(For Non-Profit Educational Use)
Filtering Information
on the Internet
Look for the labels to decide if unknown
software and World Wide Web sites
are safe and interesting
by Paul Resnick, Scientific American, March 1997, pp. 106-108.
The Internet is often called a global village, suggesting a huge but
close-knit community that shares common values and experiences. The
metaphor is misleading. Many cultures coexist on the Internet and at times
clash. In its public spaces, people interact commercially and socially
with strangers as well as with acquaintances and friends. The city is a
more apt metaphor, with its suggestion of unlimited opportunities and
myriad dangers.
To steer clear of the most obviously offensive, dangerous or just boring
neighborhoods, users can employ some mechanical filtering techniques that
identify easily definable risks. One technique is to analyze the contents
of on-line material. Thus, virus-detection software searches for code
fragments that it knows are common in virus programs. Services such as
AltaVista and Lycos can either highlight or exclude World Wide Web
documents containing particular words. My colleagues and I have been at
work on another filtering technique based on electronic labels that can be
added to Web sites to describe digital works. These labels can convey
characteristics that require human judgment--whether the Web page is funny
or offensive--as well as information not readily apparent from the words
and graphics, such as the Web site's policies about the use or resale of
personal data.
The Massachusetts Institute of Technology's World Wide Web Consortium has
developed a set of technical standards called PICS (Platform for Internet
Content Selection) so that people can electronically distribute
descriptions of digital works in a simple, computer-readable form.
Computers can process these labels in the background, automatically
shielding users from undesirable material or directing their attention to
sites of particular interest. The original impetus for PICS was to allow
parents and teachers to screen materials they felt were inappropriate for
children using the Net. Rather than censoring what is distributed, as the
Communications Decency Act and other legislative initiatives have tried to
do, PICS enables users to control what they receive.
What's in a Label?
PICS labels can describe any aspect of a document or a Web site. The first
labels identified items that might run afoul of local indecency laws. For
example, the Recreational Software Advisory Council (RSAC) adapted its
computer-game rating system for the Internet. Each RSACi (the "i" stands
for "Internet") label has four numbers, indicating levels of violence,
nudity, sex and potentially offensive language. Another organization,
SafeSurf, has developed a vocabulary with nine separate scales. Labels can
reflect other concerns beyond indecency, however. A privacy vocabulary,
for example, could describe Web sites' information practices, such as what
personal information they collect and whether they resell it. Similarly,
an intellectual-property vocabulary could describe the conditions under
which an item could be viewed or reproduced [see "Trusted Systems," by
Mark Stefik]. And various Web-indexing organizations could develop labels
that indicate the subject categories or the reliability of information
from a site.
Labels could even help protect computers from exposure to viruses. It has
become increasingly popular to download small fragments of computer code,
bug fixes and even entire applications from Internet sites. People
generally trust that the software they download will not introduce a
virus; they could add a margin of safety by checking for labels that vouch
for the software's safety. The vocabulary for such labels might indicate
which virus checks have been run on the software or the level of
confidence in the code's safety.
In the physical world, labels can be attached to the things they describe,
or they can be distributed separately. For example, the new cars in an
automobile showroom display stickers describing features and prices, but
potential customers can also consult independent listings such as
consumer-interest magazines. Similarly, PICS labels can be attached or
detached. An information provider that wishes to offer descriptions of its
own materials can directly embed labels in Web documents or send them
along with items retrieved from the Web. Independent third parties can
describe materials as well. For instance, the Simon Wiesenthal Center,
which tracks the activities of neo-Nazi groups, could publish PICS labels
that identify Web pages containing neo-Nazi propaganda. These labels would
be stored on a separate server; not everyone who visits the neo-Nazi pages
would see the Wiesenthal Center labels, but those who were interested
could instruct their software to check automatically for the labels.
Software can be configured not merely to make its users aware of labels
but to act on them directly. Several Web software packages, including
CyberPatrol and Microsoft's Internet Explorer, already use the PICS
standard to control users' access to sites. Such software can make its
decisions based on any PICS-compatible vocabulary. A user who plugs in the
RSACi vocabulary can set the maximum acceptable levels of language,
nudity, sex and violence. A user who plugs in a software-safety vocabulary
can decide precisely which virus checks are required.
In addition to blocking unwanted materials, label processing can assist in
finding desirable materials. If a user expresses a preference for works of
high literary quality, a search engine might be able to suggest links to
items labeled that way. Or if the user prefers that personal data not be
collected or sold, a Web server can offer a version of its service that
does not depend on collecting personal information.
