Slashdot Mirror

← Back to Stories (view on slashdot.org)

The Eyes Have It

Posted by michael on from the do-you-still-love-me dept.

Feelgood writes: "Yahoo is carrying a Reuters report that thermal imaging may be used in airports to detect liars. Shouldn't be a problem that 1 out of 4 liars will get away and 1 in 10 innocents will be incorrectly nailed." There's a UPI story about the lie detector possibilities and a blurb in Nature. From the UPI article, the inventor has a good appreciation of the ethical considerations. Will anyone else care?

10 of 320 comments (clear)

  1. Once more, for all the slow JBT's. by jcr · · Score: 5, Informative


    I would like to point out that while Aldritch Ames was in the process of getting a whole lot of US agents in the Soviet Union killed by ratting them out, he continued to pass his polygraph tests.

    There's no such thing as a lie detector. Polygraphs are voodoo, and so is this.

    -jcr

    --
    The only title of honor that a tyrant can grant is "Enemy of the State."
    1. Re:Once more, for all the slow JBT's. by 4n0nym0u53+C0w4rd · · Score: 3, Informative

      This technique has been used for years in psychology studies (and by bad cops on naive suspects). It's called the "Bogus Pipeline"[1]. The basic idea is that, given a quick demonstration of its effectiveness (instruct the target to tell a couple of lies, and show that the machine detects them), along with an incentive to avoid being caught (if you're honest with us, we'll be easy on you), people are more likely to behave honestly.

      [1] Jones, E. E., and Sigall, H., "The Bogus Pipeline: A New Paradigm for Measuring Affect and Attitude," Psychological Bulletin 76: 349-364, 1971.

      --

      Buy Hex-Rated Stuff, fight the DMCA!

  2. Schneier said it by coltrane99 · · Score: 2, Informative

    On the impact of seemingly acceptable success rates on large-scale systems here

  3. off topic, troll, flaimbate by mokyar · · Score: 0, Informative

    Jon Katz, where are you?

  4. Comment removed by account_deleted · · Score: 4, Informative

    Comment removed based on user account deletion

  5. It's sounds better than a polygraph. by AnotherBlackHat · · Score: 5, Informative

    Polygraphs are considered about 75% accurate, which sounds good until you consider that flipping a coin is 50% accurate.

    If they can really catch 3 out of 4 liars, and "avoid" 9 out of 10 innocents,
    (which is what the article claims inventors claim) then it's much better than 75%.
    If 1 in 100 people are "liars" then this would be nearly 90% effective.
    Which again sounds good until consider that identifying everybody as innocent would be 99% accurate.

    On the plus side, this might make wearing eye shadow a crime under the DMCA.

    Polygraphs can be beat simply by putting a thumb tack in your shoe,
    and stepping on it during the "little bad" questions and not during the "big bad" question.
    (saying that probably makes this post a violation of the DMCA ...)
    I'd bet that this device can be beat by a similar method.

    1. Re:It's sounds better than a polygraph. by Saithier · · Score: 2, Informative

      If they can really catch 3 out of 4 liars, and "avoid" 9 out of 10 innocents, (which is what the article claims inventors claim) then it's much better than 75%. If 1 in 100 people are "liars" then this would be nearly 90% effective. Which again sounds good until consider that identifying everybody as innocent would be 99% accurate.


      Using those numbers it is actually far far less than 90% effective. If 1 in 100 are liars, that means 10 in 1000 are. You have two populations to work with 990 innocent people, and 10 who lie. The test is 90% accurate on the innocent, which means it false reports 99 people as liars (10%). The test is 75% accurate on liars, so (rounding up) it reports 8 of the 10 liars as guilty, and 2 as innocent. This gives us a total of 107 people reported as liars, when only 8 of those actually are a rate of 7.47% accuracy. And that doesn't count the 2 liars it missed!

      Cheers!
      **Saithier

  6. Re:how does this do anything? by digitalunity · · Score: 2, Informative

    Would a Lunatic pass a Polygraph if he really believed what he said?

    Yes. The causes for which a lie is detected are things like respiration and heart rate. These things wouldn't be affected by someone who isn't nervous about what they're saying.

    --
    You can't legislate goodness. Let each to his own destiny, by will of his freely made choices.
  7. Comment removed by account_deleted · · Score: 3, Informative

    Comment removed based on user account deletion

  8. Comment removed by account_deleted · · Score: 4, Informative

    Comment removed based on user account deletion