Perhaps I'm overlooking something obvious, but it seems to me that a polygrapher would detect a "lie" by comparing a strong physiological reaction on one question to a series of weak reactions on the rest of the questions.  For example, if the following five questions were asked:

#1 - Did you lie in describing your prior employment?
#2 - Did you lie in describing your prior drug use?
#3 - Did you lie about any thefts?
#4 - Did you lie about any serious crimes you may have committed?
#5 - Did you lie about violating anyone's civil rights while in your custody?

And a subject had been truthful with all information except his prior drug use, it would stand to reason that his response to question two would stand out on the chart ("peak of tension") compared to the rest of the questions.  Therefore the "control" is the baseline established by the other four questions.  And it would be expected that regardless of how many times these questions were asked of the subject, the results would likely be the same.

Now if a subject is asked this series of questions two times, your suggested countermeasures include producing a deceptive response (i.e. using countermeasures, "puckering", etc.) to one or two other questions.  Lets say round one went like this:

#1 - no countermeasure
#2 - lie
#3 - countermeasure
#4 - no countermeasure
#5 - countermeasure

and round two went like this:

#1 - countermeasure
#2 - lie
#3 - countermeasure
#4 - no countermeasure
#5 - no countermeasure

Then couldn't a polygrapher say that the subject was deceptive on all questions except question 4?  Certainly deception would be easy to discern with regards to the drug use question (question #2).  Using countermeasures as you've described for this "peak-of-tension" test really just relies on trying to confuse the examiner (by producing false-positives), but if it's really that subjective, couldn't he say that the subject showed deception on all but one question, and that he consistently showed deception about drug use?

I've never taken a polygraph, but I will be taking one soon.  I really don't have anything to hide, so I haven't decided whether or not I'll use countermeasures.  I have confidence the use of countermeasures in the control-relevant examination.  But I am unconvinced that countermeasures can be reliably used in the peak-of-tension test.  I don't trust polygraphy 100% either, but I don't see why polygraphers don't use the peak-of-tension test to combat countermeasures more than they do.