content and
welcomes feedback on how to improve its accuracy at webmaster@DomeBase.com.
As requested by Afilias last week, the results of this study were delayed until
after the Sunrise Period so that potential sunrise squatters would not know what
types of fraud-detection algorithms are being used and submit registrations that
evade detection.
It is important to note that not all Sunrise registrations whose
WHOIS records technically violate the Sunrise rules are really bad. Some WHOIS records
for Sunrise registrations that are technically in violation of the Sunrise rules
may be due to computer glitches or incorrect data entry or may be consistent with
the spirit of the Sunrise rules. For this reason, this study employed limited
manual review as well as automated rule compliance algorithms.
On the other hand,
not all Sunrise registrations that escape detection using compliance algorithms are
valid trademark registrations. Some WHOIS records for Sunrise registrations that
escape detection using simple compliance algorithms may have apparently valid, but
actually fraudulent, trademark information.
For these reasons, this study
does not produce a definitive list of bad registrations. It does, however,
provide useful information to start large-scale investigation, challenge and removal
of inappropriate Sunrise registrations. Thorough evaluation of the validity
of thousands of Sunrise registrations will be will be a huge job.
This study quantifies
the frequencies of different types of inconsistency with Afilias' Sunrise rules.
Several of the WHOIS fields had missing or poor quality data, probably due to lack
of edit checks at the time of data entry. This required more complicated algorithms
incorporating data checks and made the analysis more difficult.
RESULTS OF THIS
STUDY
The following are the types and frequencies of inconsistencies and probable
inconsistencies with Sunrise Period rule based on information in the WHOIS records.
Since one record can have more than one type of inconsistency, the percentages given
for specific types of inconsistency overlap. However, the overall percentages
given at the end exclude overlap.
BAD TM NAME: Approximately 6% of the Sunrise
registrations had a missing or bad trademark name -- such as a blank, "?", "n/a",
"none", "US", or the registrant's personal non-trademarked name. An additional
4% had "trademark example" listed as trademark which, even if a computer glitch,
provides no genuine trademark name information to be checked. One interesting
example is the registration for VIENNA.INFO which says "no trademark!" in the trademark
field.
NON-IDENTICAL TM: Approximately 12% had a domain name that was not identical
to the trademark name as required by the Sunrise rules -- even after generous adjustments
for variations in hyphens, spaces, other non-alpha characters, various company designation
suffixes in different languages, and other plausible variations. For example,
the trademark "dumping" was used to register CAR.INFO, MUSIC.INFO and CALL.INFO.
The trademark name, number, and date used register the site for trademark law --
"TMLAW.INFO" -- are "n/a", "n/a", and 2001-01-01.
BAD TM NUMBER: Approximately
11% of the registrations had a missing/bad trademark number. These cases included
missing value, "unknown", "n/a", "common law", "applied", "US", "0", "123456789",
"99999999", or "e.g. 12345". For example, the trademark number given for "BIBLE.INFO"
is "1". While this may be symbolically appropriate, it does not match information
at the US Trademark and Patent Office. As of the release of this article, the
author was not able to confirm the validity of the trademark numbers "12345-14" for
"BUSH.INFO" and would like to suggest a recount.
DUPLICATE REGISTRATION FOR SAME
TM: 21% had more than one domain name for the same trademark number. If domain
name A is identical to trademark name B, and domain name C is different than domain
name A, then how can domain name C be identical to trademark name B? Registering
domain names that are subsets of a trademark is not consistent with the Sunrise rules.
For example, if someone has a trademark on "The Best Place in the World to Find a
Home or get a Job", one should not be able to register domain names for BEST.INFO,
PLACE.INFO, WORLD.INFO, FIND.INFO, HOME.INFO, GET.INFO, and JOB.INFO. Some
registrants registered scores of domain names with the same trademark number.
CHILD.INFO and NUDE.INFO were both registered with the same trademark number, as
were JEWISH.INFO and KKK.INFO.
DATE AFTER CUT-OFF: 7% had a trademark date after
October 2, 2000, including many for the year 2040, 39 years in the future.
Another 16% had a trademark date of October 1, 2000. This was a Sunday and
probably not a valid trademark start date. Another 2% and 3% had identical trademark
dates, including trademark dates for computer-related domain names dating back to
1899. Even if some of these science fiction dates are due to computer glitches, they
remain serious problems because lack of valid data confounds checks on trademark
validity.
***.INFO TRADEMARKS: 3% used a trademark for "***info" to register "***.INFO"
where "***" is a generic word. This games the trademark system to lock in generic
.INFO domain names before anyone else has a shot at registering them. If I
anticipate that ICANN will be coming out with a new extension .SHOP, is it proper
for me to lock up the entire domain space years in advance by going to some non-english
country and trademarking all the words in the english dictionary followed by the
four letters "shop"? One other observation about suffixes. Some people
used "***.com" (without a trademark for "***") as the trademark to register "***.info."
If this is allowed, then new top level suffixes do not expand access to internet
names as intended, but rather just duplicate the roster of .COM registrants.
Registrations with inconsistencies identified by the above criteria were compiled
and then received limited manual review for reasonableness. This manual review resulted
in 26% being judged plausible and removed from the inconsistent list.
We first
used conservative criteria to identify Sunrise rule violations, counting only records
which had one or more of the following problems and which had not been removed as
plausible through manual review: BAD TM NAME (excluding "Trademark Example)"; NON-IDENTICAL
TM (after generous adjustments); BAD TM NUMBER; DUPLICATE REGISTRATION FOR SAME TM;
or DATE AFTER CUT-OFF. This resulted in 16% of Sunrise Registrations.
We
then used more liberal criteria to identify Sunrise rule violations, counting all
of the records in the conservative approach, plus those which had one or more of
the following problems and which had not been removed as plausible through manual
review: TRADEMARK of "Trademark Example"; DATE OF 2000-10-01, 1995-05-06, OR 1899-12-30;
and ***.INFO TRADEMARKS. This resulted in 23% of Sunrise Registrations.
These
percentages are lower than those reported anecdotally for small samples of the most
valuable names. Some Land Rush participants have reported that almost all of
the names that they pre-registered have been taken by Sunrise squatters. This
is because the more popular names are more likely to be targeted by sunrise squatters.
Overall,
we estimate that the percentage of Sunrise Registrations that violate the trademark
submission rules is between 15% and 25%. With tens of thousands of Sunrise registrations
in total, adequate review through WIPO challenges could require thousands of challenges
at a cost of millions of dollars. The cost of WIPO challenges can be subsidized by
Afilias using registration fees received from Sunrise squatters.
CONCLUSION:
Sunrise
squatters are a serious problem, probably affecting between 15%-25% of Sunrise registrations.
WIPO challenge fees to adequately address them might be in the millions of dollars.
Algorithms to detect fraud are a useful first step, but are not sufficient.
Sorting through thousands of registrations to determine will be very labor intensive,
especially with the incomplete and inconsistent registration data. Unless these
problems can be thoroughly corrected, use of Sunrise Periods in future top level
domain names may be in doubt.