Establishing Trust
Not every label is trustworthy. The creator of a virus can easily
distribute a misleading label claiming that the software is safe. Checking
for labels merely converts the question of whether to trust a piece of
software to one of trusting the labels. One solution is to use
cryptographic techniques that can determine whether a document has been
changed since its label was created and to ensure that the label really is
the work of its purported author.
That solution, however, simply changes the question again, from one of
trusting a label to one of trusting the label's author. Alice may trust
Bill's labels if she has worked with him for years or if he runs a major
software company whose reputation is at stake. Or she might trust an
auditing organization of some kind to vouch for Bill.
Of course, some labels address matters of personal taste rather than
points of fact. Users may find themselves not trusting certain labels,
simply because they disagree with the opinions behind them. To get around
this problem, systems such as GroupLens and Firefly recommend books,
articles, videos or musical selections based on the ratings of like-minded
people. People rate items with which they are familiar, and the software
compares those ratings with opinions registered by other users. In making
recommendations, the software assigns the highest priority to items
approved by people who agreed with the user's evaluations of other
materials. People need not know who agreed with them; they can participate
anonymously, preserving the privacy of their evaluations and reading
habits.
Widespread reliance on labeling raises a number of social concerns. The
most obvious are the questions of who decides how to label sites and what
labels are acceptable. Ideally, anyone could label a site, and everyone
could establish individual filtering rules. But there is a concern that
authorities could assign labels to sites or dictate criteria for sites to
label themselves. In an example from a different medium, the television
industry, under pressure from the U.S. government, has begun to rate its
shows for age appropriateness.
Mandatory self-labeling need not lead to censorship, so long as
individuals can decide which labels to ignore. But people may not always
have this power. Improved individual control removes one rationale for
central control but does not prevent its imposition. Singapore and China,
for instance, are experimenting with national "firewalls"--combinations of
software and hardware that block their citizens' access to certain
newsgroups and Web sites.
Another concern is that even without central censorship, any widely
adopted vocabulary will encourage people to make lazy decisions that do
not reflect their values. Today many parents who may not agree with the
criteria used to assign movie ratings still forbid their children to see
movies rated PG-13 or R; it is too hard for them to weigh the merits of
each movie by themselves.
Labeling organizations must choose vocabularies carefully to match the
criteria that most people care about, but even so, no single vocabulary
can serve everyone's needs. Labels concerned only with rating the level of
sexual content at a site will be of no use to someone concerned about hate
speech. And no labeling system is a full substitute for a thorough and
thoughtful evaluation: movie reviews in a newspaper can be far more
enlightening than any set of predefined codes.
Perhaps most troubling is the suggestion that any labeling system, no
matter how well conceived and executed, will tend to stifle noncommercial
communication. Labeling requires human time and energy; many sites of
limited interest will probably go unlabeled. Because of safety concerns,
some people will block access to materials that are unlabeled or whose
labels are untrusted. For such people, the Internet will function more
like broadcasting, providing access only to sites with sufficient
mass-market appeal to merit the cost of labeling.
While lamentable, this problem is an inherent one that is not caused by
labeling. In any medium, people tend to avoid the unknown when there are
risks involved, and it is far easier to get information about material
that is of wide interest than about items that appeal to a small audience.
Although the Net nearly eliminates the technical barriers to communication
with strangers, it does not remove the social costs. Labels can reduce
those costs, by letting us control when we extend trust to potentially
boring or dangerous software or Web sites. The challenge will be to let
labels guide our exploration of the global city of the Internet and not
limit our travels.
---------------
Further Reading
Rating the Net. Jonathan Weinberg in Hastings Communications and
Entertainment Law Journal, Vol. 19; March 1997 (in press). Available on
the World Wide Web
Recommender Systems. Special section in Communications of the ACM, Vol.
40, No. 3; March 1997 (in press).
The Platform for Internet Content Selection home page is available on the
World Wide Web at http://www.w3.org/PICS/
The Author
PAUL RESNICK joined AT&T Labs Research in 1995 as the founding member
of the Public Policy Research group. He is also chairman of the PICS
working group of the World Wide Web Consortium. Resnick received his Ph.D.
in computer science in 1992 from the Massachusetts Institute of Technology
and was an assistant professor at the M.I.T. Sloan School of Management
before moving to AT&T.
