I can easily accept that Eckman's microexpressions work significantly better than chance in a laboratory setting, just like the puffer, but when you apply it to a detection system where your targets are so incredibly rare that the practitioners will never be able to see an actual terrorist in their entire careers, you are going to get predictably bad results. Just like the puffer--the false alarms necessary to get an effective detection percentage make the implementation infeasible.

Ekman is just one more contractor selling some impractical magical thinking to our overfunded professional paranoids at TSA.

Another strike against the BDO program as compared with the failed puffer program is that BDO is visual, and there aren't enough true terrorists out there to train them effectively. The best you can do is role-play with some actors or photoshop some pictures. If that isn't a load of pathetic security theatre, I don't know what is.

Again, a BDO may very well pick the terrorist-actor out of a lineup of non-terrorist-actors significantly better than chance. But that performance is more like 1-in-7 rather than 1-in-a-billion. BDO isn't something objective like using a magnet to pick a needle out of a haystack as it's being loaded by conveyor, it's more like like trying to visually estimate needle-like lengths through a B&W CCTV.

Taking this back towards the OP, it is a valid criticism of the BDO program that they fail to find criminals within TSA's own ranks. If BDOs are toning down their spidey-sense so that they don't tweak to thieves, perverts, and warrant offenders, they are also degrading their ability to detect the mythical 1-in-a-billion terrorist.