Spying and the Dilemma of False Positives

Glimpses into how US spying works, and intra-governmental wars, via Buzzfeed:

"Treasury Department's Office of Intelligence and Analysis has been illegally rifling through and filing away the private financial records of US citizens … Sources also claimed that OIA has opened a back door to officers from other intelligence agencies throughout the government, including the the CIA and the Defense Intelligence Agency. Officials from those agencies have been coming to work at OIA for short periods of time, sometimes for as little as a week, and thereby getting unrestricted access to information on US citizens that they otherwise could not collect without strict oversight."

I’ve wondered for a while how you get agencies to care more about false positives. They have strong incentives to not end up on the front-pages (“Incompetent Agency X should've prevented bad thing Y”). That is, if they see something that might be bad - a suspected terrorist wiring money, a family that might be abusing a kid, etc. - they have a strong incentive to flag it. Agencies really want to minimize the cases where they say “this isn’t a problem” when in fact there might be a problem. In other words, they care a lot about reducing false negatives. But they have much weaker incentives to care about false positives. Cases where they flag a potential problem, some action is taken, and it turned out that everything was fine. The costs are diffuse. I find this particularly worrisome with respect to privacy. If you posit a) tribal politics that impede the emergence of social consensus driving legislation to protect privacy, and b) that our future revolves around a data-driven service economy, the equilibrium is basically a panopticon.