I vaguely recall that you got pretty annoyed at me a year or so ago when I pointed out a contradiction in your reasoning. I suspect that your anger at me over that incident is informing your commentary.
I’ve had no interaction with you on this site at all, but I have read your posts on the previous Amanda Knox threads, and while I believe I have a far greater aversion than wedrifid to making statements so likely to antagonize others, I have to say I find your judgment in this case in conjunction with your position as a lawyer downright frightening.
Judges are mostly selected from among lawyers, so that would be a lot more comforting if I were confident that the selection process were a genuinely good filter for people of exceptional judgment. But I would have a lot more trust in our justice system if I thought that lawyers tended to be people who would not readily become convinced of and argue strongly for positions in the absence of good reasons for believing them true.
At least in the United States, judges are mostly selected from among prosecutors. Defense attorneys, including public defenders, aren’t very well represented on the bench. General judgment aside, this a serious systemic bias of the system.
What mechanism makes it a problem? I can see why it would perhaps be most fair for an individual to have one of a former prosecutor and former public defender presiding over the individual’s original trial and appeal. The more judges make binding rules for the system, rather than in their courtroom, the more of a problem this would be (though I am under the impression this is not the case).
I see it as a problem in the sense that prosecutors and defenders are conditioned to view the judicial process in somewhat different ways, and I don’t think that judges can simply wave away years of conditioning to see things from a prosecutorial point of view.
Individual bias isn’t great either, but at least there the bias isn’t necessarily going to be in the same direction at each step of the process. The fact that the trial judge and all the appeals court judges are all likely to be former prosecutors, on the other hand, cuts the same direction at every step.
Individual bias isn’t great either, but at least...
I think this is a legitimate but small concern.
If half of all judges at each level were former prosecutors and half former defense attorneys, half of those who had a trial and one level of appeal would face the same bias at each step (assuming even promotion, and unless this was specifically corrected for).
A more fair system than evening out representation would be embracing a type of bias and having formal rules counteracting that, e.g. have all judges be former prosecutors and have a set of judicial ruels favoring defendants.
I think it likely you are being blinded to the gross unfairness of having an individual tried under a former defense attorney when the rules are written with no regard for bias or assuming the average judicial bias is in favor of the prosecutor, or under a former prosecutor even if half of all judges are former defense attorneys, by the legitimate point that fairness would be increased if each defendant had at most one of each type between trial and appeal.
This is especially true considering how much less important appeal is than an original trial.
Only mostly? I had assumed it was an actual legal requirement. That’s interesting. Where you come from how many judges have ever not been lawyers and how on earth do they know what they are doing?
EDIT: From the looks of it some (40) states in the US allow non-lawyers to be low level judges, usually for small towns doing straightforward cases. From what I can tell in Australia (and most comparable countries) a law qualification of some sort is required.
It’s rare now, but used to be common, for prospective lawyers to pursue an apprenticeship rather than a law degree. You can still use either as a qualification to take the bar exam.
That depends on the jurisdiction, and also on how exactly you define “lawyer.” (Do you mean someone with a law degree, or a member of the local bar association or some equivalent guild? The former is usually, but not quite always, a requirement for the latter.)
If anything, in many state and local jurisdictions within the U.S., judges are elected by popular vote, and I would guess that in some of those there are no such requirements for candidates, at least in theory.
Interestingly, for a U.S. Supreme Court appointment, a law degree from a top 14 law school (and at least one academic degree from Harvard or Yale) has been a de facto requirement for decades, but as recently as the 1930s and 1940s, there have been occasional SCOTUS justices appointed without a law degree at all.
Apparently the last United States Supreme Court Justice without a law qualification was Robert H. Jackson—although he passed the bar exam without official training and was a prominent practicing lawyer. I haven’t found the last time someone was appointed to that role without passing the bar but research so far does seem to suggest it is an entirely political position, without qualifications required.
I would have a lot more trust in our justice system if I thought that lawyers tended to be people who would not readily become convinced of and strongly for positions in the absence of good reasons for believing them true.
I see your point, but I suspect the problem is more in your own judgment than in mine. Consider that I have had the experience of being wrong on these sorts of issues—and having to face it—many many times.
In this case, knowing that you persist in your assignment of a high likelihood of guilt for Knox and Sollecito given the data that’s now available to you, I feel confident in saying that your ability to say oops is too poor.
After the advent of DNA testing, retrospective analysis of many crimes exonerated people who had previously been convicted. Had this case occurred before DNA testing was available, Knox and Sollecito would most likely have been convicted, but a DNA test should have been sufficient to exonerate them.
In this case, knowing that you persist in your assignment of a high likelihood of guilt for Knox and Sollecito given the data that’s now available to you, I feel confident in saying that your ability to say oops is too poor.
FWIW that assessment happens to be incorrect. I say “oops” regularly. And even in my professional life where there is a lot of money at stake, I occasionally have to withdraw a case and apologize. Or ask for leave to amend a paper.
Had this case occurred before DNA testing was available, Knox and Sollecito would most likely have been convicted, but a DNA test should have been sufficient to exonerate them.
Exactly what DNA evidence do you believe exonerates them? I am not aware of any, but I am happy to consider it in good faith. And yes, revise my probability estimates accordingly.
My understanding is that some doubt has been cast on the DNA evidence against the two. Which isn’t the same thing as being exonerated. But I really would like to hear about such evidence.
Exactly what DNA evidence do you believe exonerates them?
The DNA evidence showing that Guede was the killer.
This is exactly the sort of thing that gets people exonerated under the Innocence Project. The only difference is, in most cases, the authorities don’t usually go back to the scenedesperately looking for new evidence to incriminate the original suspect.
The DNA evidence showing that Guede was the killer.
This is exactly the sort of thing that gets people exonerated under the Innocence Project.
I completely disagree. As far as I know, Guede’s fingerprints were found at the scene of the crime. So that if years later, his DNA was found to have been there, it should not have much of an effect of peoples’ assessment of the case. It would be a surprise if his DNA were NOT found at the crime scene.
Wait, what? You’re saying that the case against Guede is even stronger than I implied, and yet you’re disagreeing with me?
The evidence against Guede—of whatever type—is what exonerates Knox and Sollecito. The more evidence against him, the more exonerated they are.
If, years later, DNA and fingerprints were found from someone other than the person convicted, that would make it even more likely for the conviction to be overturned.
The evidence against Guede—of whatever type—is what exonerates Knox and Sollecito.
Only if you substantially rule out the possibility that all 3 were involved in the murder.
ETA: Besides, the claim on the table seems to be that if you look at the case without DNA evidence; and then ADD DNA evidence, it would substantially reduce, basically to zero, the assessed chances that Knox and Sollecito were involved in the murder. Agreed?
ETA: Besides, the claim on the table seems to be that if you look at the case without DNA evidence; and then ADD DNA evidence, it would substantially reduce, basically to zero, the assessed chances that Knox and Sollecito were involved in the murder. Agreed?
No. Or at any rate, that’s not an interesting claim, so I don’t see why it should be on the table.
So you are saying that there are only two choices: (1) privileging the hypothesis; and (2) substantially ruling it out?
No. Or at any rate, that’s not an interesting claim,
If it’s not interesting to you, then probably you should not have tried to answer my question. Here is the claim which was made:
Had this case occurred before DNA testing was available, Knox and Sollecito would most likely have been convicted, but a DNA test should have been sufficient to exonerate them.
I asked this:
Exactly what DNA evidence do you believe exonerates them?
And you said this:
The DNA evidence showing that Guede was the killer.
Since Guede’s fingerprints were (apparently) found at the scene, extra DNA evidence which confirms his presence at the scene should not have much of an effect on anyone’s assessments of the likelihood of any conclusion.
So you are saying that there are only two choices: (1) privileging the hypothesis; and (2) substantially ruling it out?
In a word: yes.
The hypothesis in question has low prior probability, and hence is “substantially ruled out” by default, unless there is strong evidence pointing to it. In this case, there isn’t; people who think there is are, almost invariably, anchoring on the probability that they assigned (or that the investigators assigned) to Knox and Sollecito’s guilt before Guede was discovered.
If it’s not interesting to you, then probably you should not have tried to answer my question.
On the contrary, I inferred from your question (correctly) that you didn’t understand the point about the evidence against Guede exonerating Knox and Sollecito.
Here is the claim which was made:
Had this case occurred before DNA testing was available, Knox and Sollecito would most likely have been convicted, but a DNA test should have been sufficient to exonerate them.
[...]
Since Guede’s fingerprints were (apparently) found at the scene, extra DNA evidence which confirms his presence at the scene should not have much of an effect on anyone’s assessments of the likelihood of any conclusion.
Most likely, the person who made the above claim simply wasn’t aware of the fingerprint evidence (or, possibly, the difference between fingerprint evidence and DNA evidence). By the principle of charity, that shouldn’t have stopped you from understanding and addressing their real point, which was that Knox and Sollecito should have been dropped as suspects as soon as the physical evidence turned out to implicate Guede and not them.
Most likely, the person who made the above claim simply wasn’t aware of the fingerprint evidence (or, possibly, the difference between fingerprint evidence and DNA evidence).
I had, in fact, forgotten about the fingerprint evidence, but I stand by my claim that Knox and Sollecito would probably have been convicted prior to DNA testing, given that even with DNA testing they were only released on appeal, and with DNA testing we have a much higher burden of expectation of evidence to meet. To say that they probably would have been convicted does not and was not intended to mean that they should have been convicted given the state of evidence that would have been available in that case.
The hypothesis in question has low prior probability,
Prior to what? Prior to finding the evidence against Guede or prior to considering any evidence at all?
I inferred from your question (correctly) that you didn’t understand the point about the evidence against Guede exonerating Knox and Sollecito.
I do understand the point, I simply happen to disagree with you at the moment. Anyway, if you want to address a different issue from the one under discussion, it would be helpful if you made it clear from the beginning.
By the principle of charity, that shouldn’t have stopped you from understanding and addressing their real point, which was that Knox and Sollecito should have been dropped as suspects as soon as the physical evidence turned out to implicate Guede and not them.
I think that’s too much of a leap. Charity must be balanced with the principle that it’s often a mistake to try to put words in peoples’ mouths.
Each component of the report in which they retested the samples (knife and bra clasp) for biological evidence of Knox and Sollecito returned negative results, as detailed in the conclusion. Having found specific reasons to doubt the results of the Italian crime labs’ testing, the retest found no positive evidence associating Knox or Sollecito with the crime, and this is a case where absence of evidence is significant evidence of absence.
Each component of the report in which they retested the samples (knife and bra clasp) for biological evidence of Knox and Sollecito returned negative results,
Let’s make sure I understand your argument: You seem to be saying that the (apparent) lack of DNA evidence on the knife and bra clasp is convincing evidence that Knox and Sollecito were NOT involved in the murder.
There was an abundance of physical evidence of Rudy Guede on the scene. If Knox and Sollecito had any physical involvement with the crime, they should have left biological evidence on a similar order, but in fact tests for evidence of their involvement did not return positive results except when corrupted by serious mishandling and poor testing procedures. The knife and bra clasp were major pieces of the prosecution’s case because they alleged that they held genetic evidence implicating Knox and Sollecito, but in fact they did not, and even if they had carried evidence of Knox and Sollecito, it would have been a suspiciously small amount of evidence compared to what one would reasonably expect had they actually been involved.
Think of the allegation of Knox and Sollecito’s involvement in the murder as a claim that there is an elephant in a room. Another party examines the room and finds...
X: Well, there does seem to be an elephantish sort of smell in the room, but it’s otherwise devoid of large mammals.
Y: But you agree that it’s a distinctly elephantish sort of smell?
X: Yes, it does smell much more like an elephant than anything else I can think of.
Y: In that case, the presence of an elephant makes it far more likely for you to observe this smell than the non-presence of an elephant, therefore it meets the bayesian definition of evidence, so you must revise your belief in the presence of an elephant upwards!
X: But hiding an elephant in a room is very difficult. If there were actually an elephant in this room, I would expect a lot more evidence than that. I can’t think of any plausible way that you could get an elephant in here and hide every sign of its presence but its smell, only very implausible ones. Your claim was sufficient evidence to promote the hypothesis of an elephant in this room to my attention in the first place, so this is less evidence than I need to maintain my prior uncertainty. To maintain my prior uncertainty, I would need to observe evidence that gave significant reason for suspecting the presence of an elephant, such as this smell, while finding some sort of arrangement that could plausibly hide an elephant. This, on the other hand, is a small enough amount of evidence that I must revise my confidence to near certainty against.
It is reflective of a suspiciously poor case that the prosecution brought so little biological evidence against Knox and Sollecito compared to the one person we can be almost certain was actually involved. The pieces of evidence were mishandled, making them suspect to begin with. This provided an opportunity for us to make our beliefs pay rent. Given the state of the evidence against them, most of us on this site concluded that Knox and Sollecito were probably not involved in the murder, so the biological evidence brought forward by the police was probably not contingent on their presence at the crime, but due to subsequent corruption and mishandling of the evidence. Thus, we could predict that the independent experts’ review of the evidence would find that when properly processed, it did not point to Knox and Sollecito’s presence at the crime. If the evidence were contingent on their presence at the crime, we would not expect to observe this.
So do you concede that your high confidence in their involvement did not predict observations as well as our belief that they were probably not involved?
There was an abundance of physical evidence of Rudy Guede on the scene. If Knox and Sollecito had any physical involvement with the crime, they should have left biological evidence on a similar order.
Well I’ll consider this argument, but let’s make sure we are on the same page about biological evidence. According to one web site I found, Guede’s DNA was found (1) on a swab of Kercher’s privates; (2) mixed with Kercher’s blood on Kercher’s handbag and the left cuff of her sweatshirt; and (3) on toilet paper in one of the bathrooms in the house.
Do you agree with that?
So do you concede that your high confidence in their involvement did not predict observations as well as our belief that they were probably not involved?
Not yet as I am skeptical of your argument that involvement by Knox or Sollecito would have most likely resulted in the same kind of biological evidence as there was implicating Guede. But I’m willing to consider your argument and I admit that I have not heard it before today.
Let’s start by making sure we agree about the DNA evidence against Guede.
That reflects my understanding of the state of the DNA evidence against Guede.
Keep in mind that that does not imply that this is the sum of all DNA evidence Guede left on the scene; when you have already found this much biological evidence, there is no reason to continue searching in order to obtain a comprehensive inventory of every trace left by the suspect.
That reflects my understanding of the state of the DNA evidence against Guede.
Well, I’m not an expert on DNA evidence but just based on common sense, it seems to me that if playing the primary role in a murder and sexual assault reasonably leaves 2 DNA traces in a person’s blood stains a 1 trace in the person’s privates, it’s plausible that playing a secondary role might very well leave no DNA traces in the person’s blood and none in their privates.
Of course the DNA traces in other parts of the house are far less interesting since Knox and Sollecito can be expected to have left DNA biological evidence there regardless of whether they played some role in the murder.
So I remain skeptical of your argument but I am happy to consider an authoritative source which says that even someone who plays a secondary role in a murder with a knife is very likely to leave biological traces behind. Do you have such an authority? Or are you just going by your own general knowledge and common sense?
Keep in mind that that does not imply that this is the sum of all DNA evidence Guede left on the scene; when you have already found this much biological evidence, there is no reason to continue searching in order to obtain a comprehensive inventory of every trace left by the suspect.
I disagree. The authorities had every reason to continue searching for DNA evidence since they suspected others besides Guede.
Well, I’m not an expert on DNA evidence but just based on common sense, it seems to me that if playing the primary role in a murder and sexual assault reasonably leaves 2 DNA traces in a person’s blood stains a 1 trace in the person’s privates, it’s plausible that playing a secondary role might very well leave no DNA traces in the person’s blood and none in their privates.
