Just because he’s cranky and once had a tree fall on him doesn’t mean he’s wrong
It’s Friday afternoon, which means it’s the time the weekly What’s New updates from Professor Robert Park, of the University of Maryland and the American Physical Society (you can subscribe to his free and entertaining weekly newsletter at the previous link) go out. This week, he reports on the conclusions of the National Academy of Sciences report on the reliability of lie detectors. Remember Wen Ho Lee? His case led DOE to implement widespread lie detector testing. The problem is that lie detectors generate a lot of false positives, and are apparently fairly easy to thwart. As Park reports
DOE carefully reevaluated its policies [in light of the NAS report] and reissued them without change, arguing that a high rate of false positives must mean the threshold for detecting lies is very low. Therefore, the test must also nab a lot of true positives. Since that’s the goal, the DOE position seems to be that the polygraph tests are working fine and false positives are just unavoidable collateral damage.
But surely there must be a better way? Park has the answer:
WN therefore recommends replacing the polygraph with a coin toss. If a little collateral damage is not a problem, coins will catch fully half of all spies, a vast improvement over the polygraph, which has never caught even one. Moreover, coins are notoriously difficult to train, making them impervious to countermeasures.