Connecting the Dots
Can the tools of graph theory and social-network studies unravel the next big plot?
In the five years since that wrenching Tuesday morning when hijacked
aircraft sliced into the World Trade Center and the Pentagon,
Americans have been living with a new undercurrent of worry and
mistrust. Naturally, there's fear of further attacks. But there's
also concern that measures taken to forestall such attacks could
erode traditional rights and liberties. In recent months,
controversy has erupted over reports that government agencies are
monitoring Internet and telephone communications as well as
financial transactions. Some of the surveillance programs are said
to be sifting through gigantic data sets, scanning for patterns that
might reveal criminal intent or activity.
The debate over these programs has focused mainly on legal and
political questions. Are constitutional and statutory safeguards
being respected? What about laws that bar intelligence agencies from
spying on American citizens? Do the programs strike an appropriate
balance between the right to privacy and the need for security?
These are important issues, but I shall leave them to others. Here I
want to ask a different kind of question: What can one expect to
learn through such wholesale screening and data-mining operations?
Do the communications patterns of terrorists have a signature so
distinctive that computer algorithms can detect signs of a
conspiracy amid trillions of other telephone calls or e-mail messages?
In addressing these questions I face an obvious impediment: Very
little reliable information on the nature and scope of the
surveillance programs has been made public. However, mathematicians
and computer scientists have tackled problems very similar to those
confronting an intelligence analyst trying to make sense of
surveillance data. And social scientists have long taken an interest
in the networks that bind people together—including networks
of criminals and terrorists. Perhaps by combining insights from
these fields we can make some plausible guesses.