Similarly, by your standard the fact that had had sex with at least 7 men by the age of 20 is “evidence.” Why? First, because it raises the probability that she suffers from borderline personality disorder and therefore is lacking in empathy.
Yes, this is all evidence according to the Bayesian definition. Calling E evidence (with respect to prior knowledge X) for a proposition H just means that p(H | E & X) > p(H | X). That is why quantifying evidence is so important. Just how much evidence is it? If all the evidence you offer raises the probability of H by only a few percent from a very low prior, then it should have practically no effect on how we treat Knox.
Yes, this is all evidence according to the Bayesian definition. Calling E evidence (with respect to prior knowledge X) for a proposition H just means that p(H | E & X) > p(H | X). That is why quantifying evidence is so important. Just how much evidence is it? If all the evidence you offer raises the probability of H by only a few percent from a very low prior, then it should have practically no effect on how we treat Knox.
I basically agree, but sometimes it’s helpful to have a story or narrative or illustration before jumping in to look at the important evidence. That’s just how the human mind works, or at least most peoples’ minds. I realize this can be dangerous, for example it can lead to a “conjunction fallacy,” but I was careful to lable my narrative as speculation.
Just today I was browsing this web site and I came across an article called “Existential Risk” which was complete with (1) a picture of the Earth; (2) a likely apocryphal story about a man who singlehandedly prevented nuclear war; and (3) a picture of a Stanford torus. Is this cheap emotional manipulation? Perhaps, but again, I think this kind of story-telling can be useful to get the mind ready to focus on the meat of the argument.
One can ask what the likelihood is that we are reaching a critical juncture where the decisions and dilligence of just a few humans in the artificial intelligence community will have a massive impact on the future of humanity. Strictly speaking, the fact that some Russian dude did (or didn’t) singlehandedly prevent a nuclear war shouldn’t have much impact on our estimate of this probability. But I think it still might be worth mentioning to demonstrate the plausibility of the claim that one person can have a big impact.
In the same way, I think it’s worth mentioning the Janet Chandler case from the 70s. But again, if you object to this approach, just ignore paragraphs 2 through 7 of my blog post.
In the same way, I think it’s worth mentioning the Janet Chandler case from the 70s. But again, if you object to this approach, just ignore paragraphs 2 through 7 of my blog post.
I agree that it’s worth mentioning Janet Chandler. It would be better to treat it seriously as evidence, rather than merely as a narrative framing device. To treat it seriously as evidence, you should use it to help establish a prior probability for Knox’s guilt (like Desrtopa did).
It would be better to treat it seriously as evidence, rather than merely as a narrative framing device. To treat it seriously as evidence, you should use it to help establish a prior probability for Knox’s guilt
It seems to me that “narrative framing device” is basically a poor man’s method of estimating a prior probability. Here’s what I said in my blog post:
The point is that there are levels of extraordinary. Claiming that Knox participated in her roommate’s murder is not like claiming that the president is actually an extra-terrestrial from Mars.
Of course in terms of assessing probabilities, it might be better if there were a lot of precedents, for example in a situation where a husband is suspected of killing his wife. But here there’s not a lot to go on.
Yes, this is all evidence according to the Bayesian definition. Calling E evidence (with respect to prior knowledge X) for a proposition H just means that p(H | E & X) > p(H | X). That is why quantifying evidence is so important. Just how much evidence is it? If all the evidence you offer raises the probability of H by only a few percent from a very low prior, then it should have practically no effect on how we treat Knox.
I basically agree, but sometimes it’s helpful to have a story or narrative or illustration before jumping in to look at the important evidence. That’s just how the human mind works, or at least most peoples’ minds. I realize this can be dangerous, for example it can lead to a “conjunction fallacy,” but I was careful to lable my narrative as speculation.
Just today I was browsing this web site and I came across an article called “Existential Risk” which was complete with (1) a picture of the Earth; (2) a likely apocryphal story about a man who singlehandedly prevented nuclear war; and (3) a picture of a Stanford torus. Is this cheap emotional manipulation? Perhaps, but again, I think this kind of story-telling can be useful to get the mind ready to focus on the meat of the argument.
One can ask what the likelihood is that we are reaching a critical juncture where the decisions and dilligence of just a few humans in the artificial intelligence community will have a massive impact on the future of humanity. Strictly speaking, the fact that some Russian dude did (or didn’t) singlehandedly prevent a nuclear war shouldn’t have much impact on our estimate of this probability. But I think it still might be worth mentioning to demonstrate the plausibility of the claim that one person can have a big impact.
In the same way, I think it’s worth mentioning the Janet Chandler case from the 70s. But again, if you object to this approach, just ignore paragraphs 2 through 7 of my blog post.
I agree that it’s worth mentioning Janet Chandler. It would be better to treat it seriously as evidence, rather than merely as a narrative framing device. To treat it seriously as evidence, you should use it to help establish a prior probability for Knox’s guilt (like Desrtopa did).
It seems to me that “narrative framing device” is basically a poor man’s method of estimating a prior probability. Here’s what I said in my blog post:
Of course in terms of assessing probabilities, it might be better if there were a lot of precedents, for example in a situation where a husband is suspected of killing his wife. But here there’s not a lot to go on.