Context: I didn’t know anything about the case; I think I’ve overheard something about it in the past, because I had a very slight sensation of déjà-vu reading your post, but I didn’t really know anything. The info I used was reading the Wikipedia pages (one you linked and one on one of the trials), then skimming the sites you linked for less than a half hour.
I have reached a “decision” about it, but I didn’t intend to post it. Then curiosity got the best of me and I read pretty much all the comments already here. My “estimate” wobbled a bit during the reading of the comments, but by the end it converged on its initial value. I decided to post my opinion, as somebody else mentioned, to avoid self-selection.
I used quotes above because my general state is confusion. Actually, it’s confusion both about the case and about the poster’s opinions (and, a bit, about your post).
The general impression I got was that all parties (except, probably, the victim) are guilty of something, including the prosecutors. But...
First, I’m not very sure exactly if your questions should be translated as “probability estimate that X killed Kercher” or not. I actually have a slight suspicion that your post might be a trick question around that. (In fact, writing now, I realize that the trial resulted in guilty verdicts on ~five counts for each, but after all the reading I have no idea what they are other than that one of each is some kind of murder.)
Regarding Knox and Sollecito, my “gut” impression is that they’re guilty of something, in the sense that I’d be quite surprised if they were just two innocent people who did absolutely nothing bad (other than that joint) and got dragged in a nightmare by chance. I also have a “gut” feeling that their “guilt” is correlated.
Regarding Guede, I’ve got a slightly stronger gut feeling of guilt (than I do for the first two).
I also feel (as in “gut”, again) that the prosecution and trial was “wrong”. (EDIT for clarification: but I can’t see any basis of estimating how wrong between “as good as reasonably possible” and “gross miscarriage of justice”.)
However, what confuses me most is the certainty of most estimates I’ve seen in the comments. I just don’t see how one can get much farther than “I don’t know, but I lean a bit towards” either way on what the posters report as their information. I mean, after only a couple hours or so of thinking (both during reading and writing this) I’m not even sure what priors might be appropriate. I feel I know so little about the case (including info like trials in general) that my response can’t be more than “I don’t know”. (If pressed a lot I’d output 0.5±e as a probability, simply because even my knowledge of probabilities is too weak to actually figure out if anything else would express my almost total uncertainty on this subject, and that’s my cached answer for binary questions.)
Put differently, given the information I have on the subject, it feels as if the confidence I should assign to any probability estimate should be very close to 0. As I said above, I tend to give probabilities as ~50% in those cases, but that’s not exactly a correct description of what I feel. A good description would be that, if forced to bet (without more time to think), I’d bet on all of them being guilty, but if given the option, I wouldn’t bet any amount at 1:1 odds.
Oh, and the answer to the fourth question is very similar. Leaving aside the fact that I’m not sure exactly how I’d quantify the “coincidence” of opinions, I’m wouldn’t put any significant confidence in any estimate. (The “gut” feeling is that you consider all three guilty, but that’s mostly a very vague impression I got from the wording of your text rather than reasoning on my part.)
As an aside, the interesting thing about the subject is that I was surprised how little confidence I assigned to my “estimates” in this case. I usually have some opinion on most questions I encounter, but in this case I find myself really unable to arrive at one.
First, I’m not very sure exactly if your questions should be translated as “probability estimate that X killed Kercher” or not. I actually have a slight suspicion that your post might be a trick question around that
No, “X killed Kercher” is right. (Or perhaps better would be, “X participated in killing of Kercher”.)
It’s an academical point right now, given your later post, but even “X participated in killing of Kercher” is not entirely clear. Any action A, from cutting her throat, to holding her, to opening the door, to simply being in the room, can be considered as “participating” in the murder, in the sense that suitably complicated scenarios can be constructed where X not doing the action A would have avoided the murder, with X knowing it and deliberately choosing to do A in order to cause Kercher not to live anymore.
I imagine what you actually wanted to ask was something like the probability that “X is guilty of whatever X was convicted”, though that only moves the uncertainty about what you ask to uncertainty about what Italian law says.
