Hacker News Reader

Comment by tyingq | Hacker News Reader

Oy. Similarly, I'm mystified with the CIA's and NSA's fascination with mandatory recurring polygraph tests. It's pretty well known the whole area is flawed bullshit. It doesn't measure anything remotely related to "truth".

<b>Edit:</b> I know we aren't supposed to discuss downvotes here, but the activity on this comment and my replies are very interesting. It's curious why discussion on this topic is so controversial. Why wouldn't the UK want it's own foreign policy distinct from the US policy? Downvotes welcome. I give zero shits about my HN "karma score".

View on HN

I think the polygraph is really just an additional tool to aid the main goal, which is to interrogate the participant. To this end the polygraph is a useful tool as it provides added discomfort, which can help the interrogator get answers from the participant.

Depends, I guess, on what they do with it. Indicators of stress could be a "good thing" in a lot of cases. It might show maturity and broad thinking versus deception. I'm a bit biased in thinking they probably bias towards loyalty and party line vs the real story. In any case, it probably excludes a lot of real talent that won't put up with the bullshit. They are inadvertently creating a very narrow culture that hurts their effectiveness.

It's just a technique to scare you into providing evidence. And in the CIA it's a test of your resilience to interrogation.

while I'm sure plenty of people in law enforcement haven't 'gotten the memo' that polygraphs don't actually work, I suspect in many cases they do them anyways exactly for the reason you describe.

Reminds me of one of my favorite scenes from The Wire, Bunk using his bag of tricks to get someone to talk. Whole thing is hilarious, but the lie detector stuff starts here:

https://youtu.be/rN7pkFNEg5c?t=92

In France, until not so long ago, analysing your hand writing was a commonly accepted form of background test for any large company!

That sounds interesting. Any links, sources?

Lots of them: https://www.google.com/search?q=french+handwriting+analysis

Similarly..unverified but word of mouth, they use face analysis in Korea for hiring.

In western Europe, job postings specifically ask not to include a picture in your application to combat implicit bias.

In Korea, the application form includes fields for height, weight and marital status.

It's pretty well known that it is flawed bullshit when applied a small number of times to a small number of people that you have limited information where you are trying to find out the truth about something very specific.

Applied repeatedly over a long time to a large group of people that you have extensive information on where you are aren't looking for information about some specific thing, by an agency with extensive experience in extracting meaningful signals from noisy sources...I would not be at all surprised if they can get something actually valid from it.

They don’t. Unless you count giving a Lebanese Hizballah illegal a security clearance, which might have been due to something other than the polygraph but is hard to reconcile with any background check actually being performed.

Polygraphs could work as advertised for 99.99% of people being processed for security clearance and still be useless, because their effectiveness against people who take the job with the intention of leaking, selling, or otherwise inappropriately handling classified information is nil. The US government has spent tons of money on polygraph research and administration over the last century despite knowing from the beginning that it was a fraud, and has never even once prevented an act of espionage with one.

Or drug-sniffing dogs.

These things all serve their purpose: providing cover for your preconceived notions.

Yeah. It's pretty interesting that popular culture like Dateline, 20/20, Forensic Files, etc, never digs into things like "talk to my lawyer", polygraphs or drug dogs. They tow the line of "cooperate with the cops or you're guilty". They regularly position this sort of thing as an admission of guilt.

I think pop culture is actually culpable here. It leads to reinforcing lay people's belief in these techniques, and especially in the US with the jury system, kay people tend to have more influence on the punishment.

For some people the very process of having to go through a polygraph can set off other triggers

Sure. But that selection bias creates a sort of monoculture of mindless drones.

AFAIK the standard script for polygraph tests is for the interrogator to say there's a "trigger", regardless of whether there's actually a statistically significant change in the polygraph output. The idea is convince the participant that they know they're lying and push them to incriminate themselves.

IMO discussion forums should show the user comments that they'd likely agree with instead of comments that some majority-of-the-moment agrees with.

Sorting comments on a majority average is such a dumb metric. The Internet would be a better place if forum discussions more closely modeled real life, where we simply don't have disagreeable opinions constantly shoved in our face because we don't hang around with those people who have them.

One simple way I'd do it: If I upvoted someone in the past, I'd give their comments and upvotes more weight. I'd also give people a way to visit the other side and break free of their bubble if they wanted to.

Insightful, thank you. It's a bit theoritcal, but I'd like a way for replies to be sorted by merit versus controversy or popularity. I'm not adverse to the idea that I'm wrong.

How would you quantify merit?

Reply on news.ycombinator.com