Ray,

First, I must admit that I agree with one of your thoughts - using your example, keeping five child molestors out of the FBI (or whatever applicable agency) is a very good thing.  However, we seem to also agree that a number of child molestors may actually move past the polygraph and perhaps even the background investigation phase.  That's bad.  One other point neglected - what about those 1 or 2 qualified candidates that might be unfairly labeled as child molestors because of this flawed process?  Is it really worth it?  Certainly the perfect tool does not exists.  But should we be using this one in the mean time?

However, my primary reason for responding to your post was your mention of an examinee that admitted to both rape and the attempted use of countermeasures.  You state "he had a hard time suppressing his relevant responses."  In this case, your subject was actually guilty.  Could you comment then on the use of countermeasures by those that do not need to lie on relevant questions but would rather reduce the chance of a false positive?  

I firmly believe that, by having knowledge of control question polygraph tests (meaning being able to effectively differentiate between control/relevant/irrelevant), an applicant is at a severe disadvantage in that responses to probable lie control questions will not be adequate simply because the "fight or flight" response allegedly measured by the machine will not necessarily be present.  In this case, should the examinee become "bothered" (in ways other than actually lying) by a relevant question and a response results, that examinee would most likely be considered to show deception for that relevant question.  Please comment and by all means correct any of my statements as needed.  Thanks.