13 ms·

A lie-detector that works is a trillion-dollar business. A great sci-fi novel that explores the implications of an accurate lie-detector is "The Truth Machine".

A lie-detector that works is a trillion-dollar business. A great sci-fi novel that explores the implications of an accurate lie-detector is "The Truth Machine". I've given a lot of thought to this topic over the years, and have concluded that the way to tackle this problem is indirectly. The problem with "truth" is that it is subjective. There is a better metric to judge: "false-confidence". There is a subtle but distinct difference between a lie and false-confidence. A lie is subjective, being a mismatch between what I believe and what is true. But false-confidence is objective, in that what I believe (regardless of the truth of what I believe) may not match what I assert. We can never know, objectively, if a person is speaking the truth. But we can know, objectively, if a person is speaking with false-confidence. This might seem like semantics, but focusing on "false-confidence" rather than "truth" is quite helpful in highlighting the real problem that is being solved: communication.

rdiddly 6y ago
The best liars, like the best actors, actually believe what they're saying though, so I don't think false confidence is the answer.
- wildermuthn 6y ago
  That's my point, though. A lie-detector tries to do the impossible. A false-confidence assessor only does what it claims to do.
lazide 6y ago
That doesn’t seem like the real problem being solved or highlighted? Communication is a vector here that is surfacing a fundamental competitive force and strategy between actors? Information is power. Knowing something, and that being representative of reality and providing predictive ability related to that thing provides you power in a number of ways (ability to predict and intervene or not depending on desired outcomes being a huge one). Someone presenting you plausibly correct, yet misleading information (a good lie) provides them a major advantage (short term, or long term depending on the situation). It allows them to influence your behavior and actions in a way they can likely predict and influence your ability to predict or react accurately to something. A lie detector (as compared to an objective truth detector) would tell you if you were being manipulated or mislead intentionally. Very different from a ‘do they actually know what they are talking about’ aka truth detector like you are referring to, but a very valuable tool if one existed in a competitive world like the one we exist in.
- wildermuthn 6y ago
  I think we're saying the same thing. People manipulate one another through communication, not only in a criminal-justice setting, but also in business and relationships. Using "false-confidence" as a metric is simply reminding us that deception has no connection with objective truth. You can lie even while communicating something that is objectively true. You can also lie while intentionally displaying outrageously high false-confidence (i.e., jokes). It is easier to reason about this topic if we choose the right name.
  - lazide 6y ago
    Reading your description though, the only way I can imagine such a machine working would be by analyzing what is said against an objective truth - and that is what the false confidence value is measuring? If I’m confident of something that is wrong, my false confidence number would be very high - especially if I believed it, right?
simonh 6y ago
A lie is a mismatch between what I believe and what I say. It has nothing to do with objective truth. We sometimes talk about lying versus telling the truth, but this is a shorthand. What we mean by telling the truth is saying what we believe to be true. No reasonable person would say that someone is lying if they say what they believe to be true, but it turns out they are mistaken.
- 13415 6y ago
  Not really, if you think about it, lying is knowingly saying what you believe to be false and what you say is also false. If what you say is accidentally true, then you tried to lie but failed. If you misdirect someone by saying the truth - e.g. because you know the person would assume the opposite of what you say or because you omit a presupposed fact such as that the gas station someone asked for is not closed -, then that's better regarded as deceiving. At least that's the theory of a dear colleague of mine who published that in various prestigious philosophy journals. I think he got it right. What I disagree with him about is the presumption that there is a "right" theory of lying. In the end, you can define it in different ways, but I do think his definition is a very good one.
  - wildermuthn 6y ago
    I love this. But this is also why I think using "false-confidence" as a metric avoids all these questions by being unambiguous about what is being measured.
  - simonh 6y ago
    That’s not right at all. Lies are false statements about your own knowledge. Even if the fact your lying about happens to be true, you are still lying about your knowledge of it. It’s important to distinguish between the act of lying, the intended result of the lie and the relationship to the true facts. Those are all independent things. You can’t “fail to lie”, though you can fail to deceive with your lie. Suppose my daughter is accused of stealing the sweets. I might lie by saying I saw her in the garden at the time they were taken. Even if she was actually in the garden then, and didn’t take the sweets, I’m still lying about my knowledge of these things because I didn’t see her in the garden. Even if I lie about remembering something, such as that I saw her in the garden (maybe it was a long time ago), and it turns out I did see her in the garden I’m still lying because I didn’t recall the memory when I said I did. >“If what you say is accidentally true, then you tried to lie but failed.” No, because you still misrepresented your own knowledge. You did not deceive someone with your lie and maybe led them to the truth, but it was still a lie.

