Testimony and Statement for the Record of
Dr. Jason Catlett
President and CEO, Junkbusters Corp.
P.O. Box 7034, Green Brook NJ 08812; (908) 753 7861
Visiting Fellow, Kennedy School of Government,
Harvard University (2001-2002)
on
The Whois Database: Privacy and Intellectual Property Issues
before the
Subcommittee on Courts, the Internet, and Intellectual Property
of the Committee on the Judiciary
House of Representatives
July 12, 2001
Hyperlinked version available at: http://www.junkbusters.com/testimony.html#whois
My name is Jason Catlett, and I am President and CEO of Junkbusters Corp.
I'm grateful for this opportunity to speak here today.
Junkbusters is a for-profit company whose mission is to free people
from unwanted commercial solicitations through media such as email,
physical mail, telephone, and faxes. (The Whois database is a major
source for contact information for all these media.) Since our web
site launched in 1996, millions of people have turned to us as a free
source for information, services and software for stopping junk messages,
particularly email. I have assisted many government organizations and
legislators on email and other privacy issues since the Federal Trade
Commission asked me to explain the mechanics of spamming at their public
workshop on the topic in 1997.
I commend the committee for holding this much-needed oversight hearing
on the Privacy and Intellectual Property Issues of the Whois Database.
I have little to contribute on the topic of intellectual property,
other than to say that it is in a sense somewhat irrelevant to the
privacy interests of an individual whether an organization owns a item of
personal information about a "data subject" (as privacy lawyers call the
individual concerned), versus whether the organization buys, licenses,
barters, scavenges, or steals the data from another party. These are
essentially commercial considerations. The key privacy questions are
whether the data subject consented to the collection, disclosure and
use of the data, whether the organization handles the data fairly and
lawfully, and what rights of redress the data subject has if it does not.
Privacy
Definitions of privacy generally fall into one of two types, both of which
are acutely relevant here. The first is "seclusion from intrusion,"
or the "right to be let alone," to use the phrase made famous in the
1890 law journal article by Brandeis. The second is "informational
self-determination," the right to control the collection, disclosure
and use of information about oneself, formulated by Alan Westin in
his 1967 book "Privacy and Freedom" and now the basis of most modern
privacy statutes worldwide. To take obvious examples in context of the
Whois database, the first definition addresses whether an individual
registering a domain receives spam or unwanted solicitations via other
media, and the second includes whether information is gathered or sold
by other parties about the registrant without her knowledge and consent.
Violations of these two types of privacy tend to be correlated, since
the gathering of contact information is a means towards the delivery
of an unwanted solicitation, and because the targeting of messages
based on further information makes the activity more economically
attractive. As an illustration, the San Francisco Chronicle reported
in 1997 that Barnes and Noble, an online bookseller, had established
software systems to search people's home pages for references to certain
authors, and emailed them solicitations to purchase new titles in the
genres mentioned. Independent of the fact that the company should have
known better than to try spamming (and soon discontinued the practice),
many people were disturbed by the idea that a profile of their reading
tastes was being assembled in this robotic manner by an unknown party,
let alone being confronted with personalized recommendations based
on them. Even fans of book catalogs might be unsettled by a physical
letter beginning "Dear Murder Enthusiast" or detailing some interest that
they intended to share only with a few friends. Given that the compilers
of marketing lists have for years used Whois registration information
as a source of personal information (in some cases scavenged free,
in others bought from registrars), concerns over the data privacy are
well justified. Most people avoid putting their home address on their
web sites, and they should be able to register a domain name without
effectively giving up this precaution.
The public policy objective of privacy law is to preserve the individual's
right to privacy, while still permitting societal participation.
This is somewhat analogous to intellectual property law, which seeks
to encourage the publication of products of the intellect by providing
certain rights to inventors and authors to control the subsequent
distribution and use of their work. The current situation with the
Whois database is unsatisfactory because individuals are effectively
required to sacrifice some of their privacy in order to participate in
a fundamental Internet activity. Courts have remarked that the Internet
has provided an unprecedented opportunity for free speech; participation
should not be dampened by avoidable erosions of privacy.
