In Defense of Metadata Platforms
 

                           Joan Feigenbaum
                        AT&T Labs - Research
                       180 Park Ave, Rm C203
                       Florham Park, NJ 07932
                        jf@research.att.com
                   http://www.research.att.com/~jf
 
 
The interplay of societal values and information technology is broad,
deep, and bi-directional.  My professional expertise (cryptology,
information security, and theory of computation) is very narrow by
comparison, and so I will not attempt to take a comprehensive position
on how this interplay should proceed.  Instead, I will address the
specific subject of Internet metadata platforms and attempt to make
two points that have not, as far as I can tell, received sufficient
attention in the debate so far.  The following discussion assumes that
the reader is familiar with the ethical, technical, and commercial
challenges presented by Internet metadata platforms in general and by
the PICS (Platform for Internet Content Selection) proposal in
particular.  An introduction to the subject can be found in the essay
by L. Cranor in these proceedings.

In my roles as Computer Science researcher, Internet user, and US
citizen, I believe that well-designed metadata platforms are
essential.  The two points I would like to make in support of this
claim are: (1) Metadata such as ratings, labels, and recommendations
are an important part of the social structure in which freedom of
expression flourishes; (2) Internet users will demand metadata as a
condition of using the net for anything serious.  If we technology
professionals don't give them good metadata platforms, they will
either use bad ones or abandon the net.

(1) Metadata systems buttress our rights to free expression

The First Amendment right to free speech is a critically important
feature of life in the US.  Moreover, it is critically important that
Americans' First Amendment rights not be weakened for the sake of
comfort, convenience, or fashion: ``Pornography,'' ``hate speech,''
and other (ill-defined) categories of expression may be offensive to
some, but they are legal and should remain so.  In particular,
explosive growth in the Internet and other relatively new information
technology does *not* require that our right to free speech be
weakened or that the principles governing the enforcement of this
right be changed.  Time-honored rights and enforcement principles are
applicable in the current technological environment.

My support for the right to speak and, indeed, to offend is
unequivocal because I do not regard Free Speech as an abstract and
lofty ideal.  Rather, I see it as a very practical, working system
that adapts and evolves along with changes in society and technology.
An important component of this practicality and adaptability stems
from a principle that is not articulated as often or as clearly as The
Right to Speak, to wit: The Ability Not to Listen.  If I abhor
``pornography'' or ``hate speech'' or any type of expression you want
to engage in, I should not be allowed to censor you.  The proper
response to my disapproval is ``if you don't like what I'm saying,
don't listen.''

Why is this an adequate response?  The Ability Not to Listen is not
written explicitly into the Constitution, and indeed the more general
question of whether Americans have ``rights to privacy'' is a subject
of ongoing debate.  This is my reason for choosing the word
``Ability'' instead of ``Right.''  Nonethess, I believe there is a lot
of evidence that, as a practical matter, US citizens can reasonably
expect to be able not to listen if they choose not to.  Our society
has developed laws, regulations, social conventions, technology, and
economic incentives to facilitate people's ability to ignore words,
pictures, and other material whose form or content offends them.
Examples of these mitigating forces at work abound:

-- Metadata such as movie and book reviews and MPAA ratings.

-- Strict regulation of the content of traditional broadcast TV 
programming vs. more lenient or no regulation of cable programming.

-- Marketing and display techniques used by publishers, bookstores,
and newsstands that make it easy for customers to find what they're
looking for and to avoid material that they think would offend them.

-- Laws against sending unsolicited, obscene, printed material through
the US mail system.  Incentives that reinforce these laws, such as the cost
of printing and mailing material to people who will never be customers and
the social opprobrium heaped on mailers of unsolicited, obscene material
when they are discovered.

-- Court decisions that require anti-choice protestors to maintain a
certain distance from abortion clinics so that patients can enter and
leave without being harrassed.

The last item on this list strongly supports my claim that our society
recognizes the Ability Not to Listen as an important complement of the
Right to Speak.  Anti-choice protestors undeniably have the right to
voice their opposition to the practice of abortion, but they do not have
the right to force others to listen to their opposition.  More generally,
everyone has the right to free expression but does not necessarily have
the right to express himself everywhere, all the time, in all manners, 
and in all media -- the privacy and safety of potential listeners have to
be taken into account.

The Internet and other relatively new electronic forums are not yet
endowed with an analogous structure of regulations, metadata,
policies, social norms, and economic incentives.  But they can be if
we give creators, users, and regulators of technology time and
motivation to develop such a structure.  With the appropriate
structure in place, an ordinary citizen, i.e., not a technology
professional, should be able to express and enforce the following
policy about material of any type that he cares about: 

     I do not want to receive it via email or other push mechanisms; 
     I do not want WWW links and other recommendation mechanisms to 
          steer me to it automatically; 
     I want it marked explicitly so that I can find it manually when 
          I consciously choose to do so and can recognize it when I
          bump into it accidentally.