rdiddly 6y ago

The best liars, like the best actors, actually believe what they're saying though, so I don't think false confidence is the answer.

wildermuthn 6y ago
That's my point, though. A lie-detector tries to do the impossible. A false-confidence assessor only does what it claims to do.

lazide 6y ago

That doesn’t seem like the real problem being solved or highlighted? Communication is a vector here that is surfacing a fundamental competitive force and strategy between actors? Information is power. Knowing something, and that being representative of reality and providing predictive ability related to that thing provides you power in a number of ways (ability to predict and intervene or not depending on desired outcomes being a huge one). Someone presenting you plausibly correct, yet misleading information (a good lie) provides them a major advantage (short term, or long term depending on the situation). It allows them to influence your behavior and actions in a way they can likely predict and influence your ability to predict or react accurately to something. A lie detector (as compared to an objective truth detector) would tell you if you were being manipulated or mislead intentionally. Very different from a ‘do they actually know what they are talking about’ aka truth detector like you are referring to, but a very valuable tool if one existed in a competitive world like the one we exist in.

wildermuthn 6y ago
I think we're saying the same thing. People manipulate one another through communication, not only in a criminal-justice setting, but also in business and relationships. Using "false-confidence" as a metric is simply reminding us that deception has no connection with objective truth. You can lie even while communicating something that is objectively true. You can also lie while intentionally displaying outrageously high false-confidence (i.e., jokes). It is easier to reason about this topic if we choose the right name.
- lazide 6y ago
  Reading your description though, the only way I can imagine such a machine working would be by analyzing what is said against an objective truth - and that is what the false confidence value is measuring? If I’m confident of something that is wrong, my false confidence number would be very high - especially if I believed it, right?

simonh 6y ago

A lie is a mismatch between what I believe and what I say. It has nothing to do with objective truth. We sometimes talk about lying versus telling the truth, but this is a shorthand. What we mean by telling the truth is saying what we believe to be true. No reasonable person would say that someone is lying if they say what they believe to be true, but it turns out they are mistaken.

13415 6y ago
Not really, if you think about it, lying is knowingly saying what you believe to be false and what you say is also false. If what you say is accidentally true, then you tried to lie but failed. If you misdirect someone by saying the truth - e.g. because you know the person would assume the opposite of what you say or because you omit a presupposed fact such as that the gas station someone asked for is not closed -, then that's better regarded as deceiving. At least that's the theory of a dear colleague of mine who published that in various prestigious philosophy journals. I think he got it right. What I disagree with him about is the presumption that there is a "right" theory of lying. In the end, you can define it in different ways, but I do think his definition is a very good one.
- wildermuthn 6y ago
  I love this. But this is also why I think using "false-confidence" as a metric avoids all these questions by being unambiguous about what is being measured.
- simonh 6y ago
  That’s not right at all. Lies are false statements about your own knowledge. Even if the fact your lying about happens to be true, you are still lying about your knowledge of it. It’s important to distinguish between the act of lying, the intended result of the lie and the relationship to the true facts. Those are all independent things. You can’t “fail to lie”, though you can fail to deceive with your lie. Suppose my daughter is accused of stealing the sweets. I might lie by saying I saw her in the garden at the time they were taken. Even if she was actually in the garden then, and didn’t take the sweets, I’m still lying about my knowledge of these things because I didn’t see her in the garden. Even if I lie about remembering something, such as that I saw her in the garden (maybe it was a long time ago), and it turns out I did see her in the garden I’m still lying because I didn’t recall the memory when I said I did. >“If what you say is accidentally true, then you tried to lie but failed.” No, because you still misrepresented your own knowledge. You did not deceive someone with your lie and maybe led them to the truth, but it was still a lie.