The current (1999) ICANN Registrar Accreditation Agreement does contain
some provisions relating to privacy, but they are inadequate in both
theory and practice. [See http://www.icann.org/nsi/icann-raa-04nov99.htm
at J.7.a and F.6.f] The agreement anticipates the possibility of a
registrant licensing a domain to another party whose contact details
are not disclosed, but this is not a satisfactory way of preventing
disclosure for the average user. The agreement also requires the
registrar to impose an undertaking not to use the email addresses from the
Whois database for sending Unsolicited Commercial Email (UCE, or spam),
but in practice this is ineffective. Spam is discussed further below,
and my statement here concludes with a set of specific recommendations
for ICANN. Mine is not the only privacy organization to seek such
reforms; see for example the Electronic Privacy Information Center's
letter of February 16 to Congressional Privacy Caucus on this topic.
[ http://www.epic.org/privacy/internet/ICANN_privacy.html ]
The requirement of the publication of registration information can
be seen as egregious and anomalous when compared to analogous media.
Telephone subscribers are universally given the option of a non-published
(unlisted) number, regardless of which local phone company they use.
The US Postal Service discloses information about the identity of a post
office box holder only if the holder solicits funds from the public.
Various statutory privacy rights have been established to protect
the nexus of contact in different media, such as the prohibition in
California against telemarketing calls to non-published numbers, so-called
"asterisk laws" in several states mandating an optional designation
in directories for published numbers that must not be telemarketed,
the federal prohibition against junk faxes, and the opportunity to issue
prohibitory orders against senders of unwanted solicitations via US mail.
This procedure was upheld by the Supreme Court in 1971, including its
restriction on the subsequent sale of the address in marketing lists.
My first recommendation below is an addition to the Whois database to
support this kind of protection for email addresses.
Given the lack of such protections in the online world, plus the ease
with which contact information may be inexpensively gathered, it is hardly
surprising that surveys routinely find privacy is the number one concern
of Internet users and a major reason for non-participation by the offline
half of the population. The basic operation of establishing a homestead
in cyberspace should not stand as an example of the lack of respect for
privacy in the architecture of the Internet, particularly when a few
appropriate curtains could be added with comparatively little effort.
To be fair to the original architects, many of their procedures were
devised at a time when the individuals involved were few and often known
personally to one another, so it is understandable that privacy does
not appear to have been a top design priority. Changes are now overdue.
Accountability
Privacy is a fundamental human right, but it is not an absolute right: it
should not provide impervious and permanent cover for criminal activity,
for example. Appropriate mechanisms should be in place for personally
identifying disclosures in the case of law enforcement investigations,
and for civil litigation such as libel, trademark and copyright
disputes. But these mechanisms should restrict disclosures to what is
necessary and fair; checks and balances should protect against misuse.
Making contact information available to everyone is as much an overkill
as if a DMV were to require people to display their drivers licenses on
their lapels when standing on the sidewalk.
Domain names do somewhat differ from other media in that they enable
the registrant to establish an identity that can be used in the role
of a publisher as well as a subscriber to a multi-way communications
channel (though fax broadcasting has a similar quality). But the actual
publication is typically performed by an Internet Service Provider,
or at least via an ISP, and ISPs do not generally require the public
disclosure of contact information for the source. Why should registrars
be any different? ISPs are accustomed to tearing down web pages or
providing subscriber information when required to do so by a court
order. The same procedures can apply to domain name registrations if
this additional step is needed.
Spamming
The problem of spamming is one of the most important and instructive
topics for analysis here. Spamming is not a criminal offense in most
states, but it is socially damaging, undermines consumer confidence in
the Internet, imposes on consumers and businesses billions of dollars
in wasted costs annually, and violates the terms of service of ISPs.
As I have said in testimony before the Senate, I believe spamming should
be prohibited by federal law, and perhaps it will be. But even if it is,
people should still be able to try to avoid spam by reducing the exposure
of their email addresses, and those who are harassed by spammers should
have the means to obtain redress, which in practical terms translates
into identifying the spammer.
The most obvious damage to privacy from the Whois database is due to
the so-called "harvesting" of email contact addresses by spammers.
(I prefer the term "scavenging" because the crop being reaped was not
planted by the scavenger.) As mentioned above, the ICANN agreement with
registrars requires the registrar to impose an undertaking not to use
the data obtained to facilitate spamming. Unfortunately spammers can
blithely ignore the "you agree not to" message attached to the responses
to their requests, because their access is essentially anonymous. Limits
are often placed on the rate at which domain name queries are answered
from any given IP address, but this merely reduces the speed with
which the addresses are obtained, and is ineffective in the long term.
