Hacker News Reader

Comment by clicks | Hacker News Reader

http://www.wikihow.com/Cheat-a-Polygraph-Test-(Lie-Detector)

has several good tips on how to cheat polygraph tests. I actually strongly encourage everyone to read up on polygraph tests and how to cheat them, here's why: they've been wrong again and again, they've gotten innocent men in trouble. There's a very real probability that the polygraph can say you're lying even when you're not lying. To save yourself from getting caught by these pseduo-scientific tests if ever you're in the position, prepare yourself now, learn about the polygraph tests and how you can use them to your benefit.

View on HN

> they've gotten innocent men in trouble. There's a very real probability that the polygraph can say you're lying even when you're not lying.

This is because the polygraph doesn't measure incidents of lies, but it measures fear. Fear of wrongful conviction can be interpreted as lying, which is how it has gotten innocent people (not just men!) in trouble.

Probably the mere fact that you've spent a lot of money on anti-polygraph training does a lot to reduce that fear, regardless of what techniques you were taught.

On antipolygraph.org, there is a free download of a much more in depth (and apparently well-researched) analysis and description of techniques called The Lie Behind The Lie Detector. Note that they very reasonably advise that unless there's pretty much no other choice, the best way not to fail a polygraph is not to take one at all.

I don't know about you, but my programs never compile the first time I write them. I wouldn't recommend anyone using these techniques on the internet unless you can test them out on a real polygraph.

The problem is it's very hard to test even with a real polygraph. You'd have to somehow induce real fear. It's not just about lying, in of itself.

The bad news is that replicating an actual polygraph scenario is basically impossible. Even if you go in for a real polygraph, it won't be the same as having something riding on it.

The good news is twofold:

1. It's seems not to be as difficult to fool a polygraph as it is to get code to compile.

2. You can measure some of the factors they look at. A heart rate monitor, for example, can aid in practicing raising your pulse.

You can replicate a real test, but it requires considerable effort:

First, you need someone capable of imitating a trained polygrapher. The simplest option is to hire one, but from what I know of polygraphy I'm pretty sure that a magician could do the job with a few hours study.

Second, you need an accurate simulation of real polygraphy equipment. Again, you can just pay for a test, or you can rig something up with a galvanometer, heart rate monitor, and spirometer.

Thirdly, as you said, you see to have "something riding on it." Hand a trusted confidant a quarter of your monthly income and tell them to give it to a charity antithetical to your beliefs if you fail the test.

An interesting idea. Actually, since various sources have gotten ahold of standard scripts, you could probably have a friend of a friend (you need someone unfamiliar) do the administration, although that won't help with the "trained interrogator" factor.

Amazon carries a $100 USB polygraph machine, but most of the reviews say it gives the same readings regardless of whether you attach it to a person. Assembling your own stuff might make more sense.

>they've been wrong again and again...

So how can you know how to cheat them?

But if they are any good, then surely the technicians recognize false negatives. "Hah! You're trying to cheat the polygraph! That's a crime!"

It ain't Kafka anymore. It's USA. I can't believe I used to worry about redneck country cops and a few days taking it in the pokey. Might as well just take your chances with a RND function if you're detained now. Of course, if you're a young black male the algo is weighed heavily against you. Money buys a lot of decimal places.

>So how can you know how to cheat them?

I don't understand the basis of your question. To say "this test does not correlate with the phenomenon it purports to measure" is not the same as to say "this test is completely random".

Counting how many times a person says the word "baby" during an answer is a piss-poor way to tell if they're lying, in large part because that factor won't correlate strongly with deception. It would also be extremely easy to fool, by simply controlling how many times you say "baby".

Didn't mean to say polygraphs are random, but that the US jusice system is so out whack that once someone is under suspicion, conviction (for something) has little to do with guilt or innocence on initial grounds.

Reply on news.ycombinator.com