Context: I didn’t know anything about the case; I think I’ve overheard something about it in the past, because I had a very slight sensation of déjà-vu reading your post, but I didn’t really know anything. The info I used was reading the Wikipedia pages (one you linked and one on one of the trials), then skimming the sites you linked for less than a half hour.
I have reached a “decision” about it, but I didn’t intend to post it. Then curiosity got the best of me and I read pretty much all the comments already here. My “estimate” wobbled a bit during the reading of the comments, but by the end it converged on its initial value. I decided to post my opinion, as somebody else mentioned, to avoid self-selection.
I used quotes above because my general state is confusion. Actually, it’s confusion both about the case and about the poster’s opinions (and, a bit, about your post).
The general impression I got was that all parties (except, probably, the victim) are guilty of something, including the prosecutors. But...
First, I’m not very sure exactly if your questions should be translated as “probability estimate that X killed Kercher” or not. I actually have a slight suspicion that your post might be a trick question around that. (In fact, writing now, I realize that the trial resulted in guilty verdicts on ~five counts for each, but after all the reading I have no idea what they are other than that one of each is some kind of murder.)
Regarding Knox and Sollecito, my “gut” impression is that they’re guilty of something, in the sense that I’d be quite surprised if they were just two innocent people who did absolutely nothing bad (other than that joint) and got dragged in a nightmare by chance. I also have a “gut” feeling that their “guilt” is correlated.
Regarding Guede, I’ve got a slightly stronger gut feeling of guilt (than I do for the first two).
I also feel (as in “gut”, again) that the prosecution and trial was “wrong”. (EDIT for clarification: but I can’t see any basis of estimating how wrong between “as good as reasonably possible” and “gross miscarriage of justice”.)
However, what confuses me most is the certainty of most estimates I’ve seen in the comments. I just don’t see how one can get much farther than “I don’t know, but I lean a bit towards” either way on what the posters report as their information. I mean, after only a couple hours or so of thinking (both during reading and writing this) I’m not even sure what priors might be appropriate. I feel I know so little about the case (including info like trials in general) that my response can’t be more than “I don’t know”. (If pressed a lot I’d output 0.5±e as a probability, simply because even my knowledge of probabilities is too weak to actually figure out if anything else would express my almost total uncertainty on this subject, and that’s my cached answer for binary questions.)
Put differently, given the information I have on the subject, it feels as if the confidence I should assign to any probability estimate should be very close to 0. As I said above, I tend to give probabilities as ~50% in those cases, but that’s not exactly a correct description of what I feel. A good description would be that, if forced to bet (without more time to think), I’d bet on all of them being guilty, but if given the option, I wouldn’t bet any amount at 1:1 odds.
Oh, and the answer to the fourth question is very similar. Leaving aside the fact that I’m not sure exactly how I’d quantify the “coincidence” of opinions, I’m wouldn’t put any significant confidence in any estimate. (The “gut” feeling is that you consider all three guilty, but that’s mostly a very vague impression I got from the wording of your text rather than reasoning on my part.)
As an aside, the interesting thing about the subject is that I was surprised how little confidence I assigned to my “estimates” in this case. I usually have some opinion on most questions I encounter, but in this case I find myself really unable to arrive at one.
No, “X killed Kercher” is right. (Or perhaps better would be, “X participated in killing of Kercher”.)
It’s an academical point right now, given your later post, but even “X participated in killing of Kercher” is not entirely clear. Any action A, from cutting her throat, to holding her, to opening the door, to simply being in the room, can be considered as “participating” in the murder, in the sense that suitably complicated scenarios can be constructed where X not doing the action A would have avoided the murder, with X knowing it and deliberately choosing to do A in order to cause Kercher not to live anymore.
I imagine what you actually wanted to ask was something like the probability that “X is guilty of whatever X was convicted”, though that only moves the uncertainty about what you ask to uncertainty about what Italian law says.