The prosecution claimed that more than one person must have been involved in the murder because, they alleged, Kercher was stabbed more times than was likely by a single assailant, and the wounds were consistent with the use of more than one knife. However, the latter part has been challenged, and I am myself aware of murders known to be committed by a single person in which the victim was stabbed more times than Kercher was. The allegation was a particularly ignorant one in the first place. However, assuming that it was true that more than one person was involved in stabbing Kercher, then even if he was the only one who sexually assaulted her, then from my own knowledge of forensics I can say that it is extremely unlikely that only one person would have left biological traces on her unless everyone else was wearing extensive protection. Keep in mind that the contrast is not 3 traces versus zero traces, but three distinct samples which have considerably more than the threshold needed to be detectable by the forensic processes versus no traces above threshold level; if anyone else was involved in physically stabbing Ms. Kercher to death, as the prosecution claimed, they would have to have done so while leaving behind no traces within orders of magnitude of what Guede left.
I have studied enough of forensics and biological labwork to make more than a common sense guess in this matter, but if my say so isn’t enough for you, I could email a forensics specialist, although you could do the same. I’m not aware of any website that explains this, although that doesn’t mean that that there are none.
I disagree. The authorities had every reason to continue searching for DNA evidence since they suspected others besides Guede.
The traces of Guede’s DNA were acquired by a simple procedure; find a spot likely to yield biological evidence, take a sample, test it. A swab from Kercher’s genitals, for example, contained DNA from Guede. They did not test to see “let’s see how many swabs we can take which register positive for Guede’s DNA,” because it’s unnecessary.
You seem to be treating the distinct samples as if they were individual hairs, where a person could plausibly leave behind a few or none at all, but that simply isn’t an accurate model.
The prosecution claimed that more than one person must have been involved in the murder because, they alleged, Kercher was stabbed more times than was likely by a single assailant, and the wounds were consistent with the use of more than one knife.
None of that, of course, was the real reason. The real reason was that they had initially arrested Knox, Sollecito, and Lumumba on the basis of crackpot psychological theories, and had egg on their face after having to release Lumumba when he turned out to have an unimpeachable alibi; so they needed Knox and Sollecito to still be guilty in order not to look like the complete morons that they, in fact, were. The bottom line was already written, and it was motivated searching from that point on.
The prosecution claimed that more than one person must have been involved in the murder because, they alleged, Kercher was stabbed more times than was likely by a single assailant, and the wounds were consistent with the use of more than one knife.
For what it may be worth, I don’t think this is a completely accurate summary of the prosecution’s argument. For example, it seems that the prosecution argued that the lack of defensive wounds suggests multiple attackers. Also, the fact that she was (allegedly) stabbed on both sides of her neck.
but three distinct samples which have considerably more than the threshold needed to be detectable by the forensic processes versus no traces above threshold level;
How much above the threshhold were each of the 3 samples which matched Guede?
I have studied enough of forensics and biological labwork to make more than a common sense guess in this matter
Well what’s your job; educational background, etc?
find a spot likely to yield biological evidence, take a sample, test it
I’m not sure I understand your point. Do you agree with me that this sampling would have been done in batches? Do you agree that Knox and Sollecito were suspects almost from the very beginning? Do you agree that therefore the authorities would have had reason to continue collecting DNA evidence even after they found crime scene samples which clearly matched Guede?
For what it may be worth, I don’t think this is a completely accurate summary of the prosecution’s argument. For example, it seems that the prosecution argued that the lack of defensive wounds suggests multiple attackers. Also, the fact that she was (allegedly) stabbed on both sides of her neck.
Had she been physically restrained by anyone else, we should strongly expect that person to have left biological traces. Besides, it’s easy to stab someone in both sides of the neck without switching hands. Like the claim that she was stabbed too many times for one person to account for, this sounds a great deal like reaching to rationalize a story.
How much above the threshhold were each of the 3 samples which matched Guede?
I don’t know, although from the fact that all sources, including the Massei-Cristiani report, agree that there was an “abundance” of genetic evidence, I can infer that the traces were well above the limit of quantification, whereas the alleged genetic material from Sollecito, when tested by the police, was amplified by a procedure which increases sensitivity by more than an order of magnitude above the base level of detection. This places a ceiling for the amount of genetic material attributed to Sollecito of about one thirtieth of what was contained in any of the samples from Guede, but that’s assuming that Guede’s samples were only slightly over the limit of quantification. To be described as an “abundance”, there was probably many times more. And the independent experts’ report determined that the DNA from Sollecito, along with three other people on the sample, was probably due to corruption of the evidence well after the crime was committed.
Amanda Knox’s DNA was found in easily measurable levels on a knife that, independent report determined, had no traces of blood on it and could not have been used in the murder at all; it was improperly flagged as a piece of evidence in the first place.
Well what’s your job; educational background, etc?
I have a Bachelor’s in Environmental Science, currently pursuing credentials to become a high school chemistry teacher. In the meantime, I tutor.
I’m not sure I understand your point. Do you agree with me that this sampling would have been done in batches? Do you agree that Knox and Sollecito were suspects almost from the very beginning? Do you agree that therefore the authorities would have had reason to continue collecting DNA evidence even after they found crime scene samples which clearly matched Guede?
Yes, yes, no. The sampling was done in batches, but not in a comprehensive “let’s test everything in the room at once” sweep. Knox and Sollecito were suspects from near the beginning, but it was on the basis of a scenario that was quickly proven to be wrong. Once it was proven that the person who they concluded that Knox’s testimony had implicated could not possibly have been involved, they had practically no case for suspecting Knox and Sollecito at all. What they did have was a great deal of motivation to find evidence implicating Knox and Sollecito, because otherwise they would look and feel like idiots.
When they did the initial sampling and found multiple traces of evidence from Guede, they had already reached a state where it was unlikely that if someone else had been physically involved, they would have missed the evidence of them. They should have revised downwards from the prior likelihood that any randomly selected murder would be committed by multiple people. They had obtained evidence such that the hypothesis of Knox and Sollecito’s involvement should have been discarded, given that under interrogation they had failed to implicate the one person who was definitely involved, the other person “implicated” by their testimony had been conclusively proven not to be there, and they had not found biological evidence of their involvement under an initial search that would probably have found it if there were any. At this point, their initial scenario was effectively disproven beyond reasonable doubt, but still they insisted on continuing to search for evidence implicating two people they now had every reason to believe were innocent.
Had she been physically restrained by anyone else, we should strongly expect that person to have left biological traces. Besides, it’s easy to stab someone in both sides of the neck without switching hands. Like the claim that she was stabbed too many times for one person to account for, this sounds a great deal like reaching to rationalize a story.
So you agree then, that you mischaracterized the prosecution’s argument?
I don’t know, although from the fact that all sources, including the Massei-Cristiani report, agree that there was an “abundance” of genetic evidence, I can infer that the traces were well above the limit of quantification, whereas the alleged genetic material from Sollecito, when tested by the police, was amplified by a procedure which increases sensitivity by more than an order of magnitude above the base level of detection.
I’m not sure I understand your point. You seemed to be arguing that the amount of DNA traces left by Guede were “orders of magnitude” above the detection threshhold, i.e. at least a hundred times higher. And that if Knox or Sollecito had been in physical contact with Kercher, they would have left at least a hundredth of the DNA left by Guede—enough to have been unambiguously detected.
Did I misunderstand your argument?
What they did have was a great deal of motivation to find evidence implicating Knox and Sollecito, because otherwise they would look and feel like idiots.
Ok, and assuming that’s true, they would have had a strong motive to do more swabs right? And if they in fact did so, then under your hypothesis, they should have found additional Guede DNA material, right?
I have a Bachelor’s in Environmental Science, currently pursuing credentials to become a high school chemistry teacher. In the meantime, I tutor.
And what classes did you take which gave you your expertise in forensics?
So you agree then, that you mischaracterized the prosecution’s argument?
Not in any way that makes their case seem weaker than it actually was.
I’m not sure I understand your point. You seemed to be arguing that the amount of DNA traces left by Guede were “orders of magnitude” above the detection threshhold, i.e. at least a hundred times higher. And that if Knox or Sollecito had been in physical contact with Kercher, they would have left at least a hundredth of the DNA left by Guede—enough to have been unambiguously detected.
The traces left by Guede were all above the limit of quantification. The only piece of evidence on the scene that contained traces above the limit of detection for anyone other than Kercher or Guede was the bra clasp, which contained DNA from Kercher as the major contributor, and DNA above the limit of detection but below the limit of quantification for several other people, according to the independent experts’ report, DNA which they claim was most likely due to mishandling and contamination. That leaves no traces above the limit of detection from any uncontaminated source for any potential perpetrator other than Guede, so if we are to posit that anyone else was involved, we must suppose that they did it while shedding orders of magnitude less biological material.
Ok, and assuming that’s true, they would have had a strong motive to do more swabs right? And if they in fact did so, then under your hypothesis, they should have found additional Guede DNA material, right?
By the time they went to retrieve further evidence, Meredith Kercher had already been buried. Her body, along with the objects on which they had found traces of Guede’s DNA, had already been removed. They did not re-swab her body to see if they could get more of Guede’s DNA off of it, nor did they attempt to see if they could get more traces from any of the other articles on which they had already found his DNA. Had they scoured the scene for more traces of Guede, I would be surprised if they did not find any, but they weren’t looking for more traces of Guede, they were looking for traces of Knox and Sollecito. Further traces of Guede found on the scene 47 days later might not even have been kept due to their lesser value as evidence since there was already plenty of opportunity for corruption and the case against him was rock solid; if they found, say, small traces of his DNA on the floor on the reexamination, they’d have no particular reason to make note of it.
And what classes did you take which gave you your expertise in forensics?
Two semesters of biology labs covered forensics. My labs in atmospheric and environmental chemistry also dealt with collection and analysis of trace chemicals from a scene.
Not in any way that makes their case seem weaker than it actually was.
Umm, does that mean yes or no?
The traces left by Guede were all above the limit of quantification.
Umm, does that mean yes or no? And what are the numerical limits for quantification and detection?
By the time they went to retrieve further evidence, Meredith Kercher had already been buried.
Umm, does that mean yes or no?
Had they scoured the scene for more traces of Guede, I would be surprised if they did not find any, but they weren’t looking for more traces of Guede, they were looking for traces of Knox and Sollecito.
But when you “scour the scene,” there’s no way to narrow your search to DNA of a particular individual. Because the matching is determined back at the lab. Agree?
Further traces of Guede found on the scene 47 days later might not even have been kept due to their lesser value as evidence since there was already plenty of opportunity for corruption and the case against him was rock solid; if they found, say, small traces of his DNA on the floor on the reexamination, they’d have no particular reason to make note of it.
If you claim that there is likely other DNA evidence against Guede which was discarded, then I will need a cite. I have an extremely hard time believing that the lab would have tested a sample; realized it was Guede DNA; and not made a note of it in the record which would have come into the trial materials.
Two semesters of biology labs covered forensics. My labs in atmospheric and environmental chemistry also dealt with collection and analysis of trace chemicals from a scene.
I don’t think that’s enough for your opinion to simply be accepted on the basis of authority. Is there no reference online for your claim that secondary participation by Knox or Sollecito—such as grabbing or holding—would have left plenty of DNA material which the Italian authorities would have been very likely to find?
At a minimum, if what you are is saying is true, then surely one of the defense experts would have testified at trial that Knox could not have been a participant because none of her DNA traces were on or about Kercher’s person. Agreed?
It means that I do not strongly expect that there would be anything left on the scene which they would file as further evidence against Guede.
But when you “scour the scene,” there’s no way to narrow your search to DNA of a particular individual. Because the matching is determined back at the lab. Agree?
Yes and no. If they already have already established previously that person X has come into contact with objects X Y and Z, further testing of objects X Y and Z is likely to reveal further traces.
For the latter last two points, I do not have a strong expectation that further evidence against Guede would have been brought to court given that later retrieval would have been less reliable, and the case against him was already watertight, nor do I have a strong expectation that the defense would have claimed that Amanda Knox could not have been involved based on the weakness of the evidence of her involvement.
If we’re going to go on though, let’s review.
Meredith Kercher is killed. Knox and Sollecito are waiting outside when the police arrive to investigate the residence, and Edgargo Giobbi tags Amanda Knox as a suspect within hours of the crime’s report because he thought the way she swiveled her hips while putting on her protective shoes at the crime scene was unusual. Edgardo Giobbi claimed in a later interview
We were able to establish guilt by closely observing the suspect’s psychological and behavioural reaction during the interrogation. We don’t need to rely on other kinds of investigation.
Having established suspicion, Giobbi now had motivation to find more reasons to suspect her. On the basis of her mannerisms being judged unusual, she was brought in for interrogation, along with Sollecito, who she had claimed to be with that night. She alleges that she was badly mistreated, which the police deny. She was not told her legal rights, provided with a lawyer, or given an official interpreter. The police told Knox that Sollecito had changed his story and no longer said that he was with her that night at the presumed time of the murder. She was persuaded to sign a statement implicating her boss Lumumba as the killer.
After giving her statement implicating Lumumba, Amanda Knox was allowed a break to receive coffee and food, and then retracted her statement within hours, saying that she had been with Sollecito as she originally claimed, and had no knowledge of the killing.
Lumumba was arrested and held for two weeks, but produced a solid alibi. Crime labs matched the print found at the scene to Rudy Guede, police substitute Guede for Lumumba in their crime scenario, while retaining Knox and Sollecito as suspects, still purely on the basis of psychological profiling (keep in mind that they did not observe everybody else Kercher had known to see if anyone else also had suspicious mannerisms, with blinding to the hypothesis so the investigating police did not know whom they were supposed to find suspicious.)
Meanwhile, Guede, the only person with hard physical evidence placing him at the scene of the crime, fails to implicate either Knox or Sollecito in his testimony.
The police continued to search for evidence against Knox and Sollecito (and I will emphasize again, did not search to see if a comparable degree of evidence could be raised against anyone else Kercher had known) and produced a knife found in Sollecito’s kitchen, which they claimed contained traces of Kercher’s DNA despite her never having been in his kitchen. The evidence was improperly tested, and the independent experts’ report found that the knife could not have been used in a murder at all.
The police returned to the scene 47 days after the murder, and searched for items which could produce evidence against Knox and Sollecito. They claimed to have found evidence on a bra clasp which had been cut from Kercher in the attack. Since the date of the first investigation, the investigators had moved the clasp four feet across the room and left it under a rug with a collection of other items. The independent experts’ report found that the clasp predominantly contained traces from Kercher, but had small traces of DNA from four other people, one of whom was Sollecito. The report found that the evidence had been mishandled and probably corrupted and that the police did not follow proper testing procedures.
At this point, what probability do you assign to the proposition that Knox and Sollecito were involved in Kercher’s murder? Do you still think that focusing a search for evidence against Knox and Sollecito in particular, beyond everyone else who could potentially have been involved in the crime scene, after they and the only person with forensic evidence placing him at the scene had mutually failed to implicate each other, was reasonable?
This is interesting. But I can see someone seeing it as rude to speculate on the chance that someone else is going to change their mind. In some contexts this could come across as condescending and can easily make worse feelings that a discussion involves others who are opponents rather than people mutually trying to get a better understanding of reality.
With that disclaimer out of the way: I’m estimate an 8% chance that Brazil will reduce his estimate below 50% in the next year and publicly say so on Less Wrong.
I think you are the one who has a bit of a problem saying “oops.” Please just admit that you misstated the prosecution’s argument. I promise you the sky won’t fall as a result.
It means that I do not strongly expect that there would be anything left on the scene which they would file as further evidence against Guede.
I do not understand what your point is here. Maybe it will help if we quantify your argument:
First, what are the numerical thresholds involved for the amount of DNA required for detection and identification.
