For five years I’ve tried to come up with a brief, economical explication of the epistemological dangers of the Doug Feith School of Intelligence. In the course of his brilliant book Angler, Barton Gellman does what I couldn’t. He starts by describing a massive, sprawling chart of connections drawn by David Wurmser, a future Cheney aide who, as part of Feith’s Policy Counterterrorism Evaluation Group, attempted to diagram the terrorist threat:

The spider chart was meant "to create a strategic picture, and that strategic picture is the foundation of policy change," Wurmser said. "It helped you visualize, because if you saw, say, twenty relationships between X and Y, and twenty between Y and Z, then there’s at least a suspicion that Z and X are interacting through Y." A map like that could bring insight, but there were perils in surmising too much. Suppose X and Y were Dick Cheney and Colin Powell. Twice they served in senior posts under presidents named Bush. In the early 1990s, they worked at the same address and were spotted together on international flights. They communicated frequently, encrypting their secrets. Back then, Cheney hired Powell for a very big job, elevating him over people far more senior. Ten years later, with Cheney in charge of recruiting, Powell got an even bigger job. From all this a person might figure that Powell was an agent of influence for Cheney. Lawrence Wilkerson, a colonel in Cheney’s Pentagon who became Powell’s State Department chief of staff, might be agent Z on Wurmser’s chart, taking Cheney’s instructions through a cutout. As it happened, though, Powell was Cheney’s bitter antagonist in the second Bush administration, and Wilkerson called Cheney every angry name from "ruthless," "dangerous," and "arrogant" to the "leader of a cabal. This business of links and relationships could get tricky.