"Skilled operator" has nothing to do with it.
Again: The claimed underlying mechanism is based on bad science, dating back to the 1920s. It presumes that anxiety (as indicated via a few physiological measures) will increase a) noticeably and b) instantly when someone is lying, and clearly that is not the case.
So-called experts at detection deception are generally full of it. They typically do no better than chance, and the best can only catch 60% of liars. Polygraphs apparently don't do much better. (e.g.
https://www.psychologytoday.com/blog/spycatcher/201203/the-truth-about-lie-detection)
Therefore, a polygraph is effective only when you want to make your subject nervous while you grill them. It's ineffective if you are trying to detect lies or deception.
I also can't help but notice that you didn't link to a
single study to support your claim that polygraphs are in any way reliable. Hmmmm.