Second, how far above that threshold was Guede. Please answer the question with a rough percentage.
Third, what is the minimum amount of DNA one would reasonably expect to be left by Knox had she participated secondarily in the murder?
If they already have already established previously that person X has come into contact with objects X Y and Z, further testing of objects X Y and Z is likely to reveal further traces.
Further traces of Person X, right?
Anyway, please give me a cite for your (apparent) claim that other DNA evidence against Guede is likely to have been discarded. Or are you just speculating?
I have a strong expectation that the defense would have claimed that Amanda Knox could not have been involved based on the weakness of the evidence of her involvement.
Again, does this mean yes or no?
Look, you are presenting what is clearly a very solid argument for Knox’s innocence if your assumptions are correct.
Your argument is as follows:
(1) If Knox had been involved in the murder, there would have been DNA traces of Knox on or about Kercher’s person which the authorities would have been very likely to find.
(2) There were no such traces found.
(3) Therefore Knox was very likely NOT involved in the murder.
If (1) and (2) are true, (3) clearly follows. And there doesn’t seem to be any dispute about (2). Further, you assert (1) based on your expertise with forensic DNA testing. Surely Knox’s experts are at least as expert as you. So if (1) is true, then Knox’s experts would have almost certainly made the this point..
I think you are the one who has a bit of a problem saying “oops.” Please just admit that you misstated the prosecution’s argument. I promise you the sky won’t fall as a result.
I agree that the prosecution made claims that I did not relay. I also did not relay that Giobbi originally flagged Amanda Knox as a suspect based on what he considered to be unusual mannerisms while putting on her safety shoes. I am happy to admit that I misstated the prosecution’s argument, absent the implication that I omitted only information that strengthened their case. I do not agree that I understated the strength of the prosecution’s case.
I do not know how far above the threshold for detection of DNA Guede’s samples were; I can make inferences from the characterizations of the evidence, both from the prosecution and defense, that they were above the limit of quantification (which as a general rule is about three times higher than the limit of detection,) without using the same amplification procedure that was used on the bra clasp, which would further increase the sensitivity of that test. However, the reports which would give the quantitative values for the traces he left relative to the limit of detection, if they are available to the public at all, are in Italian.
I am prepared to accept if you do not consider my judgment in this case sufficiently authoritative; I can respect your suspicion on that count. We could suspend this discussion in order to contact a forensics expert and ask for an estimate of the likelihood that a person could be involved in physically grappling and stabbing a victim with a knife without leaving identifiable biological traces on the scene.
However, I think you would be guilty of motivated continuing, demanding evidence far in excess of what you need to alter your decision.
I affirm that I agree with statements 1, 2 and 3. However, I am not sure of the level of forensics expertise of Knox’s defense; the independent commision certainly possesses greater expertise in forensics than I, and they affirmed that there was no reason based on the forensic data to suspect Knox and Sollecito’s involvement. As for her defense, I know that they were able to take issue with the handling of the evidence by the police, but they could have done this by investigating “what are the procedures that a forensics specialist is supposed to take in such a situation, and did the police follow them? The case of the prosecution sets a low floor on how much familiarity I must assume that a trial team in a case dealing with forensics would have to have, so I have a high degree of uncertainty regarding the expertise of the defense. Further, despite being simply wrong, “absence of evidence is not evidence of absence” is a sufficiently pervasive truism (what do you call it when people treat something false as a truism?) that I do not have a very strong expectation that, even given a high level of expertise in forensics, the defense would present an absence of evidence as evidence of absence, rather than simply arguing that there was no reason to believe in their guilt (which the defense did.)
Will you provide your current judgment of the case based on the information I provided in my previous post? That information I can provide citations for, so we won’t have to stall waiting on further expert input.
Will you provide your current judgment of the case based on the information I provided in my previous post?
I have considered your argument in good faith and I reject it. You have not been able to answer simple, reasonable questions to let me understand and evaluate your argument. You have also evaded and ignored simple, reasonable questions.
Most importantly, you have not been able to provide an authoritative cite for your implicit claim that secondary involvement by Knox would have very likely resulted in Knox leaving DNA evidence which would very likely have been found. It’s also strange to me that I have not found prominent mention of your argument anywhere, especially since it would have been available to Knox’s defense (more specifically her experts) from the get-go.
In short, your argument has not stood up to mild scrutiny. However, if you can provide answers to my questions I will reconsider.
It is my general experience that people provide arguments of the form “if X, we should strongly expect evidence Y, we do not observe Y, therefore we should strongly disbelieve X” far less often than they are applicable, which is why I do not have a strong expectation that it would have been used in this case contingent on its applicability.
I do not evade the question of how much more of Guede’s DNA was found than is necessary to meet the detection limit, I affirm that I do not know, although I maintain that I have enough information to infer that it was much more than any other potential participants must have left.
At this point, I consider you to be evading the questions of whether the evidence you already have available to you supports a belief in Knox and Sollecito’s guilt, and whether you still feel that the police were justified in holding them as suspects after they received the evidence implicating Guede. I see three ways this discussion can go from here. We can suspend it, and write to an independent expert in forensics, and thus establish whether a person with acknowledged expertise in forensics would agree with my judgment. We can continue based on the information for which we already have adequate citations, and you can provide your judgment based on that information, or we can abandon it as fruitless.
It is my general experience that people provide arguments of the form “if X, we should strongly expect evidence Y, we do not observe Y, therefore we should strongly disbelieve X” far less often than they are applicable,
That’s not my experience. In fact, the pro-Knox page linked to in the very first thread explicitly makes such an argument in general terms (not specific to DNA).
At this point, I consider you to be evading the questions of whether the evidence you already have available to you supports a belief in Knox and Sollecito’s guilt,
Can you please quote the question I evaded?
I see three ways this discussion can go from here
It’s up to you. I’ve considered your argument in good faith and rejected it. But the door is still open if you want to provide the information and citations I requested.
At this point, what probability do you assign to the proposition that Knox and Sollecito were involved in Kercher’s murder? Do you still think that focusing a search for evidence against Knox and Sollecito in particular, beyond everyone else who could potentially have been involved in the crime scene, after they and the only person with forensic evidence placing him at the scene had mutually failed to implicate each other, was reasonable?
If you do not trust that I have sufficient knowledge to support the argument that the state of the biological evidence in this case more or less completely rules out the possibility of Knox and Sollecito’s involvement, I am prepared to accept your discounting that argument (although I find your belief that the defense would necessarily have used it if it were valid naive, if arguments of that form were used as commonly as they merit, “absence of evidence is not evidence of absence” would not have become a popular saying in the first place.) We can of course try to contact an independent expert, but is your openness to the likelihood of Knox and Sollecito’s innocence contingent on that particular proof?
I think it’s pretty clear that my assessment of the probabilities has not changed after having considered your argument, but let me make it perfectly clear: my assessment has not changed.
Do you still think that focusing a search for evidence against Knox and Sollecito in particular, beyond everyone else who could potentially have been involved in the crime scene, after they and the only person with forensic evidence placing him at the scene had mutually failed to implicate each other, was reasonable?
Yes.
but is your openness to the likelihood of Knox and Sollecito’s innocence contingent on that particular proof?
I’ve already stated that your argument is very solid if your assumption is correct. So I am conceding in advance that if you can support your assertion, my probability assessment will change a lot.
Can you give a numerical probability? You claimed that your assessment had shifted somewhat in light of the information that Knox and Sollecito had been found not guilty on appeal.
I’m going to do a run down of my probability estimates for Knox, Sollecito, and Guede, given the evidence available to me excluding the argument I’ve already given for the unlikelihood of Knox and Sollecito’s presence at the scene given the state of the forensic evidence. Could you explain what, if any of it, you disagree with, and how your own analysis would differ?
Prior probabilities. Common wisdom is that most people are killed by someone they know. I’m going to go with 80% likelihood of murder by an acquaintance, although I can’t actually find a single figure all sources agree on. So we’ll weight the prior probability of guilt such that people Kercher knows account for 80%, and everyone else accounts for 20%. Unfortunately, I can’t find a good source estimating how many people most people know, and it depends how close a standard of acquaintance you’re measuring. I’m going to give a ballpark estimate of about 1000, and thus apportion an average of .00125 prior likelihood of guilt between them.
For Knox, let’s adjust up for closeness of acquaintance (we’ll call it a factor of 10:1) for a likelihood of .0125. She had no prior arrest record; I’ll adjust down by 70 percent to .00875, and she was a woman who was neither a close family member nor an intimate partner, so I’ll adjust it down by another 90%, to .000875. I don’t feel that the prosecution established any good reason to suspect a motive, but I’ll be charitable to their case and not revise further down for that.
I’ll call Sollecito’s probability .0007, adjusting upwards for his increased likelihood as a male, but downwards based on his much lesser closeness to the victim; I have to take probability mass away from slight acquaintances if I’m adding it to close ones.
Guede was male, only slightly acquainted with Kercher, had been involved in breaking and entering prior to the case, and also did not have a clear motive. People with prior criminal records account for a majority of violent crimes despite being a minority of the population. The population without a criminal record is giving up a majority of their probability mass to the minority with, so this is a pretty significant boost. I don’t know the proportion of Italy’s population that has a record, but if we assume similar rates to the U.S. this will increase Guede’s likelihood by about 11 times. So let’s call his prior probability about .0077.
There’s a considerable amount of fudge factor in these numbers, and you are of course free to take issue with them, but I think they give us a reasonable ballpark order to start with.
First, we’ll deal with Guede. His handprint in blood places him conclusively at the scene of the crime (I doubt that anyone has ever left a handprint in blood at a scene at which they were not present) and he does not deny having been at the scene, so we’ll call the odds of his presence infinitesimally below 1. He changed his story repeatedly over the course of his interrogation, beginning with “another anonymous black man did it while I was in the bathroom after having sex with her.” The prior likelihood of this is very low, and Guede left DNA traces he was unlikely to have left if he was not involved in killing her. I’d adjust Guede up from the prior by an order of about 10^7, which would leave a likelihood of guilt that approximates to 1-(10^-5).
Let’s move on to Amanda Knox. Giobbi first flagged her as a suspect based on the belief that her mannerisms at the crime scene were odd. I expect that a person is significantly more likely to behave oddly at a crime scene (relative to the norm for an uninvolved person) if they were involved than if they were uninvolved. I’d put an odds ratio of maybe 15 to 1 on it, except that Knox was already known to have odd mannerisms (such as doing yoga stretches in places considered inappropriate,) so I’d revise that down to maybe 5 to 1. So that gives us a probability of involvement for Knox of about .0044. All the evidence brought forward by the police after this point is filtered evidence, because they focused their investigations on Knox and Sollecito, and put more effort into finding reasons to suspect them than anyone else who might have been involved.
In the United States in 1976-1996, 16% of murders were committed by more than one person; since I don’t have any better figures I’ll assume similar rates for present day Italy. Ignoring the failure of forensics to turn up evidence of anyone else, I find the arguments posed by the prosecution for believing that the murder had multiple perpetrators extremely uncompelling. Arguments such as the number of stab wounds and the fact that Kercher had wounds on both sides of her neck are extremely weak evidence; particularly taken alongside claims such as one which they offered for the break-in being staged, that the pattern of the broken glass was inconsistent with a break from outside, which was tested and found to be false, these arguments sound to me like scrambling for anything that sounds like evidence. Ignoring the failure of anyone else to leave biological traces, I’d be willing to extend perhaps a 3 to 1 odds ratio on their evidence for multiple attackers, so together with the prior unlikelihood of a murder having multiple perpetrators, that brings Knox’s likelihood of guilt to about .0022.
Knox and Sollecito failed to implicate Guede, and Guede failed to implicate Knox and Sollecito; for their mutual failure to implicate each other, I’d reduce Knox and Sollecito’s likelihood about 20 times, bringing Knox to about .00011.
I do not find any evidence presented by the prosecution compelling with regards to establishing a motive or likelihood of presence at the crime for Knox or Sollecito, their case seems to me to be a collection of behaviors and actions attributed to Knox with the implication “this seems more unusual for a person who was not involved than a person who was.” Since this evidence is filtered, I only find it reasonable to update on it to the extent that I believe the body of evidence is stronger than what I should expect for anyone with my current assignment of guilt subject to a similar police investigation. I think if you really looked and sincerely believed you had reason to find anything, you could find a substantial body of reasons to be suspicious of anyone. One of the most common ways that people err is in believing that a false proposition must have no valid evidence, and every argument for increasing your confidence in it must be refutable. I would have to assume that people I know are generally significantly weirder and more suspicious than most of the population to assign an odds ratio of even 10 to 1 on the prosecution’s evidence, but my social circle probably is fairly odd, so I’ll give them that, but I think I’m being generous.
All in all, that gives approximately .001 odds of guilt for Knox. Sollecito was tagged as a suspect mainly through association with her, so I would say that he is at most as likely as she is to have been involved.
Of course, there is plenty of uncertainty in these figures, and this does not portray the entirety of the relevant evidence even ignoring the improbability of other people involved in the murder not leaving biological traces (one particular piece of evidence I even left out deliberately, you can try and guess what,) but I think it reflects a reasonable probability estimate given that omission for an amount of evidence that a person can reasonably be expected to process without a substantial time commitment.
Could you explain what, if any of it, you disagree with, and how your own analysis would differ?
I don’t have the energy to do a point by point response, but I will point out what I think is the central flaw in your reasoning.
The flaw, it seems to me, is that you seem to dismiss the circumstantial evidence against Knox (and Sollecito) as “filtered evidence.” There are two problems with this.
First, the evidence is not filtered in the sense that the only evidence we (or the jury) saw is that which was cherry-picked by the prosecution. Knox and Sollecito both had the opportunity to gather and present evidence which would tend to exculpate them. To be sure, they did not have the same ability to get cooperation and compel the production of evidence as the authorities, but they surely had quite a lot. And they had the advantage over the authorities of being firsthand witnesses to their own whereabouts and actions.
Second, even filtered evidence can be very probative. To take Eliezer’s hypothetical a step further, imagine that a particular coin is flipped exactly 1000 times and our hypothetical advocate tells us that between 641 and 1000 of the flips resulted in heads. Even though we do not know how the actual data was filtered, and even though our advocate probably has some kind of agenda, we can be pretty confident that the coin is biased.
If we know (or can make a good guess at) the advocate’s agenda, we can get an even better handle on the situation. For example, if somebody tells you that he attends a Top 6 law school, it’s pretty likely that he goes to Columbia, NYU, or Chicago. And not Harvard Yale or Stanford.
How is your probability affected by what former FBI agent Steve Moore has to say?
It’s not, since Moore obviously is trying to spin the case in favor of Knox’s position. For example, he falsely asserts that there was not evidence of Knox’s presence at or near the crime scene. So I don’t put a lot of stock in his authority.
Since he is a qualified expert on forensics making essentially the same claim that I was about the state of forensic evidence, I’m curious to know what you would consider to be adequate corroboration for that point.
Why do you condition strongly on the prosecution’s arguments, when their evidence has been found in many points to be false, mishandled, or exaggerated, but not condition at all on the arguments of Moore?
Why do you condition strongly on the prosecution’s arguments, when their evidence has been found in many points to be false, mishandled, or exaggerated, but not condition at all on the arguments of Moore?
I’m not sure what you mean by “condition on,” but I would be similarly skeptical (probably even more skeptical) if the prosecution assured us that it’s perfectly plausible for the secondary participant in a murder such as this not to have had DNA traces found on or about the victim’s person. Even if the assertion were made on behalf of the prosecution by a board certified forensic investigator.
By “condition on,” I mean “update your probability in light of.” To refuse to update at all, you must suppose that it is no more likely that Moore would claim that the state of the evidence effectively rules out Knox and Sollecito’s involvement if they were not involved than if they were.
