Tuesday, May 02, 2006

Bad security is worse than no security

One of the big arguments between US and UK intelligence is over the value of polygraph tests. In the US their use is ubiquitous, in the UK they are considered to be pseudoscientific quackery.

Psychology has long held the dubious distinction of being the scientific field most susceptible to psuedoscience. The Rorschach ink blot test was used for decades despite being a transparent piece of hokus pokus. Many of the crazier notions of Freud and Jung are still accepted by self described 'psychoanalysts'.

So when I can across the www.antipolygraph.org site which is run by a polygraph skeptic I started to think about the problem of testing the effectiveness of polygraphs in blind trials. If polygraph testing is scientific their effectiveness should be empirically verifiable. The National Academy of Sciences has found no evidence to support the claims made for polygraph test. The theoretical basis for the claims is weak, empirical studies do not support the high accuracy claims made.

All of which should be a serious cause for concern as the security services increasingly rely on polygraph tests despite ample evidence that they don't work. Aldrich Ames passed two lie detector tests while he was spying for the Soviets. Instead of throwing the junk out the CIA stepped up their polygraph testing.

No comments: