5 ms·

Interestingly you just quoted the very definition of ad hominem. What someone tends to say makes no statement about their current argument. You can choose to

by jbjohns 13y ago

Interestingly you just quoted the very definition of ad hominem. What someone tends to say makes no statement about their current argument. You can choose to dismiss them out of hand because they tend to lie, but to claim that what they are currently saying is false because they are liars is a fallacy. Remember the boy who cried wolf? The take away from that story should be that every statement has to be evaluated in isolation.

lisper 13y ago
> What someone tends to say makes no statement about their current argument. Not so. If someone has lied in the past, that is evidence that their conscience permits them to lie, and so they are more likely to lie in the future. > The take away from that story should be that every statement has to be evaluated in isolation. Again no. The takeaway is that you should not lie, because if you do people will be less likely to believe you in the future.
- jbjohns 13y ago
  > Not so. If someone has lied in the past Fine. It makes no conclusive statement about their current argument. You still have to check it if you want to know if it's true or not. > The takeaway is that you should not lie, because if you do people will be less likely to believe you in the future. This is the over simplified view of the story. In my opinion, the subsequent attack on the village was 100% the fault of the others. The boy did his job and warned them of a wolf and they engaged in an ad hominem and dismissed his argument without even a cursory check. If they weren't willing or able to work with the boy then it's their fault again for giving him a job they didn't trust him to do.
  - lisper 13y ago
    You need to brush up on your Bayesian reasoning. In the story, the boy lies repeatedly before finally telling the truth once. If you do the math you will see that the villagers acted properly (or at least rationally). The Bayesian prior on P(wolf) is pretty small to begin with, and the posterior probability of P(wolf|boy-cries-wolf) gets smaller with every observation of ~wolf|boy-cries-wolf. You are being distracted by the fact that the data comes from a boy rather than an impersonal predictor like, say, some astrological sign. Your Bayesian prior on P(X|someone-says-X) is high, and that is as it should be. Evolution sees to it that people are basically honest. But evolution produces a lot of variation, including pathologies like lying. You're failing to update your estimate of P(Y-is-a-pathological-liar|Y-lied).
    - jbjohns 13y ago
      It has nothing to do with Bayesian, we're talking logic here. An astrological sign would not fall under logic, but rather the domain of science where it can be demonstrated to have no predictive power. A human can decide to stop lying, a sign can't decide to suddenly have predictive scientific properties. The village could certainly reason that the boy usually lied. But it is a fallacy to say "this argument is false because he is a liar". He might not be lying this time (indeed, the last time he was not). It would have been wise of them to get someone else to do the wolf watching, however, since they couldn't work with this boy.
      - lisper 13y ago
        > It has nothing to do with Bayesian, we're talking logic here. You really don't understand this at all. There is no such thing as single Platonic "logic" independent of any qualifier. There are many different kinds of logic. There is propositional logic, first order logic, higher order logic(s), modal logics, extensional and intensional logics. Bayesian reasoning is a (particular kind of) logic. Specifically, it is a logic of probability, which is to say, the kind of logic that applies when you don't know for certain whether something is true of false. Which is to say, the only kind of logic that actually applies to real-world situations. > A human can decide to stop lying, Can they? All of them? Are you sure? There are humans who can't decide to (say) stop smoking, how can you be certain that there aren't humans who can't decide to stop lying? > a sign can't decide to suddenly have predictive scientific properties. Of course a sign can't "decide" anything because it doesn't have a brain, but that doesn't mean it couldn't suddenly start to have predictive power. You can't rule out the possibility that the laws of physics could change tomorrow on any logical basis. The best you can (logically) do is observe that there is a ton of EVIDENCE that thaw laws of physics don't change capriciously, and conclude that therefore the probability they will change tomorrow is low (and Bayes's theorem will tell you exactly how low if you want to crunch the numbers). Likewise, there is a ton of EVIDENCE to support the theory that the boy in the story is a pathological liar who is not capable of deciding not to lie, or for whatever reason chooses to lie notwithstanding that he might have the ability not to. The only issue is how many lies it would take before one's estimate of the posterior probability of the boy telling the truth was indistinguishable from zero. This has to do with one's Bayesian prior on the probability of a particular random individual being a pathological liar, but that's beside the point. The point is that there is some finite number of lies after which the theory that what the boy says is in any way correlated with the truth (except by chance) is no longer logically tenable.

by jbjohns 13y ago

lisper 13y ago

> What someone tends to say makes no statement about their current argument. Not so. If someone has lied in the past, that is evidence that their conscience permits them to lie, and so they are more likely to lie in the future. > The take away from that story should be that every statement has to be evaluated in isolation. Again no. The takeaway is that you should not lie, because if you do people will be less likely to believe you in the future.

jbjohns 13y ago
> Not so. If someone has lied in the past Fine. It makes no conclusive statement about their current argument. You still have to check it if you want to know if it's true or not. > The takeaway is that you should not lie, because if you do people will be less likely to believe you in the future. This is the over simplified view of the story. In my opinion, the subsequent attack on the village was 100% the fault of the others. The boy did his job and warned them of a wolf and they engaged in an ad hominem and dismissed his argument without even a cursory check. If they weren't willing or able to work with the boy then it's their fault again for giving him a job they didn't trust him to do.
- lisper 13y ago
  You need to brush up on your Bayesian reasoning. In the story, the boy lies repeatedly before finally telling the truth once. If you do the math you will see that the villagers acted properly (or at least rationally). The Bayesian prior on P(wolf) is pretty small to begin with, and the posterior probability of P(wolf|boy-cries-wolf) gets smaller with every observation of ~wolf|boy-cries-wolf. You are being distracted by the fact that the data comes from a boy rather than an impersonal predictor like, say, some astrological sign. Your Bayesian prior on P(X|someone-says-X) is high, and that is as it should be. Evolution sees to it that people are basically honest. But evolution produces a lot of variation, including pathologies like lying. You're failing to update your estimate of P(Y-is-a-pathological-liar|Y-lied).
  - jbjohns 13y ago
    It has nothing to do with Bayesian, we're talking logic here. An astrological sign would not fall under logic, but rather the domain of science where it can be demonstrated to have no predictive power. A human can decide to stop lying, a sign can't decide to suddenly have predictive scientific properties. The village could certainly reason that the boy usually lied. But it is a fallacy to say "this argument is false because he is a liar". He might not be lying this time (indeed, the last time he was not). It would have been wise of them to get someone else to do the wolf watching, however, since they couldn't work with this boy.
    - lisper 13y ago
      > It has nothing to do with Bayesian, we're talking logic here. You really don't understand this at all. There is no such thing as single Platonic "logic" independent of any qualifier. There are many different kinds of logic. There is propositional logic, first order logic, higher order logic(s), modal logics, extensional and intensional logics. Bayesian reasoning is a (particular kind of) logic. Specifically, it is a logic of probability, which is to say, the kind of logic that applies when you don't know for certain whether something is true of false. Which is to say, the only kind of logic that actually applies to real-world situations. > A human can decide to stop lying, Can they? All of them? Are you sure? There are humans who can't decide to (say) stop smoking, how can you be certain that there aren't humans who can't decide to stop lying? > a sign can't decide to suddenly have predictive scientific properties. Of course a sign can't "decide" anything because it doesn't have a brain, but that doesn't mean it couldn't suddenly start to have predictive power. You can't rule out the possibility that the laws of physics could change tomorrow on any logical basis. The best you can (logically) do is observe that there is a ton of EVIDENCE that thaw laws of physics don't change capriciously, and conclude that therefore the probability they will change tomorrow is low (and Bayes's theorem will tell you exactly how low if you want to crunch the numbers). Likewise, there is a ton of EVIDENCE to support the theory that the boy in the story is a pathological liar who is not capable of deciding not to lie, or for whatever reason chooses to lie notwithstanding that he might have the ability not to. The only issue is how many lies it would take before one's estimate of the posterior probability of the boy telling the truth was indistinguishable from zero. This has to do with one's Bayesian prior on the probability of a particular random individual being a pathological liar, but that's beside the point. The point is that there is some finite number of lies after which the theory that what the boy says is in any way correlated with the truth (except by chance) is no longer logically tenable.