So it seems that you are claiming that it is much more likely that the prosecution would produce the evidence and arguments that they did if Knox and Guede were guilty than if they were innocent (enough to multiply the likelihood of their guilt by orders of magnitude,) but no more likely that Moore, who has neither pay nor reputation riding on the outcome of the case, would claim that there is dramatically less evidence than we would expect to see in the case of Knox and Sollecito’s involvement, if they were not involved than if they were.
Would you agree with that?
You gave a probability estimate of 90% back in Komponisto’s first thread, and early in this one you said that you had revised your confidence on the fact that Knox and Sollecito had been found innocent on appeal, but now you are asserting exactly the same probability that you did back then. Why is that?
Also, I would appreciate it if you would answer my question about what you would consider to be adequate corroboration for my argument that if Knox and Sollecito had been involved we would strongly expect forensic evidence that we did not see.
Assuming that the various evidence I am considering comes from the prosecution, yes.
You gave a probability estimate of 90% back in Komponisto’s first thread, and early in this one you said that you had revised your confidence on the fact that Knox and Sollecito had been found innocent on appeal, but now you are asserting exactly the same probability that you did back then. Why is that?
Here’s what I said early in the thread:
The main evidence was that the appeals court in Italy reversed the convictions of Knox and Sollecito. This undermined my confidence a bit. On the other hand, arguing that Knox and Sollecito were involved in the murder has made me a bit more confident in my beliefs. I hope that this is because arguing has given me the opportunity to think more carefully about the case, but it may also be the false confidence which comes from emotional investment in a position. Either way, I doubt it’s made much of a difference since I was pretty confident from the beginning.
My probability estimate remains roughly 90% and I think that’s pretty clear.
Ok, now a question for you: Why are you so eager to find a contradiction in my posts?
Also, I would appreciate it if you would answer my question about what you would consider to be adequate corroboration for my argument that if Knox and Sollecito had been involved we would strongly expect forensic evidence that we did not see.
First of all, please do not weasel, i.e. change your position without acknowledging it. Your original point was about DNA evidence.
Anyway, what would support your argument is if you can produce a quote from a textbook, police manual, or other authoritative source which says, more or less, and in general terms, that even physically restraining another person will almost always leave biological traces which can be traced back to the person who did the grabbing or holding.
Since he is a qualified expert on forensics making essentially the same claim that I was about the state of forensic evidence, I’m curious to know what you would consider to be adequate corroboration for that point.
I’m not sure it’s the same claim. As I recall, your claim was limited to traces of Knox and Sollecito on or about Kercher’s person. Moore’s point is more general.
But anyway, what I had in mind was a statement which is not specific to the case. For example, a quote from a forensics textbook or a police manual.
Yes, Moore’s point is more general than mine, he points to additional evidence that it had not occurred to me that we would expect. Moore’s point is stronger than mine.
I do not have a forensics textbook or police manual, but Moore probably does. If I ask him for a citation, and he provides one, would you consider that adequate?
Yes, Moore’s point is more general than mine, he points to additional evidence that it had not occurred to me that we would expect. Moore’s point is stronger than mine.
Moore’s point is stronger if his assumptions are correct.
I do not have a forensics textbook or police manual, but Moore probably does. If I ask him for a citation, and he provides one, would you consider that adequate?
If it says more or less, and in general terms, that even physically restraining another person will almost always leave biological traces which can be traced back to the person who did the grabbing or holding.
And by the way, if you really plan to research this, what would also be helpful to me would be a cite for the proposition that the Italian forensics labs may have (or did) find additional material attributable to Guede which did not make it into the record.
What sort of odds ratio would you assign to the evidence presented by the prosecution given the knowledge that it’s filtered? If your other figures are not similar to mine, it would need to be more than two orders of magnitude higher than what I assign, or about 3,000 to 1.
My low weighting of the prosecution’s evidence hinges, not simply on the fact that it is filtered, but that it is not much stronger than the evidence that I would expect them to find against anyone else given a similar level of investigation. So it is not analogous to dismissal of the evidence that “coin flips between 641 and 1000 were heads” on the basis that while true, it is filtered and so can be discounted. It is more analogous to receiving the evidence “at least six of ten coin flips came up heads,” and deciding “If you had stronger evidence than that, you probably would have told me, and that is not much more evidence than I would expect to see from an average unbiased coin.”
The defense extensively pointed out the weakness of the evidence brought forward by the prosecution, they did not put similar resources to the prosecution in order to raise similar evidence against anyone else to show how easy it would be. If people sincerely want to produce reasons for believing something, they will tend to both find and overweight them. I would bet at 1:1 odds that had the prosecution investigated me with a similar level of prior suspicion, they would have found as much or more reason to suspect me, and I may be atypical, but I do not think I am 1 in 3,000 suspicious. Conditioned on the facts that Knox was already known to have odd mannerisms, and was a foreigner from the perspective of the police (one of the original pieces of evidence that the police used to narrow her down as a suspect, for instance, was that she texted “see you later” to her boss, whom she was not about to see,) I think you will need to provide some very strong arguments to convince me that you are not dramatically overweighting.
What sort of odds ratio would you assign to the evidence presented by the prosecution given the knowledge that it’s filtered?
I’m not sure what you mean by “odds ratio” and I don’t know exactly what evidence was presented by the prosecution, so it’s hard to say. My assessment of 90% is based on the evidence I am aware of.
but that it is not much stronger than the evidence that I would expect them to find against anyone else given a similar level of investigation
I disagree. I shared some of my arguments about my interpretation of the evidence in the last couple threads.
If you accepted my assertion that an appropriate assignment of guilt not accounting for the lack of forensic evidence is about 0.1%, you would adjust your estimate to 30%?
No, if I accepted your assertion that even grabbing or holding someone is very likely to leave biological traces which can be identified by the authorites, then I would adjust my estimate down to 30%.
Can you explain why? That sounds like a claim that makes no sense in terms of probability theory, and I suspect that you either do not know enough about calculating probabilities to understand why it doesn’t make sense, or you are failing to apply that understanding.
We often make inferences like that implicitly, even here, revising our beliefs much less in light of new information than actually makes sense. One of our weaknesses as humans is that we will tend to stay much closer to the first number that comes into our heads than makes sense on further information. That’s why will will sometimes sanity-check our estimates by going through a numerical breakdown of what we know, like I did earlier. It gives a chance to look and see “Hey, I assigned a probability of 20%, but when I plug in the most reasonable numbers I can come up with and try and calculate a probability, I come up with .03%! To believe that the probability is really 20%, I would have to guess that this piece of evidence is a million times more probable if the belief is true than if it’s false, and that seems like way too much, or that the odds of this particular other thing happening are orders of magnitude off from what this study says, or some other combination of things that all look really improbable in light of the information I have! It looks like my first instinct was not just wrong, but very wrong, so I’d better change it by a lot.”
Before I put time and effort into providing more evidence, I think it’s important to establish that you actually know how to process evidence in a way that makes sense mathematically, not just intuitively. You can you go through a rundown of how you arrived at your numbers? I know you said before that you didn’t have the energy, but I think it will save a lot more time and energy in this conversation if you do it than if I go on trying to retrieve information for you while expecting you to update not at all or far less than makes mathematical sense.
You sure? 90% to 30% isn’t really an orders of magnitude difference. Of course, taking that into account, “beyond a reasonable doubt” should be something like “at least 99.9% probability of guilt” (or whatever incidence of innocent people you’re willing to throw in prison / execute), in which case a shift from 90% to 30% would be a shift from “not guilty” to “slightly more not guilty”.
I don’t understand what you’re asking if I’m sure about. I’m saying that given the premise that brazil84 accepts the forensic argument against Knox and Sollecito’s guilt, it does not make sense to update his probability by so little.
He already stated that given that he accepts the argument, he would update to 30% confidence, which conveniently frees us from the need to consider the probability outside the argument.
It’s possible that he really meant “given that I accept the argument at low confidence, as opposed to not at all, as I currently do,” but that would make his statement awfully misleading, and brazil84 has demonstrated too little familiarity with Bayesian reasoning for me to read his statements with that much charity.
He should mean “given that I assign the argument a comparable credence to other things I would state that I accept”—which shouldn’t be 1. I think this probably shouldn’t be a big enough difference to matter (I missed the “given...” in the context, in my response), but it would be interesting to update the numbers to verify.
Can you explain why? That sounds like a claim that makes no sense in terms of probability theory, and I suspect that you either do not know enough about calculating probabilities to understand why it doesn’t make sense, or you are failing to apply that understanding.
I could try, but first please explain to me what does not make sense about my statement. If I start at 90% based on other evidence, where should I end up (in your view) if your assertion about biological traces is established, and why?
ETA: Better yet, why don’t you provide your calculation which starts at 90% and ends up at some number very different from 30%.
9/.43 is 20.9. So if you have proposition A, which you believe with 90% confidence, but then you get some piece of evidence which Not-X makes 20.9 times more likely than X does, you should reduce your confidence to 30%. If you had evidence that Not-X made a thousand times more likely than X, then that should reduce your odds to .009, which is equivalent to a probability of slightly over .009, or .9%
So we can work backwards and figure that, given the premise that you accepted the piece of evidence and adjusted your probability from 90% to 30%, then if you did the right sort of calculations you must have decided that it was a bit more than twenty times more likely that Knox and Sollecito would have failed to leave behind the forensic evidence that would implicate them given their innocence than given their guilt.
Considering the number of pieces of evidence that Moore claims that they would almost certainly have left if they were guilty, which they did not leave, I don’t think that passes a basic sanity check.
Converting to log-odds, for the practice reading log-odds units...
In bits, we have:
90% initial confidence = lg(90%/10%), which is about 3.2 bits likely
30% final confidence = lg(30%/70%), which is about −1.2 bits likely, or 1.2 bits unlikely
total evidence needed to shift our view that far: 4.4 bits
In decibels:
90% initial confidence = 10 log(90%/10%), which is about 9.5 decibels likely
30% initial confidence = 10 log(30%/70%), which is about −3.7 decibels likely, or 3.7 decibels unlikely
total evidence needed to shift our view that far: 13.2 decibels of evidence
Could you please just answer my question? I think it’s very simple and reasonable.
(1) In your view, if I start at 90% based on the other evidence; and your assertion about biological traces is established, then where should I end up? 5%? 1%? 88%? 0.01%? It’s a very simple question.
I opened my response box before you edited your comment to include the question (I actually left the house in the meantime.) I think that you do not know the necessary math well enough to believe that you can defend your assertion, and so are trying to evade the question by trying to make me answer further questions and hoping that you will find something you can take issue with.
Nevertheless, I will show you a calculation. I will use what I believe represents a reasonable lower limit on the weight one could assign to the evidence. I will not respond to requests to cite unless you agree to provide citation indicating that the weight implicit in your own estimate is more reasonable.
We start with a probability of .90. I will assume that murderers leave forensic evidence with 99.9% frequency, and non-murderers have false positives with .1% frequency.
Out of a population of 10,000 people at this confidence level of guilt, we will expect 9000 murderers and 1000 innocent people. We’ll assign a 99.9% confidence that a person who is not guilty will not leave forensic evidence implicating them (with .1% chance of false positive) and a 0.1% chance that a person who is guilty will not leave forensic evidence implicating them.
Out of 9000 murderers, we will expect 9 to be false negatives (not leave forensic evidence)
Out of 1000 non-murderers, we will expect 999 not to leave forensic evidence, with one false positive.
That leaves 1008 people who have not left forensic evidence, 999 of whom are innocent. So if a person has not left forensic evidence, their odds of guilt are 1-(999/1008), or about 0.9%
Edit: I actually made an embarrassing mistake in my calculations the first time around, and overestimated the posterior probability. Results now corrected.
To make myself clear, this is not the weight I assign to the evidence, I am saying that it would be unreasonable to assign less weight than this.
I opened my response box before you edited your comment to include the question
As I recall, the original part of my post specifically asked you to give me the probability I should end up with. The addition merely asked you to provide your calculation. You provided neither.
Anyway, the main flaw in your reasoning consists in conflating the propositions that (1) Knox (and Sollecito) were physically involved in the murder (i.e. holding, grabbing, and or stabbing); and (2) that Knox and Sollecito were guilty. For example, it’s possible that Knox and Sollecito encouraged Guede from a few feet away to attack Kercher but never actually touched her. And perhaps did not even intend for him to kill her.
Doing the same calculation as you, but substituting 95% for 99.9% (actually you seem to have used 99% in your calculation) gives 450 false negatives and 999 correct negatives. So out of 1449 people who have not left biological traces on or about Kercher’s person, 450 are actually guilty. ETA: 31%.
Anyway, I still look forward to your cites that (1) physically participating in a knife murder (i.e. holding, grabbing or stabbing) is extremely likely to result in biological traces being left on or about the victim’s person; and (2) the Italian authorities may have (or did) find additional traces from Guede on or about Kercher’s person which did not end up in the record.
Yes, I already stated before that an odds ratio of approximately 20:1 would result in a posterior probability of 30%. I have also asserted that this weight does not pass a basic sanity check. You have stated that if you accept the argument from forensic evidence, you would assign a probability of 30%. Moore, who is a forensics expert, judges that for Knox and Sollecito to have been involved in the crime, based on the state of the evidence, is nigh impossible. 5% likelihood is not a sane estimation.
I am extremely tired of looking forward to how you will attempt to evade each of my arguments and ask me to do more work in exchange for promises of “good faith consideration” which amount to looking for any possible excuse not to update. As I stated in my previous comment, I will contact an expert for citations if you agree to provide a citation for the claim that 5% likelihood of failure to leave forensic evidence is a credible assignment. One of us is asserting a confidence many orders of magnitude apart from a qualified expert,and one of us has been repeatedly demanding further evidence for the other’s confidence assertion, and in a reasonable argument, these would not be the actions of two separate people.
You have stated that if you accept the argument from forensic evidence, you would assign a probability of 30%
I’m not sure what you mean by “the argument from forensic evidence,” but let me ask you this: Do you understand that the probability that Knox (and Sollecito) were physically involved in the murder (i.e. holding, grabbing or stabbing) is not necessarily the same as the probability that they were involved in the murder?
I am extremely tired of looking forward to how you will attempt to evade each of my arguments
Sorry, but you are the one who is being evasive. I explained to you EXACTLY what the flaw was in your argument, and you ignored it.
The probability that Knox (and Sollecito) were involved in the murder is necessarily greater that the probability that they were physically involved. Agreed?
And do you agree that if they were involved in the murder but did not hold, grab, or stab Kercher, then there is a good chance that they would not have left biological traces on or about Kercher’s person?
These are reasonable, yes or no questions. Please do not evade them.
I’m not sure what you mean by “the argument from forensic evidence,” but let me ask you this: Do you understand that the probability that Knox (and Sollecito) were physically involved in the murder (i.e. holding, grabbing or stabbing) is not necessarily the same as the probability that they were involved in the murder?
Yes, but Moore also points out forensic evidence that we should expect to see if they were simply in the room at the time of the murder which we do not see. I agree that if they were involved in the murder, did not physically engage with Kercher, there is a good chance they would not have left biological traces on her person. However, if they were physically in the room at the time of the murder, there is still forensic evidence which we should expect but do not see. If we assume that they were conspirators in her murder, the forensic evidence doesn’t rule it out, but the prosecution did not allege that Knox and Sollecito conspired with Guede but were not physically present, because they do not have the evidence of communication between them to support it.
You are consistently refusing to stick your neck out and engage with questions that will expose your analysis of the case to criticism. I have just answered your questions, yes I agree that if Knox and Sollecito did not physically engage with Kercher, we should not expect traces on her, but we still have an expectation of evidence around the room, and yes they could have been involved through conspiracy without being present and leaving physical evidence, but we don’t have the evidence to support that either.
