[roel_v]

Yes, and I built a very simple one in law school, to illustrate the concept (well to be honest I was an engineer who went to law school as a hobby, and then I took every excuse I could find to build technology things as a hobby within a hobby). Anyway it was just a multimeter, some washers soldered to copper wire and some tape to keep the washers in their place on the subject's hands. When people didn't try to game it, it actually sort of worked, although I have no proof of statistical significance; I just had a feeling of 'better than random'. (our experiments were more focused on how to ask questions to elicit the highest response, as questioning with a polygraph requires some skill at asking questions that are leading or not, and in just the right ways).

Conclusion is still that it's nowhere near reliable enough to be used for anything, but the principle that people show physiological responses when answering truthfully or not about things is sort of true (I don't think that was ever under debate, still was a fun experiment).

[astine]

Just a guess, but I wonder if a polygraph is useful as an intimidation tactic? Maybe you can't detect lies with any sort of accuracy but you can intimidate people into making more mistake when answering questions and that can be used to figure out if someone is lying? OR perhaps they can pressure some people into revealing more information than they otherwise would?

[roel_v]

Yes that's how it's used. That depends on people believing polygraphs work though, which is why polygraph companies for decades have been litigating against people who question polygraphs. Sometimes just the idea that there will be a polygraph test is enough to get people to be more truthful. It's like placebo drugs - if it works somehow, should we care about whether it actually works? By that reasoning, should we allow homeopathic drugs for healthcare insurance reimbursements? If Josie doesn't have a headache after taking pills, does it matter whether those pills are paracetamol or water with a story of it being very potent water? Fun discussions, not very productive, but still fun (for a while).

[astine]

"That depends on people believing polygraphs work though"

Are you sure that's true? In medicine, apparently placebos have an effect even if you know that it's a placebo. I imagine that even if you know that it's theoretically useless, being hooked up to a machine that can read your vitals while lying would still be nerve wracking.

[ampdepolymerase]

Having an effect is different from being able to sell a product based on that effect. For regulated fields like pharmacy, the FDA regulatory approval test requires proving efficacy against a placebo. A doctor may (under certain specific circumstances) treat a patient using a placebo (because medicine's aim is to do whatever that is necessary to alleviate the problem or reduce suffering, it is different from pure biological natural sciences/pharmacy where inferences must be backed up by scientific theory and validated by empirical evidence) but it does not necessarily mean that a placebo can be sold/marketed as a cure.

[roel_v]

To a degree of course, just 'but it might work' would be enough I guess, the same way someone needs to at least not be told something is a placebo for a double blind study to be "valid" ("methodologically sound" would a better term I think?). I did read quite a lot on the topic and much has been written about it over the years, but this was 10+ years ago for me, I don't remember enough to really make a coherent argument that addresses the details.