Wandersmann, from what I've seen, polygraph continues to be BS (belief system) based. It is far closer to a religious cult than it is to a hard science, in my opinion.

For example, the American Polygraph Association claims average accuracy for incident-specific exams is 89%. However, their research, to the best of my knowledge, does not encompass across-the-board randomness of examiner expertise, which varies widely. Further, the APA's research does not include a vigorous countermeasure component.

In light of those two fatal shortcomings, I'd say that real-world polygraph accuracy is far closer to 59% than it is to the APA's claim of 89%.

As for the government's use of polygraph, it seems clear that their security-oriented "war footing" mentality trumps concerns over collateral damage.

For example, we see the same justification in drone strikes targeting high-value targets. Innocents are often sacrificed in order to get the bad guy.

Save for the blood and guts, the same overall dynamics are at play in the gummint polygraph suite.