Web browser to know how to interpret the
P3Pv1 link tag’s value. Although Netscape
and Internet Explorer can’t interpret the tag
in their current versions, it’s technically
simple to incorporate this feature into the
next version of both browsers.
This platform is a first step toward help-
ing us learn every Web site’s privacy policy
without requiring us to actually read its
statement, but it does not prevent malicious
Web site behavior, and it does not replace
seals of approval. Simply publishing a pol-
icy statement in machine-readable format
does not mean that a watchdog organiza-
tion reviewed the Web site, and it does not
specify what will happen if the Web site
changes its privacy policy or disobeys it.
Apart from embedding the seal of approval
as an entity tag into the machine-readable
policy document, another possibility would
be to store the policy information on the
watchdog organization’s Web site, as in the
following fictitious example:
<HTML>
...
<link rel=“P3Pv1” href=“http://
www.truste.org/P3P_seal?
company=www.shopping_
mall.com”>
...
</HTML>
Intelligent agents to the rescue?
If most Web sites agree to publish privacy
policies and seals of approvals in machine-
readable format, they pave the way for dis-
tributed intelligent agents that could act on
our behalf and protect our privacy. Privacy
agents might act as current antivirus software
updates act over the Internet. Personally, I’ve
always been concerned about artificial agents
who record every move I make when I browse
the Web. I prefer the analogy of a gate guard,
which stays still and sleeps most of the time,
but awakens at particular times under my
control. For example, my watchdog agent
might awaken only when it’s time to examine
the privacy statement of the Web site I’m
visiting. At that time, my agent could query a
collective of watchdog user agents across the
Internet to make sure that the Web site in
question is not guilty of infringing its privacy
statement. It might be technically and legally
challenging to implement such a distributed
system, but at least the burden will shift from
the user—who currently must do all the work
of watching what he or she does—to the
machine, which might be able to perform
routine verification tasks faster than a human.
The current burst of activity in the Fed-
eral Trade Commission and the US Congress
demonstrates that privacy is a growing con-
cern in the US. The two tools presented here
only scratch the surface of possible solutions
to this problem. In the meantime, I must ad-
mit that I am a news junkie, and I will hap-
pily trade my privacy for convenience so that
I can try out all the new personalized Web
news services in my never-ending quest for
the perfect personalized Web page.
MAY/JUNE 2000
13
Illustration by Sally Lee.
•
To convince the authorities, fuller evalua-
tions are necessary, especially regarding
the nontechnical difficulties (socioeco-
nomic, political, and so on) and social
benefits. In particular, evaluations must
consider effectiveness, safety, and eco-
nomics, and the consortium needs to
study the legal implications to determine
the responsibilities of the different play-
ers. LaRA also needs to consider potential
barriers to and side effects of implementa-
tion. The effort required for evaluation is
on a scale with the potential benefits.
So far, LaRA has studied Scenarios B
and C in detail and has proposed to French
authorities an action plan that distin-
guishes between research and industrial
projects. LaRA has also built Livic, a re-
search and testing facility near Versailles
that is now operational and has approxi-
mately 20 full-time employees (see www.
inrets.fr/ur/livic/index.html).
Jean-Marc Blosseville heads the Livic labora-
tory, a joint research laboratory between Inrets
and LCPC, two French research institutes deal-
ing with transportation, safety, and infrastruc-
tures. He is also a research director of Inrets,
where he works on methods for measurement,
control, and surveillance of road traffic. He is
the author of several patents in this field along
with several traffic control methods successfully
applied in France and several European coun-
tries. He has a master’s in mathematics and a
PhD in statistics from the University of Paris.
Contact him at Institut National de Recherche
sur les Transports et leur Sécurité, Av. du
Général Malleret-Joinville, 94114 Arcueil
cedex, France; jean-marc.blosseville@inrets.fr;
www.inrets.fr/ur/livic/blosseville.htm.
Michel Parent is the program director of Inria’s
R&D for La Route Automatisée, which he initi-
ated with Inrets while he was working in the
Praxitele project. He lead the development at
Inria of the CyCab vehicle, which includes drive-
by-wire and automatic control and is now manu-
factured by Robosoft, a start-up company. He has
an engineering degree from the French Aeronau-
tics School, and a master’s in operation research
and a PhD in computer science, both from Case
Western Reserve University. Contact him at Insti-
tut National de Recherche sur l’Informatique et
l’Automatique Domaine de Voluceau, B.P. 105,
78153 Le Chesnay cedex, France; michel.parent@
inria.fr; www.lara.prd.fr/Michel. html.
Continued from page 11
Coming in
July 2000
in
IEEE Intelligent Systems
Humanoid
Robotics
Guest edited by
Mark L. Swinson, DARPA
David Bruemmer, Strategic Analysis
