Fortunately, I found out about the "Beat the Lie Detector" secondary story in the most recent episode of MythBusters before I watched it. So I knew to fast-forward through the lie detector story and just watch the other one.
This was entirely for the sake of my health. There's no way I could watch someone claiming that polygraphs are "80% to about 99% accurate", and then see a screen shot of software saying "Probability of Deception is Greater Than .99", without dangerously elevating my own metabolic markers.
(But yes, I've skimmed through the lie detector story now, just to make sure the complaints are valid. They are.)
The icing on the cake is the fact that the person making the "80 to 99%" claim, and later administering the polygraph tests, was "Doctor" Michael Martin. Who, apparently, bought his doctorate from a diploma mill.
You certainly don't need university qualifications to be knowledgeable about a subject, but fake degrees are anti-qualifications. Nobody who bought a diploma to make themselves look qualified in an area should be believed about anything, until they say they're sorry and take the unearned honorifics off their business card.
In reality, it is arguable that the polygraph is not entirely useless. (This may set some sort of record for damning with faint praise.)
The polygraph looks especially good if you, unfairly, count the cases in which it's used merely as an intimidation device to trick a guilty person into confessing. Wen Ho Lee, for instance, passed his polygraph test with flying colours - but that was no problem for the Feds, who just said he'd failed. It's like police interrogators telling a suspect that their buddy has already confessed, when no such thing has actually happened.
Contrary to not-a-real-doctor Michael Martin's statement, the polygraph's history is one big losing streak. Nobody's ever actually been able to demonstrate, in proper controlled tests, that the darn thing is actually worth using. Not that many governments or corporations seem to listen when the National Academies of Science tell them as much.
The NAS actually concluded that, although the polygraph is the best lie-detection device created so far, it's still worse than useless, thanks to its high false-positive rate. The essential randomness of the polygraphic process means that although it certainly is possible to "beat" a polygraph test, there are no guarantees; no matter how innocent you look (because you know the tricks, or because you really are innocent), the polygraph operator may still decide you're guilty.
The popular conception that a polygraph actually does "detect lies" in any straightforward sense is entirely wrong. An honest TV show should make this clear, and not give air time to someone who proudly states the opposite, whether or not that person has valid qualifications.
MythBusters genuinely does make an effort to get things right, which makes them almost unique in the "reality TV" field, and quite unlike certain other shows in their own niche.
This time, though, they appear to have dropped the ball very seriously indeed.
This isn't just a procedural error, oversimplification or scientific mistake on the level of getting the shape of a raindrop or the principles of operation of a wing wrong. It's a big ol' slab of prime-time bullshit.