It cannot prevent scavenging any more than a supermarket could prevent
shoplifting by limiting the numbers of bags shoppers are allowed to
carry out of the store.
The observation has often been made that Whois contact information can
help track down spammers, and I certainly agree that this is sometimes
the case. Unfortunately it is rarely much help against career spammers,
who have registered large numbers of domains with contact addresses
such as the Martian embassy and phone numbers such as 202-555-1212.
Beyond these patently false addresses lie more plausible but incorrect
entries. Experienced spam hunters tend not to rely on such self-reported,
unauthenticated and too-often inaccurate information; rather they examine
the header information on the email and use software utilities such as
"traceroute" to establish the ISP that originally carried the spam,
and then ask the ISP to terminate the account. The casual spammer will
usually desist after a warning from his ISP. Furthermore, almost all
spammers give other generally more reliable clues to their identity in
the content of their emails, which are seldom abstract messages such
as "Sin no more." They often ask the addressee to visit a particular
web site, which can be tracked via traceroute and the hosting ISP,
or in the case of a site accepting credit card payment, through the
banking system. Many spams ask directly for checks to be sent to a
post office box specified in the email, which can also be followed.
In practice, self-reported contact information is like a weak door
lock that keeps out the honest unintentional intruder while presenting
no serious challenge to the dedicated burglar. I do not believe the
benefits of tracking amateur spammers via the self-reported contact
details from the Whois database outweigh the damage to privacy caused
by the public availability of the information.
Reducing personally identifiable information
Various other benefits of contact details being public have been cited,
but none of them persuades me that administrative contact must be made
public. Technical contact information is certainly useful for maintenance
tasks, but most technical contacts are business-title roles at ISPs,
not individual registrants. The fact that consumers find it useful to
authenticate a business using the administrative contact information
from the Whois database is no reason to require it of all registrants,
any more than residential phone subscribers should be forced to have
yellow pages entries. Businesses that consider it beneficial can elect
to do so, as proposed in my second recommendation below.
ICANN states in the preamble to its June 2001 survey that
more than 70% of its registrations are by organizations.
[See http://www.icann.org/dnso/whois-survey-en-10jun01.htm under
Background] The remaining twenty-something percent still adds up to
a very large number of individuals whose privacy is being compromised
by their registrations. A policy question arises whether organizations
should be treated differently to individuals. Only natural persons have
privacy rights; entities such as corporations do not, though they may
have an interest in confidentiality: considerable public speculation has
arisen from domain names registered by large companies such as Amazon
and Microsoft. In the case of sole proprietors, the entity may appear
to be an institution when it is in many ways more like an individual.
For these reasons it seems to me appropriate to give institutional
registrations exactly the same control over admin and billing contact
information as individuals have for personal registrations.
I further believe that it may be desirable and feasible for domain names
to be registered with a pseudonym (such as a registrar-issued customer
number), so that no personally identifiable information is provided,
not even to the registrar to whom payment was made (presumably with
a money order). Anonymity and pseudonymity are the most reliable
ways to protect privacy: there is no possibility of personal data
being disclosed or used inappropriately, because it does not exist.
(The difference between anonymous and pseudonymous speech is that while
neither is identified as originating from a specific individual, the
pseudonym allows continuity of interaction and attribution.)
If participation in the digital network without identification raises
concerns in your minds about accountability, consider how routinely
this occurs on the telephone network: with a payphone, using a popular
privacy-enhancing technology called coins. Doubtless some crimes are
facilitated by this opportunity, but nobody would consider this as a
justification for retrofitting the nation's payphones with credit card
readers or for abolishing the quarter. In some countries, including
Italy, it is even possible to subscribe to a prepaid mobile telephone
service without identifying oneself to either the carrier or the
government. If the phone appears to be involved in criminal activity,
law enforcement can have the service suspended or obtain the identity
of subscriber by examining the numbers called or by wiretapping calls.
The situation for pseudonymous domain names would be analogous.
Notice that the registration itself is unlikely to be considered criminal:
even if the text of the domain name were arguably libelous or blasphemous,
is there any prospect of real harm merely from its presence in the
Whois database? Registrars have already addressed the question of
obscene domain names, and can decline to register them if they consider
them offensive. Even in the case of trademarks, it is far from clear
clear that merely registering FamousNameSucks.org without publishing
a corresponding web site would constitute infringement. Rather, it
is activities other than registration that constitute the wrongdoing,
and those activities entail their own means of tracing the malefactor:
the Whois database cannot reasonably be expected to serve that purpose,
any more than the white pages should be expected to deter harassing
phone calls.
Where it is found appropriate to revoke a domain name, it is obviously
just as easy to terminate domain name service for a pseudonymous
account as it is for one registered to Thomas Paine or the Federalist
Publishing Company. The Famous Name Corporation can still sue a John
Doe defendant, seek his identity from an ISP, and persuade a court to
have the registration transferred to it.
If a Unabomber wishes to publish his manifesto anonymously, he is
likely to find other options preferable to registering the domain
ExplodeTechnologists.org. Even if he did wish to establish such a
web site, he would be more likely to give his administrative contact
address as Mauritius rather than Montana. The FBI would be no more
hampered by pseudonymous registration than the false details in this
registration; its agents would probably sooner seek the assistance
of the ISP hosting the domain rather than sending field agents to the
Indian Ocean. Some spammers favor disposable return email addresses,
which pseudonymous registrations could provide, but they are already have
that by claiming to be from the Martian embassy, or less flagrant false
addresses. Also, free web-based email services have a cost advantage
to the spammer over domain name registration. In short, pseudonymous
registration of domain names seems unlikely to lower the practical level
of accountability for objectionable behavior, because such behavior can
more reliably and appropriately traced by other means.
Pseudonymous registration does raise some logistic questions, such as
how renewal notices are to be sent (perhaps by anonymous remailers),
but I believe that deliberation would likely find practicable solutions,
so I suggest that ICANN investigate the question.
This is one of the following several specific recommendations I
respectfully submit to ICANN and the committee to improve the privacy
of registrants and Internet users.
Recommendations
1) UCE field: The addition to the registration database of a field
indicating the registrant's disposition towards Unsolicited Commercial
Email from any party to email addresses within the domain (not merely
the one provided as part of the registration). At least three possible
registrant responses should be supported: unwilling, willing, and not
indicated.
This measure has similarities to the "do-not-call" lists and "asterisk
laws" that several states have passed against telemarketers. The UCE
field may be usable under existing state anti-spam legislation such as
California's, and possibly by future federal and state legislation.
2) Disclosure election: Registrants should be given the opportunity
to indicate their disposition toward disclosure by their registrar of
billing and admin contact information. At least three possible registrant
responses should be supported: unwilling, desired, and not indicated.
I believe ICANN should require this of registrars. This option should
apply not merely to email address, but to all contact data. Domain name
registrants receive a great deal of junk physical mail as a result of
registering (some due to their registrar actively selling the contact
details as a mailing list). Registrants should not have to be burdened
with this.
In the case of Registrars who wish to sell for marketing purposes contact
information about their registrants (versus distributing it via the
Whois database), separate affirmative consent should be required (opt-in).
3) Population of fields: A program to encourage or require registrars
to seek and process customers' elections for the above two fields (UCE
and disclosure).
Registrants need not be immediately pestered for a response, but
the process should be easily available via the registrar's web site,
and the question should be posed prominently at the time of renewal.
Consideration should be given to whether the registrant's response ought
to be made public as part of the Whois database; this transparency may
be beneficial in seeing whether registrars are withholding or providing
data about registrants who have made no election.
4) Plaintiff's procedures: The development of standard procedures for the
processing by registrars of requests for the on-forwarding of messages
to, or the disclosure of contact information about, registrants who have
elected against disclosure of their contact information.
A typical question here is what should happen when a trademark owner
wishes to send a cease-and-desist notice to the operator of a web site.
The procedure should not impose undue burdens or liability on registrars.
5) Development of appropriate legal mechanisms to support the three
points above.
Privacy rights require an enforcement mechanism with a sound legal basis.
For example, if a registrar discloses a registrant's personal data
contrary to her instructions, what procedures does she have for redress?
6) Pseudonymous registration: The development of appropriate mechanisms
to support pseudonymous registrations.
I believe that the steps I recommend above would greatly improve
the privacy of Internet participants without significant deleterious
side-effects.
I appreciate the opportunity to speak with you today. I would be pleased
to answer your questions.