Now I am going to ask you to provide a rundown of your numbers, as I did myself, and I asked you to earlier, but you avoided with further questions. If you ask further questions without providing one, I am going to have to write you off as too logically rude to engage with.
Yes, but Moore also points out forensic evidence that we should expect to see if they were simply in the room at the time of the murder which we do not see
That’s a different issue. If you want to abandon your original claim, fine but you should acknowledge it. The question I answered was how my estimation of the chances of Knox’s (and Sollecito’s) guilt would change if I were very confident that physical participation in the murder would have left DNA traces of Knox (or Sollecito) on or about Kercher’s person.
That’s the question I answered. If you want to ask me a different question, fine, but please stop pretending that we are talking about a different question.
I have my own rules of debate, and I don’t engage with people who strawman me, i.e. people who insist on attributing to me positions which I did not take. So please knock it off.
Anyway, Moore’s assumption is that Knox and/or Sollecito were physically involved in the murder. Re-read his piece.
However, if they were physically in the room at the time of the murder, there is still forensic evidence which we should expect but do not see
Exactly what? If Knox and Sollecito were standing just inside the doorway, encouraging Guede to attack Kercher, and then turned and ran out when he started stabbing her, what evidence would we expect to see? Certainly not bloody footprints of Knox or Sollecito.
Now I am going to ask you to provide a rundown of your numbers, as I did myself, and I asked you to earlier
I already provided my calculation, changing your 99% to 95% to reflect the possibility that Knox and/or Sollecito were guilty but did not do anything like hold grab or stab Kercher. I think that there is a 5 to 10% chance that Knox and Sollecito were guilty but did not physically participate.
And again my question:
The probability that Knox (and Sollecito) were involved in the murder is necessarily greater that the probability that they were physically involved.
You are consistently refusing to stick your neck out and engage with questions that will expose your analysis of the case to criticism
Please quote 3 questions which you think I have avoided and I will try to answer them.
And please do not ignore or evade my question above:
If Knox and Sollecito were standing just inside the doorway, encouraging Guede to attack Kercher, and then turned and ran out when he started stabbing her, what evidence would we expect to see?
ETA: By the way, I totally concede that if Moore’s assumptions are correct then the probability that the prosecution’s scenario is correct is close to zero. I also concede that if your assertion about biological traces is correct, then the probability that the prosecution’s scenario is correct is close to zero.
I already provided my calculation, changing your 99% to 95% to reflect the possibility that Knox and/or Sollecito were guilty but did not do anything like hold grab or stab Kercher. I think that there is a 5 to 10% chance that Knox and Sollecito were guilty but did not physically participate.
What I asked for is a run down of your numbers leading to your estimation of your probability of guilt, such as I provided here.
Please quote 3 questions which you think I have avoided and I will try to answer them.
I did not ask for your calculation of the weight of the forensic evidence, simply substituting in another number for mine, I calculated for you what weight you would have to assign the evidence, and I asked you why you thought that was a reasonable probability, and you responded with more questions instead of attempting to provide an answer.
As I stated in my previous comment, I will answer further questions from you if you provide a rundown of your numbers that form the basis for your probability estimate. This would be a simple request for you to fulfill, and would save time and energy over the course of this conversation. Take as long as you want, it’s not going to take as long as this conversation will if you hedge around.
What I asked for is a run down of your numbers leading to your estimation of your probability of guilt, such as I provided here.
It didn’t seem to me that you were doing that. My impression was that you were asking how I got from 90% to 30%; not how I got to 90% in the first place. The topic under discussion was whether my probability estimate had changed in light of various information and it looks to me like you are trying to play a gotcha game.
Anyway, I will give you a kind of run down with the understanding that if you ignore or evade any of my questions from now on, we are done.
The numbers I am giving you represent a rough approximation of the intuition in my thinking. At this point, 90% is not the result of an explicit mathematical calculation. Anyway, in my opinion, the prior probability that Knox is guilty is pretty low (i.e before one looks at the various circumstantial and forensic evidence). On the one hand, it’s extremely rare for someone of her demographic to take part in a murder; on the other hand it’s pretty common that when somebody is murdered it was someone in their household. On balance, perhaps about 1 in 500 is my prior estimate of the probability of Knox’s guilt. At the same time, the chances that an innocent person would have this kind of circumstantial evidence against them (changing alibis, staged break-in, forgotten phone calls, and so forth) is much lower. Perhaps 1 in 5000. (This is including the exculpatory aspects of the evidence as well.) Of course a guilty person would be very likely to have this kind of evidence against them. So out of 100,000 hypothetical Knoxes, 200 are guilty and roughly 190 have this kind of circumstantial evidence against them. At the same time, 95,000 are innocent and about 20 of those innocent Knoxes have this kind of circumstantial evidence against them. So of 210 people who have this kind of circumstantial and forensic evidence against them, roughly 90% are guilty.
A source for your probability assignment of the strength of the forensic evidence.
I’m not sure what you are asking for here, but the source of my estimate of the strength of the forensic evidence (taking Moore’s assertions as true) is the observation that Knox (and Sollecito) may have been involved in the murder but not in a way which would have put them close enough to Kercher to leave the sort of evidence Moore discusses. I have estimated this probability at about 5 to 10%. i.e. I think there is a 5 to 10% chance that IF Knox and Sollecito were involved in the murder, they were involved from at least a few feet away.
For example, it’s possible that Knox and Sollecito were encouraging Guede from just inside Kercher’s door, and turned and ran away shortly after he started stabbing her.
So here’s a couple questions for you. (And if you ignore or evade them, we are done.)
Do you agree that in the scenario I described where Knox and Sollecito encourage Guede and then run, it’s pretty likely that they would not have left any DNA traces on or about her [ETA: Kercher’s] person?
Do you agree that in this scenario, it’s pretty likely that Knox and Sollecito would not have left the kind of evidence described by Moore in his article?
Moore, who is a forensics expert, judges that for Knox and Sollecito to have been involved in the crime, based on the state of the evidence, is nigh impossible.
That’s not what Moore says. His reasoning is based on the assumption that if they were involved, they were physically involved. Read his article again.
An explanation for why you revise your probability estimate strongly on the arguments of the prosecution, who have a vested interest for arguing Knox and Sollecito’s guilt, when their evidence has been found in many points to be false, exaggerated or mishandled, but do not revise your probability estimate at all on the arguments of Moore, who in addition to being a qualified expert, does not have money or reputation riding on the outcome of the case
I am not sure what you mean by “arguments,” since my probability estimate is based mainly on the evidence. Evidence is not the same thing as arguments. And neither is the same thing as an opinion. I am not basing my probability estimate on the prosecution’s arguments.
Moore’s actual argument is very solid. The problem is his opinion. I am skeptical of his opinion since he has an agenda. I am also skeptical of the prosecution’s opinions.
And another two questions:
The probability that Knox (and Sollecito) were involved in the murder is necessarily greater that the probability that they were physically involved. Agreed?
And do you agree that if they were involved in the murder but did not hold, grab, or stab Kercher, then there is a good chance that they would not have left biological traces on or about Kercher’s person?
It didn’t seem to me that you were doing that. My impression was that you were asking how I got from 90% to 30%; not how I got to 90% in the first place. The topic under discussion was whether my probability estimate had changed in light of various information and it looks to me like you are trying to play a gotcha game.
I wouldn’t call it a gotcha game, but I was of course expecting to find something in your probability analysis that I would object to. My primary dispute has all along been that I think your probability estimate of Knox and Sollecito’s guilt is much too high. If our posteriors disagree, then we have an implicit disagreement about what our priors should be and/or how much we should update on the available evidence. If we do a breakdown of our probability analyses, then we can determine the source of our disagreement.
At the same time, the chances that an innocent person would have this kind of circumstantial evidence against them (changing alibis, staged break-in, forgotten phone calls, and so forth)
The “staged break in” claims have been more or less discredited; all the evidence thus far has been found consistent with a real break-in. Unless you’re referring to some argument presented by the prosecution that I haven’t been able to find, the “forgotten phone calls” refers to a three second phone call from Knox to Kercher, which I do not think accounts for any weight as evidence at all, as I have argued here. The “changing alibis” was done,according to Knox, under extreme duress; she claims that she was treated abusively during interrogation, and encouraged to implicate Lumumba, as the police suspected him on the basis of her text message to him, until she began to imagine that perhaps she and Lumumba really had been involved after all. After she signed the statement implicating him, had a meal and some time to gather her thoughts, she went back to saying that she had no recollection of having been involved, and that the statement was collected under coercion.
A prior probability of .002 for guilt strikes me as a fairly credible starting point (I assigned a higher prior, but in retrospect I think I should have adjusted down considerably based on the fact that Knox and Sollecito meet none of the FBI dangerous individuals criteria,) but it seems that you are adjusting strongly upwards on evidence that had already been discredited, or is already very weak due to high likelihood of coercion.
That’s not what Moore says. His reasoning is based on the assumption that if they were involved, they were physically involved. Read his article again.
In addition to the forensic evidence, Moore also lays out sixteen FBI criteria that serve as indicators of violent inclinations, out of which Knox meets zero. For a person to cheer on the attacker of a close acquaintance they had no motive to kill themselves, I would strongly expect them to show signs of a violent or sociopathic personality, which Knox and Sollecito do not.
Do you agree that in the scenario I described where Knox and Sollecito encourage Guede and then run, it’s pretty likely that they would not have left any DNA traces on or about her [ETA: Kercher’s] person?
Do you agree that in this scenario, it’s pretty likely that Knox and Sollecito would not have left the kind of evidence described by Moore in his article?
Yes, I agree. However, I think that this scenario has a dramatically low prior probability. The calculations I’ve given there describe the upper limits of my confidence interval on the likelihood I would assign to such an event; I believe that even one person, who has no motive to kill the victim, is not a sociopath, and meets none of the FBI dangerous individuals criteria, witnessing a violent attack by a person they do not have strong affiliations with (if the person were, say, a fellow gang member or mobster it would be a different matter,) and cheering on or otherwise encouraging the attacker, rather than say, intervening to help the victim, calling the police, or panicking and running away, would be an event completely unprecedented in the history of criminology. For two people to do so together goes beyond incredible into downright ludicrous.
I agree that the forensic evidence does not, as far as I am aware, rule out such involvement, but the evidence of Knox and Sollecito’s failure to meet any of the FBI dangerous individuals criteria does weigh heavily against it, and in my prior probability distribution for Knox and Sollecito’s involvement in the murder, I would only assign such a scenario a fraction of the likelihood that I would assign to the already unlikely “Bizarre sex game goes out of control” scenario, which the forensic evidence does rule out.
The probability that Knox (and Sollecito) were involved in the murder is necessarily greater that the probability that they were physically involved. Agreed?
And do you agree that if they were involved in the murder but did not hold, grab, or stab Kercher, then there is a good chance that they would not have left biological traces on or about Kercher’s person?
Agreed and agreed, but i would assign most of the prior probability mass for their being involved in the murder, but not physically, to the possibility that they were involved through conspiracy, and the lack of evidence of communication between Knox and Sollecito and Guede weighs heavily against that. The prior probability of their being involved via happening to witness the attack, and encouraging the attacker, represents such a miniscule part of my probability distribution that I would need a great deal more evidence to raise it to the point of being worthy of attention. It took me a while to realize that you were positing such a scenario because its initial plausibility was so low that it didn’t occur to me that you would suggest such a thing.
Given that many of the prosecution’s claims have been discredited, or simply do not constitute evidence in favor of Knox and Sollecito’s involvement, I think that a 5000:1 odds ratio is much too much weight to assign to their case. However, even if we do assign that much weight, I don’t think that it comes anywhere near to overcoming the prior improbability of the sort of involvement you’re suggesting.
The “staged break in” claims have been more or less discredited; all the evidence thus far has been found consistent with a real break-in.
I’m not sure what you mean by this. The fact that a piece of evidence is “consistent with” a real break-in does not mean that such a piece of evidence is not indicative of a staged break-in. For example, when Sollecito called the police, he allegedly advised them that there had been a break-in but nothing had been taken even though there was no way he could have known this if it had been a real break-in. This evidence is “consistent with” a real break-in in the sense that it’s possible to come up with an innocent explanation for Sollecito’s statements. But it doesn’t change the fact that it’s a piece of evidence which supports the staged break-in hypothesis.
the “forgotten phone calls” refers to a three second phone call from Knox to Kercher,
No, I was mainly referring to Knox’s trial testimony where she claimed to have forgotten about a call she made to her mother on the day the body was discovered. The call was made extremely early in the morning, Seattle time.
The “changing alibis” was done,according to Knox, under extreme duress
So what? Look, I agree that one can look at each piece of circumstantial evidence against Knox and come up with an innocent explanation for each piece. It doesn’t change the fact that when you look at all of it together, it’s extremely unlikely that an innocent person would have all of that evidence against him.
For a person to cheer on the attacker of a close acquaintance they had no motive to kill themselves, I would strongly expect them to show signs of a violent or sociopathic personality, which Knox and Sollecito do not.
I’m not sure what your point is here. I agree that Knox is not a likely murderer. That’s why my estimate of the probability of her guilt is only 90%. If she had been Kercher’s ex-lover; or if she came from a more violent demographic, my estimate would be higher.
Yes, I agree. However, I think that this scenario has a dramatically low prior probability
Well let me ask you this: Assuming that Knox was involved in the murder, what do you think the chances are that she was not physically involved?
I’m not sure what you mean by this. The fact that a piece of evidence is “consistent with” a real break-in does not mean that such a piece of evidence is not indicative of a staged break-in. For example, when Sollecito called the police, he allegedly advised them that there had been a break-in but nothing had been taken even though there was no way he could have known this if it had been a real break-in. This evidence is “consistent with” a real break-in in the sense that it’s possible to come up with an innocent explanation for Sollecito’s statements. But it doesn’t change the fact that it’s a piece of evidence which supports the staged break-in hypothesis.
Knox had already returned to the residence she shared with Kercher, and witnessed the signs of the break-in, but not, according to her own testimony, entered Kercher’s room and seen the murder scene (if she had entered the room, she would probably have left footprints and/or gotten blood on her shoes, which the police would have been able to discover, so we can assume that she probably did not.) She returned to Sollecito’s house after this, had breakfast, and made her calls. If their house had been robbed, it is probable that they would have taken things that she would not have to enter Kercher’s room to find missing, so “nothing was taken” would be a reasonable inference based on the information she had available to her. I agree that the claim that nothing was taken shifts the probability in favor of a staged break-in, but I don’t think that it shifts it very much.
The prosecution’s case for a staged break-in also rested on arguments such as the pattern of broken glass being inconsistent with a break from outside, and it being impossible to get in through the fire escape, both of which were simply false and represent no weight as evidence at all.
No, I was mainly referring to Knox’s trial testimony where she claimed to have forgotten about a call she made to her mother on the day the body was discovered. The call was made extremely early in the morning, Seattle time.
Well, this is not what I thought you had in mind, but I do not think that this is something Knox is much more likely to testify that she forgot if she was involved in the murder than if she was not. Whether she was innocent or guilty, we can assume that she was under a fair amount of stress, and the contents of the phone call weren’t incriminating anyway.
So what? Look, I agree that one can look at each piece of circumstantial evidence against Knox and come up with an innocent explanation for each piece. It doesn’t change the fact that when you look at all of it together, it’s extremely unlikely that an innocent person would have all of that evidence against him.
I strongly disagree. I would bet in favor of police finding a body of evidence of at least equal strength against me, were I to be investigated for a murder I was uninvolved in. I forget phone calls frequently, make statements based on inference which I do not have highly certain knowledge of on a regular basis, would probably change my alibi if I were abused and exposed to strong suggestion by the police, and have unusual mannerisms. This is why I originally brought up the point about the evidence being filtered; I am unconvinced that the evidence brought forward by the prosecution is more than I would expect them to find for the average person to whom I assigned the same prior probability of guilt.
Well let me ask you this: Assuming that Knox was involved in the murder, what do you think the chances are that she was not physically involved?
I’d call 99.99999% a fair estimate. But I’d estimate about 0.00000000000001% likelihood of her being physically involved, based on my previous likelihood calculation, plus the evidence that she doesn’t meet any of the FBI dangerous individuals criteria, and the lack of evidence of conspiracy with Guede, both of which I didn’t account for before, and the evidence of her failure to leave forensic evidence. So if you multiply the chance that she was involved physically by the number of times more likely I think it was that she was involved non-physically, you get about 0.0000001% likelihood. Most of my probability mass for that would be that she somehow managed to conspire with Guede without producing evidence that the police would discover, rather than the even more unlikely hypothesis that she happened upon the murder and encouraged it.
My probability estimate that she and Sollecito were both involved is, of course, lower than my probability estimate that Knox alone was.
She returned to Sollecito’s house after this, had breakfast, and made her calls.
At the time Sollecito reported that there was a break-in and nothing was taken, he was aware of the broken window in Romanelli’s room but Romanelli had not returned and inspected her room. Agreed?
and it being impossible to get in through the fire escape
I don’t recall the prosecution making this claim. In fact, I don’t recall hearing anything about a fire escape. Can you give me a link?
but I do not think that this is something Knox is much more likely to testify that she forgot if she was involved in the murder than if she was not and the contents of the phone call weren’t incriminating anyway.
I disagree. The contents of the call were very inculpatory and it was extremely convenient for her to forget about it. Anyway, even if she was under stress, it seems pretty unlikely to me that she would forget the call and continue to forget the call even after she was reminded of both the call and its contents. Especially given that it was an unusual call; made at a time when she was wide awake but knew her mother would be sleeping.
I forget phone calls frequently,
I doubt you would forget an unusual phone call made to your mother at a time when you knew you would be waking her up. Especially if you were reminded about the time and contents of the call.
I would bet in favor of police finding a body of evidence of at least equal strength against me
Well in your view, what are the strongest pieces of circumstantial evidence against Knox?
I’d call 99.99999% a fair estimate
Ok, then you should not object if my probability of Knox’s guilt doesn’t drop much if I add the assumption that Moore’s assertions are substantiated.
By the way, I am STILL waiting for cites backing up your claims about (1) trace DNA ending up on or about Kercher’s person if Knox (or Sollecito) grabbed, held, or stabbed her; and (2) further evidence of Guede DNA traces being discarded and not making it into the record.
At the time Sollecito reported that there was a break-in and nothing was taken, he was aware of the broken window in Romanelli’s room but Romanelli had not returned and inspected her room. Agreed?
I disagree. The contents of the call were very inculpatory and it was extremely convenient for her to forget about it. Anyway, even if she was under stress, it seems pretty unlikely to me that she would forget the call and continue to forget the call even after she was reminded of both the call and its contents. Especially given that it was an unusual call; made at a time when she was wide awake but knew her mother would be sleeping.
She was a college student without a lot of experience living on her own. Her home is broken into, she’s in a panic, she calls mom and asks “My home has been broken into, what do I do?”
Forgetting such a call, even repeatedly, would be entirely characteristic of me.
I don’t recall the prosecution making this claim. In fact, I don’t recall hearing anything about a fire escape. Can you give me a link?
I had remembered komponisto, who translated the Massei-Cristiani report, making such a statement, but applying a search function over his posts, I can’t find any such claim. I suspect that someone else made a statement along those lines and I misremembered, but I don’t know that they would be a credible information source, so I’m prepared to drop that claim.
Well in your view, what are the strongest pieces of circumstantial evidence against Knox?
Unusual mannerisms, and the dryer.
Ok, then you should not object if my probability of Knox’s guilt doesn’t drop much if I add the assumption that Moore’s assertions are substantiated.
That’s my estimate for the relative probability after the forensic evidence is accounted for. My likelihood estimate of Knox and Sollecito’s guilt after accounting for the forensic evidence is approximately equal to my probability estimate prior to it minus all the probability mass I assigned to their physical involvement. You already acknowledged that if you accept Moore’s arguments, it reduces the likelihood of Knox and Sollecito’s physical involvement to near zero. If your prior estimate assigns 90% probability to Knox and Sollecito being guilty, and 85% probability to any of the prosecution’s scenarios being correct, then that leaves at most 5% of your probability mass assigned to the possibility that Knox and Sollecito were involved non-physically, since the prosecution did not put forward any scenarios in which Knox and Sollecito were involved non-physically. So if you subtract nearly all of the probability mass that you assigned to Knox and Sollecito being involved physically, that should reduce your probability estimate of Knox and Sollecito’s guilt to just over 5% at most, less by any overlap between “Knox and Sollecito physically involved in the murder” and “Prosecution’s scenarios wrong.”
By the way, I am STILL waiting for cites backing up your claims about (1) trace DNA ending up on or about Kercher’s person if Knox (or Sollecito) grabbed, held, or stabbed her; and (2) further evidence of Guede DNA traces being discarded and not making it into the record.
For reasons I have already explained, this would be difficult for me to produce. I would need to contact an expert. I will agree to put in that effort if you also attempt to find a cite for your implicit premise, that 1:2 relative odds of non-physical to physical involvement is a credible distribution of probability mass in a murder with no evidence of conspiracy. You may agree to do this, and I will attempt to get a citation from an expert, you may offer some other assertion of yours that you are willing to go to the effort to acquire a citation for, and if I agree that it’s a reasonable substitution, I will make the effort to get the citation. If you do neither, I will not make the effort to get the citation, and I will have to take issue with your making further requests.
Ok, then I still don’t understand what you mean by your claim that the evidence is “consistent with” a real break-in. This is one piece of evidence which you seem to concede supports the staged break-in hypothesis.
She was a college student without a lot of experience living on her own. Her home is broken into, she’s in a panic, she calls mom and asks “My home has been broken into, what do I do?”
Assuming you are accurately summarizing the contents of the call, it’s suggestive of guilt. The fact that you (seem to) have no idea why is telling.
but I don’t know that they would be a credible information source, so I’m prepared to drop that claim.
You should also be prepared to confront the reality that you have very poor understanding of the case against Knox and Sollecito. This is the second time you have confidently mischaracterized the prosecution’s case. You don’t understand the significance of the forgotten phone call. And you seriously believe that Knox’s mannerisms are among the strongest pieces of circumstantial evidence against her.
It’s obvious that you are suffering from confirmation bias, i.e. you aren’t paying serious attention to evidence (or argument) which contradicts your views about the case.
That’s my estimate for the relative probability after the forensic evidence is accounted for.
Ok, then tell me your probability estimate ignoring the “forensic evidence.”
I will agree to put in that effort if you also attempt to find a cite for your implicit premise, that 1:2 relative odds of non-physical to physical involvement
Actually the odds I have been using are roughly 1:20. You seriously think that’s way off-base? (Please don’t evade this question.)
Anyway, your (apparent) claim about Italian police and legal procedure is one that even you need to concede you do not have the expertise to make.
Assuming you are accurately summarizing the contents of the call, it’s suggestive of guilt. The fact that you (seem to) have no idea why is telling.
I too have no idea. Yes, it’s something a person who has something to hide and doesn’t want to involve the police might do. It’s also something a confused college kid in a foreign place might do on encountering any hiccup. Completely consistent with either guilt or innocence means it sways me very little.
You should also be prepared to confront the reality that you have very poor understanding of the case against Knox and Sollecito. This is the second time you have confidently mischaracterized the prosecution’s case. You don’t understand the significance of the forgotten phone call. And you seriously believe that Knox’s mannerisms are among the strongest pieces of circumstantial evidence against her.
You asserted before that you yourself do not have a great deal of familiarity with the arguments made by the prosecution, and yet you find it adequate to support 90% confidence of guilt.
I agree that this statement was probably a mischaracterization of the prosecution’s argument, but can you name any other piece of evidence regarding the break in which is more probable in light of a staged break-in than a real one?
I think you are dramatically underestimating the frequency with which college students unused to living alone and exposed to frightening and unfamiliar situations will default to the action of “panic and call mom.”
I am probably subject to some degree of confirmation bias in this case, but I would need to weight the evidence in favor of Knox and Sollecito’s involvement many, many times more heavily for it to become a credible assertion. I think that you are yourself subject to an extremely strong confirmation bias here; weighting Knox and Sollecito’s changing of their alibis under probable abuse more heavily than Knox, Sollecito and Guede’s mutual failure to implicate each other, for instance, seems purely nonsensical.
Ok, then tell me your probability estimate ignoring the “forensic evidence.”
Some quick googling doesn’t give me any results relevant to the question of what proportion of murders involve conspiracy, but I would suggest a prior of maybe 1:100. However, the lack of evidence to support a conspiracy would weigh heavily against that, so I’d shift down to maybe 1:10,000.
“Suspects witness victim, who they have no motive to kill, being attacked, and cheer on the attacker,” would approximate to zero percent of my probability mass.
If you can point to any evidence that such an event (non sociopathic, FBI zero threat rating individuals witnessing an attack on someone they have no motive to kill, by someone they do not have strong affiliation with, and cheering on the attacker,) has ever happened before, I would agree that I am underestimating the probability. Can you point to any such evidence?
Actually the odds I have been using are roughly 1:20. You seriously think that’s way off-base? (Please don’t evade this question.)
I think it’s somewhat off base as a prior, and very very off base in light of the lack of evidence of conspiracy, and if you assign a significant fraction of that to “witnessed attack and cheered it on,” I think it’s positively ludicrous.
Anyway, your (apparent) claim about Italian police and legal procedure is one that even you need to concede you do not have the expertise to make.
You asserted before that you yourself do not have a great deal of familiarity with the arguments made by the prosecution, and yet you find it adequate to support 90% confidence of guilt.
Nonsense, I clearly stated that my probability assessment was based mainly on the actual facts as opposed to arguments.
I agree that this statement was probably a mischaracterization of the prosecution’s argument,
Your second mischaracterization. Not only that, you seriously believe that one of the best pieces of circumstantial evidence against Knox is her “mannerisms.” And not only that, you do not understand the significance of the forgotten phone call. Just admit that you don’t really understand the case against Knox (or Sollecito). I promise you the sky won’t fall.
I am probably subject to some degree of confirmation bias in this case, but I would need to weight the evidence in favor of Knox and Sollecito’s involvement many, many times more heavily for it to become a credible assertion.
Given that you clearly don’t know the strength of that evidence, you shouldn’t have an opinion one way or another.
I think that you are yourself subject to an extremely strong confirmation bias here; weighting Knox and Sollecito’s changing of their alibis under probable abuse more heavily than Knox, Sollecito and Guede’s mutual failure to implicate each other, for instance, seems purely nonsensical.
Please show me where I made such a weighing. Please QUOTE me. Failing that, please admit that I made no such weighing and apologize.
what proportion of murders involve conspiracy,
What exactly do you mean by “conspiracy”? It seems pretty clear that if Knox and Sollecito were involved in the crime, they acted in concert with Guede.
In the scenario I describe where Knox and Sollecito encourage Guede from a few feet away and then run, would you count that as a “conspiracy”?
If you can point to any evidence that such an event (non sociopathic, FBI zero threat rating individuals witnessing an attack on someone they have no motive to kill, by someone they do not have strong affiliation with, and cheering on the attacker,) has ever happened before, I would agree that I am underestimating the probability. Can you point to any such evidence?
I think I probably could. But let’s make sure I understand what you are saying.
You are claiming that Scenario A (non sociopathic, FBI zero threat rating individuals witnessing an attack on someone they have no motive to kill, by someone they do not have strong affiliation with, and cheering on the attacker) is at least a thousand times less likely than scenario B ((non sociopathic, FBI zero threat rating individuals witnessing an attack on someone they have no motive to kill, by someone they do not have strong affiliation with, and physically participating in the attack).
You are also claiming that Knox and Sollecito are known to be non-sociopathic and known not to have had a motive to harm Kercher.
Do I understand you correctly?
Can you explain what this apparent claim is?
That if additional samples taken from on or about Kercher’s person matched Guede, it would not have become part of the record in the case.
ETA:
but can you name any other piece of evidence regarding the break in which is more probable in light of a staged break-in than a real one?
A single piece of evidence? Sollecito’s phone call comes close, but probably not.
Your second mischaracterization. Not only that, you seriously believe that one of the best pieces of circumstantial evidence against Knox is her “mannerisms.” And not only that, you do not understand the significance of the forgotten phone call. Just admit that you don’t really understand the case against Knox (or Sollecito). I promise you the sky won’t fall.
No, I don’t concede that. If you can explain to me why it’s so much more likely that Knox would make that phone call if she were guilty than if she were innocent, perhaps it will be something I failed to consider, and I will change my mind. But as-is, it seems to me like such a weak enough piece of evidence that had the police not already promoted Knox and Sollecito to attention, they would probably not have treated it as worthy of notice.
Please show me where I made such a weighing. Please QUOTE me. Failing that, please admit that I made no such weighing and apologize.
I inferred that you made such a weighting from this statement
At the same time, the chances that an innocent person would have this kind of circumstantial evidence against them (changing alibis, staged break-in, forgotten phone calls, and so forth) is much lower. Perhaps 1 in 5000. (This is including the exculpatory aspects of the evidence as well.)
If I was incorrect, then I apologize for that. But I don’t see how you’re getting an odds ratio of 1:5000 in favor of guilt for changed alibis (very high likelihood of coercion,) a forgotten phone call, a phone call stating that nothing had been taken in the break-in (which they could reasonably infer,) versus mutual failure to implicate, absence of evidence of motive, absence of evidence of dangerous personalities or inclinations, absence of evidence of conspiracy, and absence of evidence of physical presence at the scene.
What exactly do you mean by “conspiracy”? It seems pretty clear that if Knox and Sollecito were involved in the crime, they acted in concert with Guede.
In the scenario I describe where Knox and Sollecito encourage Guede from a few feet away and then run, would you count that as a “conspiracy”?
By conspiracy, I mean that they intended for Kercher to be murdered, and made plans to that effect in advance with her murderer.
I think I probably could. But let’s make sure I understand what you are saying.
You are claiming that Scenario A (non sociopathic, FBI zero threat rating individuals witnessing an attack on someone they have no motive to kill, by someone they do not have strong affiliation with, and cheering on the attacker) is at least a thousand times less likely than scenario B ((non sociopathic, FBI zero threat rating individuals witnessing an attack on someone they have no motive to kill, by someone they do not have strong affiliation with, and physically participating in the attack).
You are also claiming that Knox and Sollecito are known to be non-sociopathic and known not to have had a motive to harm Kercher.
Do I understand you correctly?
No. I am saying that Scenario A is much less likely than the possibilities that dominate the probability mass for “Knox and Sollecito physically involved in Kercher’s murder.” This would include possibilities such as “violent argument over cleaning roster gone downhill,” “sex game gone wrong,” etc. The prior probability for “Knox and Sollecito witness Guede attacking Kercher, and proceed to assist Guede in the attack,” as with the probability that they would cheer him on, is infinitesimal.
In retrospect, I think that the relative prior probability that I assigned to their involvement through conspiracy (1:100) was too low, but it’s still very low in light of the absence of evidence that Knox and Sollecito conspired with Guede, and the likelihood that they would observe an attack in progress and support the attacker is infinitesimal.
Also note that if Knox and Sollecito came to the residence Knox shared with Kercher, and witnessed an attack in progress, they cannot have let Guede in. Any evidence in favor of a staged break-in is evidence against this hypothesis.
The prosecution tried quite hard to find a motive for the killing, and was unable to establish one, unless you account “crazy sex game gone wrong” or “satanic ritual” as motives, neither of which they were able to present evidence for. At this point, the notion that Knox and Sollecito had a motive to murder Kercher deserves a very low probability.
Amanda Knox was known by her acquaintances for having an low tolerance for conflict, and being uncomfortable if she and people around her were in disagreement, behavior which is highly inconsistent with Antisocial Personality Disorder. Further, if the prosecution was able to find evidence that she or Sollecito had APD, they had strong motive to make record of it, but the Massei-Cristiani report contains no such evidence. So the probability that Knox and Sollecito would witness an attack on Kercher and support the attacker may be dominated by the possibility that they were secretly sociopaths and had reason to want her dead, but we have more evidence than for the average person to suspect that neither was the case.
That if additional samples taken from on or about Kercher’s person matched Guede, it would not have become part of the record in the case.
I never said that I expected that if they found additional samples implicating Guede, it would not have become part of the case, I said that I did not have strong expectation that if further evidence been found against Guede, it would have been raised in the case (or to be more precise, I do not have a strong expectation that they would have brought the evidence to the case and I would have found out about it; Guede’s case was not controversial and so was not heavily reported on, and he opted for fast track trial in which he waived the right to contest any evidence against him, and all the primary documents are in Italian.)
Because I do not have a great deal of expertise regarding the Italian legal system, my model of reality would not have taken much of a hit either way.
If you can explain to me why it’s so much more likely that Knox would make that phone call if she were guilty than if she were innocent, perhaps it will be something I failed to consider, and I will change my mind.
It’s not really my job to explain it to you. It’s your job to do your homework to find out what the best and most important evidence is on both sides before you make your probability assessment. I have done that with both sides. On the other hand, you obviously have a very poor understanding of what the case against Knox is. As far as I can tell, you haven’t even tried to defend your claim that Knox’s mannerisms was among the best circumstantial evidence against her.
If I was incorrect, then I apologize for that.
Yes, you are incorrect. My statement was clearly looking at the body of evidence as a whole.
By conspiracy, I mean that they intended for Kercher to be murdered, and made plans to that effect in advance with her murderer.
Ok, then I don’t see how the concept of a “conspiracy” adds much to the discussion. If there was a conspiracy, Knox and Sollecito plausibly may still have been physically involved. If there was not a conspiracy, Knox and Sollecito may plausibly not have been physically involved.
No. I am saying that Scenario A is much less likely than the possibilities that dominate the probability mass for “Knox and Sollecito physically involved in Kercher’s murder.”
In that case you are comparing apples and oranges, since you are comparing one narrow scenario I proposed with much bigger set.
I think that the relative prior probability that I assigned to their involvement through conspiracy (1:100) was too low, but it’s still very low in light of the absence of evidence that Knox and Sollecito conspired with Guede, and the likelihood that they would observe an attack in progress and support the attacker is infinitesimal.
You are ignoring other possibilities here. A more likely scenario is that Knox and Sollecito conspired with Guede to harm Kercher in some way but did not intend for her to die.
For example, consider the murder of Janet Chandler which took place back in 1979. Her roommate and co-worker—a girl named Laurie Swank—apparently helped to organize a sexual attack on her. Swank admitted to cheering on the attackers who ultimately killed Chandler. There was no DNA testing back then, but if there had been, there would not have been any of Swank’s DNA on or about Chandler’s person.
Also note that if Knox and Sollecito came to the residence Knox shared with Kercher, and witnessed an attack in progress, they cannot have let Guede in. Any evidence in favor of a staged break-in is evidence against this hypothesis.
I agree but so what? I’m not proposing that Knox and Sollecito walked in on Guede assaulting Kercher and spontaneously decided to help him. I’m proposing that they acted in concert for some unlawful purpose (which by hypothesis pretty clearly happened) and that for whatever reason, Knox and Sollecito were not close to Kercher when Guede stabbed her to death.
So basically there are two possibilities, assuming that Knox and Sollecito were involved in the murder: (1) They acted in concert and were all physically involved; and (2) They acted in concert but only Guede was physically involved. You seem to be saying that (1) is far more likely than (2). I don’t see why, since both scenarios are (a priori) pretty remote. I don’t see any (a priori) reason to think that one is 100 or 1000 times more likely than the other.
But let’s be specific: What do you think is the probability ratio of (1) and (2)?
I said that I did not have strong expectation that if further evidence been found against Guede, it would have been raised in the case
I don’t think that’s what you said, but anyway it seems to me that Knox and Sollecito (or their supporters) would have had incentive to raise this issue. If there were 100 traces of Guede on Kercher’s person and none of Knox, there’s a decent—perhaps compelling—argument that the prosecution was wrong. Certainly Moore did not make this argument even though it would have fit in nicely with the rest of his essay.
If you can explain to me why it’s so much more likely that Knox would make that phone call if she were guilty than if she were innocent, perhaps it will be something I failed to consider, and I will change my mind.
It’s not really my job to explain it to you. It’s your job to do your homework to find out what the best and most important evidence is on both sides before you make your probability assessment.
Downvoted for this remark and most of the next paragraph. In this context, Derstopa is clearly aware of the phonecall and doesn’t understand why one would think it is evidence of guilt. Explaining why one would think that it is non-trivial evidence of guilt (if one thinks it is) is what one should do if one trying to have an actual discussion. Refusing to explain that and telling someone to just do their homework is rude, logically rude, and generally not productive for people trying to understand your viewpoint.
Explaining why one would think that it is non-trivial evidence of guilt (if one thinks it is) is what one should do if one trying to have an actual discussion.
I would say it depends on what is being discussed. My point in that post was not so much that Knox is probably guilty as that Derstopa is ignorant of the case against Knox. Note also that Derstopa is aware of the phone call only because I mentioned it to him.
I’m not Desrtopa, but I had heard about the call way back in 2009. I dismissed it as essentially irrelevant, because no one who has mentioned it has ever said anything convincing about how it was inculpatory. Apparently it was supposed to be obvious? But it wasn’t, not to me, nor apparently to Desrtopa. At best I can make strawman arguments which I find easy to knock down. Having you respond to those would be a waste of everyone’s time. So please, just tell me why you found this call suspicious.
I don’t speak Italian. I have been limited to secondary sources, rather than the prosecutor’s own arguments directly, but I really have looked closely at arguments by those who think Knox and Sollecito are guilty. At the time, they did not articulate any reason for me to find the phone call suspicious. I still haven’t found anything after looking again.
but I really have looked closely at arguments by those who think Knox and Sollecito are guilty. At the time, they did not articulate any reason for me to find the phone call suspicious.
My point in that post was not so much that Knox is probably guilty as that Derstopa is ignorant of the case against Knox. Note also that Derstopa is aware of the phone call only because I mentioned it to him.
The first sentence is irrelevant to my observation. The second claim (about only knowing about the phonecall from your remark) seems inaccurate to my reading of the conversation. From my skimming the thread again the only bit I see that could give you the impression that Derstopa wasn’t aware of the phonecall is this exchange where Derstopa doesn’t say that. In that context, Derstopa thinks you are talking about a different phonecall. Nothing in that is a useful indication that Derstopa didn’t know about this phone call. Moreover, if you thought that that was the problem then telling Derstopa that he should read up on the prosecutions case would have made minimal sense as a reply at that point, not a reply three days later to where Derstopa asks about how one thinks the phonecall is relevant, especially when they’ve already explained why it seems only marginally relevant.
Nothing in that is a useful indication that Derstopa didn’t know about this phone call.
Here’s what he said:
Unless you’re referring to some argument presented by the prosecution that I haven’t been able to find, the “forgotten phone calls” refers to a three second phone call from Knox to Kercher.
It seems pretty clear to me that he was not aware of the phone call I was referring to.
Moreover, if you thought that that was the problem then telling Derstopa that he should read up on the prosecutions case would have made minimal sense as a reply at that point,
As I recall, at that point, I was only just beginning to realize just how profoundly ignorant Derstopa was about the case against Knox.
It’s not really my job to explain it to you. It’s your job to do your homework to find out what the best and most important evidence is on both sides before you make your probability assessment. I have done that with both sides. On the other hand, you obviously have a very poor understanding of what the case against Knox is. As far as I can tell, you haven’t even tried to defend your claim that Knox’s mannerisms was among the best circumstantial evidence against her.
JoshuaZ’s comment reflects everything I have to say about this.
You are ignoring other possibilities here. A more likely scenario is that Knox and Sollecito conspired with Guede to harm Kercher in some way but did not intend for her to die.
Then why is there no evidence of prior communications between Knox, Sollecito and Guede to support this? Knox had only met Sollecito a week before Kercher’s murder, and had just started dating him, and there’s no evidence that Sollecito and Guede had ever even met as of the time of the murder. The prosecution has no record of phone calls between Knox and Guede, and has no evidence or testimony of any sort to suggest that the two were more than slightly acquainted.
I agree that this scenario had greater prior likelihood, but an absence of evidence that Sollecito and Guede had ever even met is a heavy weight of evidence against it.
I agree but so what? I’m not proposing that Knox and Sollecito walked in on Guede assaulting Kercher and spontaneously decided to help him. I’m proposing that they acted in concert for some unlawful purpose (which by hypothesis pretty clearly happened) and that for whatever reason, Knox and Sollecito were not close to Kercher when Guede stabbed her to death.
You think that there’s enough evidence to think that Knox and Sollecito, for whom there is an absence of evidence of any motive, and who had just recently met, deliberately collaborated with Guede, who there is no evidence that Sollecito had ever met, or that Knox had ever called or met with privately, conspired with Guede to harm Kercher? When did they do the conspiring, after 9:10 PM on the night that Kercher was killed?
The prosecution never alleged conspiracy between Knox, Sollecito and Kercher; the information they had available to them didn’t render it tenable.
So basically there are two possibilities, assuming that Knox and Sollecito were involved in the murder: (1) They acted in concert and were all physically involved; and (2) They acted in concert but only Guede was physically involved. You seem to be saying that (1) is far more likely than (2). I don’t see why, since both scenarios are (a priori) pretty remote. I don’t see any (a priori) reason to think that one is 100 or 1000 times more likely than the other.
If Knox and Sollecito were complicit in the crime, but not participants in the crime scene, they would most likely have to have premeditated some sort of harm against Kercher, and the prosecution found no evidence at all to suggest they would have any motive for such. If they were physically involved, they could have premeditated harm against Kercher, but they could also have become involved in the heat of the moment. Most murders are not premeditated, particularly in cases with no apparent motive prior to the crime.
I would say that my previous assignment of 1:100 may have been excessive; a prior ratio of 1:20, as you assigned, might be more reasonable. But the evidence that renders any before-the-event conspiracy between the three highly unlikely weighs against it.
Just how little evidence do you expect prosecutors with a heavy dose of motivated cognition to find for an innocent person, that you think what the prosecution found weighs so heavily against Knox and Sollecito? The data from the Innocence Project suggests an error rate higher than this for average cases prior to the use of DNA evidence, and the evidence against Knox and Sollecito is far weaker than for the average suspect, let alone convict, and the evidence pointing to their innocence far stronger.
JoshuaZ’s comment reflects everything I have to say about this.
As I said to JoshuaZ, the main point of that passage was that you are ignorant of the case against Knox. Which you do not seem to deny anymore.
Then why is there no evidence of prior communications between Knox, Sollecito and Guede to support this?
The prosecution has no record of phone calls between Knox and Guede, and has no evidence or testimony of any sort to suggest that the two were more than slightly acquainted.
Not all communications result in a phone or other record. Particularly if some of those communications are between drug users and elements of the criminal underworld.
I would say that my previous assignment of 1:100 may have been excessive; a prior ratio of 1:20, as you assigned, might be more reasonable.
Ok, then it was reasonable for me to go from 90% to 30%. Unless you think there is an error in my math?
Just how little evidence do you expect prosecutors with a heavy dose of motivated cognition to find for an innocent person, that you think what the prosecution found weighs so heavily against Knox and Sollecito?
I would say a lot less than here. Maybe one or two pieces of inculpatory evidence at most.
By the way, can I take it you are conceding that Knox (and/or her supporters) would have had an incentive to introduce and publicize evidence that there were more traces of Guede on or about Kercher than the two I mentioned earlier?
As I said to JoshuaZ, the main point of that passage was that you are ignorant of the case against Knox. Which you do not seem to deny anymore.
I do deny this. You accuse me of not defending my claim that Knox’s mannerisms are among the strongest evidence against her. I have defended at several points my reasons for assigning very low weight to pieces of evidence put forward by the prosecution. You, on the other hand, while asserting that the prosecution’s arguments collectively deserve high weight, have not defended your reasons for assigning high weight to any of them. In response to requests for a defense of such, you’ve told me it’s not your job to do my homework.
Ok, then it was reasonable for me to go from 90% to 30%. Unless you think there is an error in my math?
Only if you assign zero weight to the evidence rendering conspiracy between them unlikely.
Not all communications require physical meeting or result in a phone record. But when you have no evidence of private meeting, no phone records, no email records, and a very short time frame in which any sort of communication could have taken place, and you don’t adjust your likelihood of communication downwards at all, then it’s clear that you’re desperate to come to a particular conclusion and don’t want to perform any sort of update which would render it more unlikely.
I would say a lot less than here. Maybe one or two pieces of inculpatory evidence at most.
Then why are so many primary suspects and even convicts later proven to be innocent? Your estimation of the amount of evidence that it is reasonable to expect against an innocent person seems strongly contradicted given our information on false conviction rates.
By the way, can I take it you are conceding that Knox (and/or her supporters) would have had an incentive to introduce and publicize evidence that there were more traces of Guede on or about Kercher than the two I mentioned earlier?
I have stopped trying to argue that point because, in the face of a qualified expert without any professional stake in the case claiming that it is enormously unlikely for Knox and Sollecito to have left an absence of evidence if they were physically involved in the case, you insist on demanding a particular piece of evidence which I have already provided repeated and abundant explanations as to why it would be difficult for me to produce.
In fact, when I put it in words, I have to wonder I’m still arguing about any of this. Thus far you have shown a tremendous unwillingness to update in favor of any exculpatory evidence (when I think evidence from the prosecution is weak, I explain why I do not think it is much more likely if Knox and Sollecito are guilty than if they are innocent. You refuse to update on evidence entirely with an explanation of how it does not disprove all possible guilt scenarios.) You have been rude, logically rude, and I suspect you could go on arguing for Knox and Sollecito’s guilt indefinitely, increasing your confidence in your probability assignment no matter how much evidence mounts against you.
You have given me every reason to believe that if exposed to further evidence against your position, you will refuse to change your assessment at all. This simply isn’t worth my time.
If brazil84 weren’t a lawyer, I would have written this debate off as pointless a long time ago. It’s raising a negative affect around all the time I spend on this site, but simply walking away knowing that he’s going to continue practicing law without realizing that he’s been making some big mistakes of judgment that demand commensurately big corrections has up till this point been even more frustrating.
As a lawyer, I feel obligated to defend the profession a little. Lawyers exist to maximize the utility of the client. But a lawyer who has no grip on what is true cannot tell the client when it makes sense to throw in the towel. And that lack of judgment is not rewarded.
That said, lawyers are relatively well trained in behavior that is orthogonal to truth but correlated with winning. But for the most part, the legal system (i.e. the judge) is trying to determine truth within the limits of its empirical system.
Well said. But note that the behavior is not entirely orthogonal to truth-seeking. Being trained to discuss a dispute through the techniques of debating will surely decrease one’s ability to say “oops”. (not in the sense of it being a necessary decrease, but in the sense of it being a tendency to decrease such that my prior reduced confidence is justified)
This is actually something I struggle with, being trained as a polemic philosopher.
You, on the other hand, while asserting that the prosecution’s arguments collectively deserve high weight
You are talking about the evidence, right?
