The focus of this lucid and worthwhile paper about US intelligence efforts is not computing, as successful intelligence work in the future will require human spies and cooperation among agencies, as well as technology.
A recurring theme in the criticism of intelligence agencies is that the US has become too reliant on technical intelligence, rather than human intelligence. One task that may be amenable to a computing solution is winnowing information from the staggering amount of data collected, including that collected from wiretaps and from eavesdropping satellites and airplanes. The volume of data collected daily may approach that in all the books in the Library of Congress. Wiretapping optical fibers, especially those under the oceans, is much harder than wiretapping electrical wires, which radiate energy; the navy probably has special submarines to tap into trans-oceanic cables.
Decryption has its uses, but evidently not against al Qaeda, whose operatives do not encrypt, but use codes (as organized crime does). Packet switching poses another major problem to eavesdroppers; the packets making up a message may go through different channels, and the order of reception is not predictable. Financial crime is an important component of present day terrorism, and one that computing technology may combat. Intelligence against present day terrorism is much harder than it was during the Cold War.