80-95%? What about nearly every single espionage case in this country where the suspect was polygraphed repeated and passed with flying colors?
Yes, there are some limitations just like any technique under the sun has them. As long as you're aware of these limitations, it can still be useful.
This is a terrible way to think about a polygraph. The point being made is that there are methods (actual investigative techniques) that return far more accurate results than polygraphs do.
Is there a reason these methods cannot be used in conjunction with a polygraph?
Neither that article nor I are saying that polygraphs are a be-all and end-all way of doing investigation. But bearing in mind their limitations, they *can* be useful. From the number of unsolved crimes and wrongly convicted people, it's obvious that evidence and "actual investigative techniques" are not always conclusive and hardly perfect often due to the human factor. I don't see why discard something that in its particular domain (determining whether people are lying) is demonstrably better than humans. Those who can fool polygraphs can fool police detectives as well, so nothing was lost. And some will fail even though they're telling the truth. You can check them out more closely - nowhere did I say that polygraphs should be taken as a proof on their own.
For employee selection I agree that it's rather stupid since it may disqualify good candidates. Though this is also true of regular interviewing - it's quite common that a competent candidate will be dismissed based on a 15 minute phone conversation because of some trivia or because the recruiter didn't get a "good vibe". Many companies are simply more willing to miss out on a good candidate, than potentially hire a bad one. It sucks, but that's the way it is.