Security was tight as NPR's David Kestenbaum entered the inner sanctum. Six staffers hovered around him.

It wasn't the CIA. Or a secret air base. Or Fort Knox. Not even the Playboy mansion. It was the closely-guarded Medicare databank in Baltimore.

The databank is enormous, petabytes of data (a petabyte is a 1 with 15 zeros after it, Kestenbaum explains). The rows and rows of shelves, a half-million cassettes of computer data, so vast that a robot is required to navigate it. The library holds medical records of decades of Medicare patients. And their doctors. It could potentially tell us much about quality and performance and efficiency of just about every physician in the United States.

Except it's a secret. Some researchers and consumer advocacy groups have fought to open it up (not the patient records, but the doctor data). And doctors have fought to keep it closed.

Arnold Milstein, a physician and researcher who has advised the White House on health care economics, wants it open. Doctors don't even know how they stack up against their peers, against standards. The database could tell them.

"Sitting there like a diamond on the desert floor was the Medicare database, essentially the nation's biggest source of information on whether or not doctors are or are not adhering to what are called scientific clinical guidelines," he says. "It's the only source of information on the total amount of health insurance fuel that they burn on solving a given clinical problem.

But doctors want it kept confidential. They say raw claims data doesn't really depict all the complexities of a specific patient, that this is not the right benchmark for measuring the excellence (or mediocrity) of one doctor who may be part of a large and complex care team.

This being the U.S.A. in the early 21st century, the controversy went to court. Consumers' Checkbook sought the information about individual physicians under the Freedom of Information Act. The government (backed by the AMA) resisted. Courts ruled this way, then that.  It may eventually end up in the Supreme Court.

We wonder if there's a middle ground somewhere.This specific issue is not one that we at New America have really delved into in all its complexity. However some of the health policy research that has been done here (on diabetes care at Baylor, physician performance in Grand Junction) suggests that the right data about physician performance can compel physicians to perform better. That's good for the doctors and for the patients -- and for the system. We can't help but think that there must be some way of harnessing this massive trove of secret data in a way that can push us -- and our doctors -- toward quality and efficiency.