Good story.  Interesting stat about 1,600 false positives and 2 false negatives trying to catch the 10 spies out of 10,000.  Where else would a 16 to 20 percent failure rate for a technological device be acceptable?

Also interesting that the FBI director of security states that 25 percent of applicants fail the polygraph.  If he admits to that number, I bet its higher -- maybe closer to the 50 percent quoted on this board.  Of course it would not make the FBI look good if they reported a number that high because no one would believe that 50 percent of applicants are lying, unworthy scum.

In any case, taking the conservative estimate of 16-20 percent error rates and 25 percent failure rate, the numbers shake down to something like this.  The article states that the FBI gives 8000 polys a year.  If half of those are new applicants (4000) then approximately 640 (4000 x .16) applicants are erroneously rejected.  That means that about 360 (4000 x .09 -- the remaing portion of the 25 percent failure rate) applicants are correctly rejected.  And perhaps 90 (based on the 20 percent false acceptance rate) "bad guys" slip through.

Again those are conservative numbers and it is entirely likely that by using countermeasures, an even higher number of bad guys can get through and an even higher number of good guys get rejected.  That means they are disporportionately hurting themselves with the rejections of people who could help bring about much-needed improvements, while accepting people who really shouldn't be there.

good article. unfortunatley not damning enough, no word on how CMs can beat the poly, and no mention of the push to ban polygraph testing in pre-employment screening.

FBIReject:

Actually the article is encouraging to the polygraph industry.  The opening paragraph states that the feds are using polygraph more than ever. According to the CIA McCarthy is no longer working for them because she was found to be deceptive duringa poly test. I question the validity of the so called test that included 10 spies and only 8 were caught but without the poly these 8 would never have been caught  so the poly was of great service to the intelligence community.

FBI-Reject wrote on May 1^st, 2006 at 8:25pm:


It should be noted that Eggen and Vedantam quoted the NAS report out of context. They failed to mention that in the example they cited, the NAS assumed, for the sake of argument, a 80% accuracy rate for polygraphy and set aside the issue of countermeasures in order to illustrate the base rate problem. The NAS did not conclude that polygraphy has a 80% accuracy rate and did not postulate that 8 out of 10 spies would actually be found deceptive by the polygraph.

retcopper wrote on May 1^st, 2006 at 6:49pm:


Yes 2 caught 8 free and you only have to fire 16% of your workforce to catch them.  

I postulate that if you fire 100% of your work force you will absolutely have zero spies.


Edit--
Here is what I figured out using Sixth Grade math:

Let me offer a little sense of proportion:

Assuming 50,000 CIA employees undergo a 5 year periodic screening (which Mahle’s polygraph probably was not).  You would have 10,000 polygraphs a year.  I don’t know the exact year the CIA started polygraphs but lets assume 1976 just to make the math easy.

You have a total of (10,000 polygraphs a year X 30 years) 300,000 polygraphs and the best you have to show for it is an employee who may or may not have leaked information to a reporter.  It will be interesting to see if any criminal prosecution follows.

Also please keep in mind most of the information in the original story was derived from open source flight logs.  The role of any person leaking information is far from clear.

In baseball you'd be batting 3.333 X 10E-6 (The E-6 means you put 6 Zeros in front of the 3)

" Actually the article is encouraging to the polygraph industry."


You notice how retcopper could care less about all the innocent people?  He seems to only care about his career.  this is the sense I get from the pro-polygraph people.  I find it a bit chilling that we have people in law enforcement who are completely driven by personal gain.