He should be able to express and enforce an entirely different policy 
about another type of material that he cares about: 

     I want this stuff pushed onto my screen as fast as possible, and 
          I am willing to pay extra to see that it is.

Definitions of ``Type of Material'' should be abundant and flexible.
I believe wholeheartedly that such a structure is achievable.  Alarmist
censorship efforts such as the Communications Decency Act and its
inevitable knock-offs should be resisted precisely for this reason.

Note that the Ability Not to Listen is not absolute and unqualified in
traditional, mature media any more than the Right to Speak.  For
example, patrons of movie theatres and other places of assembly are
basically forced to see the Exit signs, just as owners and managers of
such places are forced to display them.  Public safety concerns
justify these restrictions on speakers and listeners.  The Internet
and other new forums may ultimately require a similarly nuanced Right
to Speak and Ability Not to Listen, and thus the mechanisms and rules
that evolve will have to accommodate complexity and subtlety.

(2) Internet users will demand metadata systems

The PICS proposal and related efforts of the W3C (World Wide Web
Consortium) to create standardized, general-purpose metadata platforms
have met with widespread criticism, as explained by Cranor.  One
argument against such systems is particularly interesting and worthy
of serious consideration in any discussion of Ethics and Technology.
Some opponents claim that metadata platforms, which were proposed to
thwart censorship efforts such as the Communications Decency Act may
themselves inhibit free expression.  The argument given by the original
PICS proponents was similar to the one I've just put forth in
different words: The right to speak does not imply the obligation to
listen; if Internet users are technologically empowered to avoid
material they wish to avoid, there is no valid reason to banish any
material from the net.  The generality and flexibility of PICS was
offered as a starting point for the rich, multi-faceted structure of
Internet rules and incentives that I have just advocated -- one that
would serve the same function as its counterparts in the world of
traditional media.

It is precisely this potential for PICS or a similar metadata platform
to make the Internet more like traditional media that alarms PICS
opponents.  The argument goes like this, as I understand it: In the
areas of, e.g., movies or television, it is not government censorship
that limits the expression or dissemination of material -- it's the
market.  MPAA ratings, ``family hours'' on broadcast channels,
proposed V-chip ratings, and similar cues and conventions facilitate
the creation of mass markets and what is tantamount to ``official''
tastes, without the need for direct government intervention.  In
principle, a movie's creators could refuse to have it rated, because
MPAA ratings are not mandated by law; moreover, any group could, in
principle, develop an alternative rating system and publicize its
ratings of whatever films it sees.  In practice, however, the
movie-going public has a finite supply of time and energy to devote to
ratings, and the MPAA has a monopoly position in this market.  The
entrenched segmentation of the movie-going public into audiences
labeled G, PG, PG13, R, and X makes it extraordinarily difficult to
raise the money needed to create and distribute a film that won't be
labeled for one of these audience segments.  The standardized metadata
system has been so successful that it now drives the creation of the
primary data.

Critics rightly fear that a similar level of homogeneity and
mass-market mentality could ultimately take over the Internet.  In
principle, flexible platforms like PICS enable anyone to create a
rating system, anyone to be a rater, and any user to choose which
systems and which raters he or she trusts.  In practice, a few systems
and raters may rise to prominence and be shipped to users in software
bundles that make it difficult to use alternative ratings or to access
material to which prominent systems give bad ratings.  This would
severely damage the Internet's ability to achieve its potential as a
forum in which anyone can be a producer as well as a consumer.  Miriad
authors, artists, and animators could continue to produce, but only
those whose work was approved by the prominent raters would reach a
large audience.  Ultimately, all artists who wanted to reach a large
audience would find a way to achieve the desired ratings -- exactly
the situation that now prevails in US popular-culture industries.

While I acknowledge that the fear of mass-marketization is justified,
I do not think that technologists can responsibly abandon work on
metadata systems.  Like the critics, I appreciate the free-wheeling
atmosphere of the early years of the Internet, and I like the idea
that ``anyone can be a publisher.''  However, we are now in a state in
which anyone can publish anything *and* can distribute it to anyone at
any time and at basically no unit cost.  Junk email, unlike junk paper
mail, can be sent to tens of millions of people essentially as cheaply
and as quickly as it can be sent to thousands.  That means that it has
the potential to dominate many people's email boxes.  People will not
stand for that.  Their pre-Internet experience has led them to expect
the Ability Not to Listen.  If they are not given a user-friendly way
to sort their junk email from their valuable email, they will stop
using email.  The ``push'' nature of email makes the need for metadata
(which would permit flexible filtering) readily apparent, but the need
is there in ``pull'' media such as the web as well.  People will stop
using the web for business if they are unable to distinguish links to
experts from links to crackpots.  They will stop using it for
entertainment if too much of what they find offends, bores, or enrages
them.  The miriad authors, artists, and animators who could
potentially flourish in a medium with low barriers to entry will still
have a small audience, because frustrated potential audience members
will abandon the net.  That would be a shameful waste, and information
technology professionals should not let things get to that point.