Hi,
I always think of Gary Ridgway, the green river killer, when this subject comes up.
He came under suspicion much earlier in the police investigation, took a polygraph and passed it. The police moved on to another, innocent man who failed a polygraph and Ridgway went on to murder dozens more women.
I read somewhere that psychopaths are supposedly more mentally adept at fooling the polygraph, and for certain people it has about as much use as astrology or a crystal ball in determining someones guilt.
I like to look at it like hypnosis. Some people are susceptible to it while others are not. Since some are not it can have very dangerous consequences if too much stock is put in the results. The results can't be used to convict someone because of the unreliability. As you pointed out there are people who flunk simply because they are nervous (not unlike people whose blood pressure rises substantially while being examined by a doctor) while there are guilty people who pass. A trial verdict is much too serious to let ride on a crap shoot.
Similarly it is a gamble when investigators use it to clear someone and concentrate their efforts elsewhere. Fortunaely the example of the Green River killer has not been reapeted too often as far as we know.
The UK is not using it for investigative purposes in the traditional sense though. It is more like military psy ops.
They try to instill fear into the released criminals. They want the parolees to be scared of committing crimes out of fear that the lie detector will catch them lying because they know they will definitely be subject to lie detector tests during all parole interviews. They also want the parolees to be scared to lie during said interviews and those who are so scared are much more forthcoming with information.
So maybe it does in a way inhibit crime through psy ops and also results in those who did commit crimes again to admit they did it so they can be punished.
Hard core criminals though are not scared of lie detector tests, are not dettered by the fact they have to take them and will probably pass the tests or realize even if they don't pass nothing will happen to them. So it likely offers nothing at all as far as the worst offenders are concerned. They are the ones I would be worried about most of all. But is doesn't hurt deterring more minor crimes so I guess it is worthwhile if it actually works as they claim.
I would never hang my hat on the results, particulary in an investigative context yet there are countless examples of police in the US doing just that and it sometimes does backfire with diablicial criminals are erroneously cleared. The odd thing is that there are still cops out there who have not learned their lesson and still swear by results anyway. But there are cops so desperate they turn to psychics though psychics never helped break any major cases.
Cops who are stumped seem to be as superstitious as sport stars. Maybe they even have lucky socks they wear to help the solve crimes who knows...