Last week two revelations about US intelligence collection exploded in the press: a court order directing Verizon to turn over call logs from its telephone network was revealed along with details of the Prism program that gives the US intelligence community access to servers through which foreign communications are routed.

How would these collection capabilities actually help prevent attacks? Would they not simply swamp analysts? First, a law enforcement agency must have some hard knowledge – a name, a photograph, a location. Computers can then use these data to draw links among all the various lines out from the initial factual starting point: what numbers were called, at what location and of what duration.

Once networks emerge, computers can analyse them to determine which nodes are central to them (Mohamed Atta, the 9/11 leader, clearly emerges from such an exercise). Rather than overloading agents doing actual surveillance, this cross-hatching is done by a computer that eventually turns up the most promising leads. After the 9/11 atrocities, it was revealed that two of the terrorists were known to the FBI. Had their names been cross-referenced only with credit card accounts, frequent-flyer numbers and mobile phone numbers, US agencies might have picked up on the 17 other men linked to them flying on the same day at the same time.