In response to requests for a defense of such, you’ve told me it’s not your job to do my homework.
In response to your request regarding the phone call, yes. The issue there is not that you have considered the argument on this point and rejected it; the issue is that you are completely unaware of the significance of the forgotten phone call. I identified the phone call merely as an example of the body of evidence against Knox.
Only if you assign zero weight to the evidence rendering conspiracy between them unlikely.
Well, assuming for the sake of argument that (1) Knox and Sollecito were involved in the murder; and (2) there was absolutely no conspiracy as you have defined that term, are you saying that your a priori probability estimate of physical involvement by Knox and Sollecito is 1000 times higher than no physical involvement?
If not 1000, then what?
But when you have no evidence of private meeting, no phone records, no email records, and a very short time frame in which any sort of communication could have taken place, and you don’t adjust your likelihood of communication downwards at all,
Are you saying that I have done so?
Then why are so many primary suspects and even convicts later proven to be innocent?
I dispute that “so many” primary suspects and even convicts are later proven to be innocent. How often do you think this happens? What is the percentage of convicts who are later completely exonerated? Surely it is far less than 10% and I have allowed a 10% chance that Knox and Sollecito are innocent.
In fact, when I put it in words, I have to wonder I’m still arguing about any of this. Thus far you have shown a tremendous unwillingness to update in favor of any exculpatory evidence
The only exculpatory evidence you have offered (which I have not already considered) is based on your non-expert opinion. I have patiently explained to you the effect it would have on my probability estimate if you were able to document your opinion.
Or is there some other piece of exculpatory evidence you think I am not considering in my probability estimate?
you insist on demanding a particular piece of evidence
Look, you are the one who is proposing that additional traces of Guede on Kercher would not have made it into the public record. I am asking for you to back up that claim.
Seriously, if you want to continue this debate, show me some evidence that you will actually update in the face of more evidence against your position. In the beginning of this debate, you said that you were aware that the DNA evidence against Knox and Sollecito had been challenged. In the face of the independent experts’ report indicating that there is no valid DNA evidence against them, and Moore’s expert claims for all the evidence we should expect to see in the case of the sort of involvement alleged by the prosecution that we do not see, you update your probability by… nothing. You say you would update your probability were I to provide the right references, but when I offer to make the effort to obtain those references in exchange for a comparable effort on your behalf to produce evidence for your position (I offered you to substitute evidence for any of your assertions if you suggested the substitute first and I agreed that it was appropriate,) you evade. Why should I think this conversation is worth my time? You’ve given me every reason think that if I go to further efforts, you’ll just find more excuses not to update.
Why should I think this conversation is worth my time? You’ve given me every reason think that if I go to further efforts, you’ll just find more excuses not to update.
Seriously, if you want to continue this debate, show me some evidence that you will actually update in the face of more evidence against your position
I already conceded that my probability estimate would drop from roughly 90% to roughly 30% if you could substantiate your claim about transfer of DNA traces.
But anyway, I am not particularly interested in debating Knox’s actual innocence with you given your ignorance of the case against Knox.
In the face of the independent experts’ report indicating that there is no valid DNA evidence against them
That’s not what the independent expert report said, as far as I know.
but when I offer to make the effort to obtain those references in exchange for a comparable effort on your behalf to produce evidence for your position (I offered you to substitute evidence for any of your assertions if you suggested the substitute first and I agreed that it was appropriate,) you evade
Nonsense, you are the one who is evading; since it’s not even clear that you dispute my “position” on the point and you refuse to clarify your position. You demanded evidence for a strawman position you set up.
I will ask you one last time, if you evade or ignore we are done:
Assuming for the sake of argument that (1) Knox and Sollecito were involved in the murder; and (2) there was absolutely no conspiracy as you have defined that term, are you saying that your a priori probability estimate of physical involvement by Knox and Sollecito is 1000 times higher than no physical involvement?
If not 1000, then what?
ETA:
Why should I think this conversation is worth my time?
I don’t know, what exactly are you trying to accomplish?
I don’t want to get sucked in by replying to more of brazil84′s comments, but in the spirit of this quote (edit, link fixed,) in case it will help me to do better in future...
By the way, if any lurkers are reading this, I will point out the fatal flaw in Desrtopa’s reasoning on the 90% to 30% issue. See, the prior probability of Knox and Sollecito being involved in the murder but not physically is quite low. But the prior probability of Knox and Sollecito being involved in the murder physically is also low. What matters for my calculation is the ratio of the two probabilities.
I suspect that’s why Desrtopa had to bob and weave with my questions on this issue.
Did it look to anyone else like I was bobbing and weaving with my responses to these questions? My perception was that I answered these questions in a straightforward way, but I may have been suffering from illusions of transparency.
You definitely got somewhat sucked into “debate mode”, and you often had far too many points lobbed at you to have caught all of them. It would seem to be possible for brazil84 to honestly have thought you were evading something, but I don’t think an outside observer should have thought so.
That you seem to have kept your cool throughout this ridiculous exercise is a credit.
We are done. You’ve given me no more reason to think there’s any point in having a conversation with you.
Ok bye.
By the way, if any lurkers are reading this, I will point out the fatal flaw in Desrtopa’s reasoning on the 90% to 30% issue. See, the prior probability of Knox and Sollecito being involved in the murder but not physically is quite low. But the prior probability of Knox and Sollecito being involved in the murder physically is also low. What matters for my calculation is the ratio of the two probabilities.
I suspect that’s why Desrtopa had to bob and weave with my questions on this issue. Because he knew deep down that there is no basis to think that this ratio is wildly lopsided. And if the ratio is not lopsided, then it’s perfectly reasonable to go from 90% to 30%.
I enjoy debating and I am fascinated by situations where the popular view is wrong. Also, I like debating where there isn’t anything important at stake.
Once in a while I learn something from debating with someone who disagrees with me.
I enjoy debating and I am fascinated by situations where the popular view is wrong. Also, I like debating where there isn’t anything important at stake.
I strongly suspect that you are incorrect about which side of this debate has more adherents. Neither LW nor Wikipedia are likely to be representative.
Once in a while I learn something from debating with someone who disagrees with me.
I hope to learn something in every discussion I have. Debate is for changing people’s minds. Debate for its own sake is pointless (aka—I don’t debate Flat-Earthers).
You don’t seem interested in saying something that I can glean insight from, and you don’t seem open to changing your mind.
Actually, I’ve done that before. It helps a surprisingly large amount in getting one to realize how astronomy can easily pay rent. And it really tests how well one understands how much historical scientists did really careful, clever observations. That said, I can’t plausibly argue that this is a good use of one’s time, merely that is probably marginally better use than say playing Farmville.
I strongly suspect that you are incorrect about which side of this debate has more adherents.
On this discussion board, the popular view seems to be that Knox is most likely innocent. I make no claims about what is believed in the world in general.
You don’t seem interested in saying something that I can glean insight from, and you don’t seem open to changing your mind.
The second part is incorrect—I am open to changing my mind. As far as the first goes, I don’t know where you are at in terms of insight so I couldn’t say.
ETA:
I hope to learn something in every discussion I have.
On this discussion board, the popular view seems to be that Knox is most likely innocent. I make no claims about what is believed in the world in general.
On this he is clearly correct.
In any case I don’t think this particular comment deserved to be as down voted as it was.
I dispute that “so many” primary suspects and even convicts are later proven to be innocent. How often do you think this happens? What is the percentage of convicts who are later completely exonerated? Surely it is far less than 10% and I have allowed a 10% chance that Knox and Sollecito are innocent.
First of all, just because they aren’t completely exonerated does not mean they are not innocent. A better question is what the percentage chance of innocence is.
Second of all, this assumes that letting a guilty person go free and sending an innocent person to get punished is of equal weight, which it is not.
Pardon for the digression, but isn’t Knox’s physical involvement in the murder the prosecutor’s theory of what happened? Otherwise, why should we be interested in Knox DNA on the victim’s bra?
I admit that it is conceptually possible that P(Knox is guilty) != P(prosecutor’s theory is correct), but in practice they tend to converge.
ETA: I’ll even agree that P(Knox is guilty) >= P(prosecutor’s theory is correct). Nonetheless, P(Knox is guilty) less P(prosecutor’s theory is correct) is a small number.
Pardon for the digression, but isn’t Knox’s physical involvement in the murder the prosecutor’s theory of what happened?
Yes, as far as I know.
Otherwise, why should we be interested in Knox DNA on the victim’s bra?
Actually I think the allegation here is that Sollecito got his DNA on the bra while cleaning up the crime scene.
I admit that it is conceptually possible that P(Knox is guilty) != P(prosecutor’s theory is correct), but in practice they tend to converge
I agree, but in this case there is less of a convergence than usual since the prosecution had no choice but to speculate. In any event, I’m allowing a pretty modest probability—like 5 or 10% -- that Knox and Sollecito were guilty despite a lack of physical involvement as alleged by the prosecution.
ETA: For example, the scenario I described where Knox and Sollecito encourage Guede from just inside the doorway and then turn and run when the blood starts to spill.
For example, the scenario I described where Knox and Sollecito encourage Guede from just inside the doorway and then turn and run when the blood starts to spill.
This scenario you describe pushes very hard on the distinction between moral guilt and legal guilt, at least in American law. Are we debating Knox’s legal culpability or moral culpability?
I agree, but in this case there is less of a convergence than usual since the prosecution had no choice but to speculate. In any event, I’m allowing a pretty modest probability—like 5 or 10% -- that Knox and Sollecito were guilty despite a lack of physical involvement as alleged by the prosecution.
So, from your point of view:
P(Knox guilty) = .9
P(prosecutor correct) = .85
And to clarify: “Knox guilty” = knowing Knox’s relevant behavior and thoughts from that time, every rational person would agree that she had committed a violation of the law that proximately caused the victim’s death.
Also, I think that if P(defendant is guilty) less P(prosecutor’s theory is correct) is in the range of .05, that is strong evidence that the prosecutor is utterly incompetent. After all, .05 probability is one in twenty. I confess unwillingness to believe that the prosecutor in Knox’s case is that incompetent. (More rigorously, my belief in that level of incompetence approaches .001, and even that seems high).
This scenario you describe pushes very hard on the distinction between moral guilt and legal guilt, at least in American law. Are we debating Knox’s legal culpability or moral culpability?
Well, it would still be illegal, although I admit that I interpreted brazil84′s meaning of “involved in the murder” more narrowly. I would have to make a different probability calculation for the the odds that they did not plan or participate in, but discovered and encouraged, the attack on Meredith Kercher. It would still contain much of the same evidence that I discussed before, and given that I have never even heard of such a thing happening, I would assign a much lower prior, particularly for someone who, as Moore points out, meets none of the FBI criteria for a dangerous personality.
It’s even more unlikely when you consider that we’re dealing with two people rather than one. If Knox and Sollecito did not plan to kill Kercher, and then found her being attacked in her home, the odds that one of them would be crazy enough to hang back and encourage the attacker, incredibly low as they are, are much higher than the odds that both would.
This scenario you describe pushes very hard on the distinction between moral guilt and legal guilt, at least in American law. Are we debating Knox’s legal culpability or moral culpability?
Well, it would still be illegal.
That’s why I specifically reference American law, which contains no general duty to rescue. That is one of the differences between English-descended Common Law, and French-descended Civil Code (which the common model in most of continental Europe).
I’m going to do a probability estimate for Knox and Sollecito having discovered and encouraged the attack. I’d do it in response to one of brazil84′s comments, but I’m already waiting on him to respond to some other questions and don’t want to sidetrack him.
Let’s start with the prior probability that Knox and Sollecito would witness the murder. Knox and Kercher lived in the same apartment, and Knox and Sollecito spent a lot of time together. Let’s call it .2
Neither Knox nor Sollecito meet the criteria for Antisocial Personality Disorder or any of the FBI dangerous individuals criteria. What should we call the probability that such a person, upon witnessing a violent attack on an acquaintance they have no motive to harm (the prosecution having established no motive) would stand back and encourage the attacker? I’ll call it >.0001. So if you have two individuals, the chance that they’d both be the sort of person who would hang back and encourage the attacker is >10^-8. But both of them would probably expect the other person to consider this behavior crazy and horrible, much like a person is less likely to commit actual murder in front of another person who they don’t know to be sympathetic, so I’ll assign a probability of .1 that if both were the sort of person to encourage the attacker, either would make the first move to start doing so. So at this point we have a probability of about 2x10^-10.
Factor in the mutual failure of Knox and Sollecito and Guede to implicate each other; I’ll assign this the same .05 probability that I did before. So now we have a probability of >8x10^-12.
If brazil84 assigns .85 probability to the prosecution’s account being correct, and .9 probability to their involvement, (I believe he may have been simply giving this as an example rather than actually asserting this,) then let’s suppose that the remainder of the probability lies in Knox and Sollecito being physically present and encouraging the attacker, but not entering the room or participating themselves. How much weight would we have to give to the prosecution’s evidence against them to get from our current probability to 5%? Their evidence would have to be more than 6.2x10^9 times more likely given Knox and Sollecito’s guilt than given their innocence.
As I stated in my earlier comment, I think that crediting them with a likelihood ratio of 50 times, including Knox’s mannerisms at the scene, is generous to their case.
Some additional thoughts on the arguments of the prosecution...
As I said in my first probability analysis, the arguments the prosecution put forward are mostly in the form of “This action seems more suspicious for an innocent person than a guilty person.” But for much of their case, I think even that is giving them too much credit. Let’s look at some examples.
The three-second phone call from Knox to Kercher. This is unusual behavior, right? Well, I’ve done this more than a few times myself, but I agree, it is pretty unusual. So how much weight would I assign that evidence?
None. The question is not how unusual the behavior is, but how much more unusual if she’s innocent than if she was involved in the murder. If I know that Amanda Knox was going, later that day, to come across Meredith Kercher being attacked, and encourage or participate in the attack, the three second phone call will not be any less confusing to me. Stalking and incessant phone calls are certainly more easily explained in light of a murder, but I know of no evidence whatsoever to suggest that murderous inclinations and unusually short phone calls are associated, nor can I think of any particular reason they would motivate one.
The time at which Amanda Knox claims to have woken up. The Massei-Cristiani report finds suspicion in the fact that Knox claimed to have gotten up at 10:00-10:30 AM that day, while her roommates characterized her as an early riser. For a person to violate their established sleep habits is unusual, although still normal enough that their say-so is generally enough to establish it as probably true absent additional reason to doubt them. I am personally not an early riser, but this morning I got up and left the house for a walk at 6:00 AM, while the stars were still bright out (which is in fact what got me thinking about the prosecution’s report in the first place.) But Amanda Knox was a murder suspect, doesn’t that give us additional reason to doubt her claim?
Well, no. Whether she was up at 10:30 or hours earlier has no bearing on the strength of her alibi. We know that the murder didn’t happen in the morning, and we know that she was in other places later in that day prior to the murder; it’s not as if she failed to establish that she ever left the house. Being involved in the murder doesn’t give her any additional reason to claim that she got up at 10:30.
Combined with misrepresentations and mistreatments of forensic evidence, including claims like the pattern of glass being inconsistent with a break-in, which was found to be simply false when tested, a large part of the prosecution’s body of arguments does not represent any weight as evidence at all.
This scenario you describe pushes very hard on the distinction between moral guilt and legal guilt, at least in American law.
Generally speaking under American law, if Knox and Sollecito intended for Guede to commit a serious crime against Kercher (such as rape or robbery), and Guede killed Kercher in the process, then Knox and Sollecito would be guilty of murder.
So, from your point of view:
P(Knox guilty) = .9
P(prosecutor correct) = .85
Basically yes, if by “prosecutor correct” you mean the prosecutor is correct that Knox and Sollecito physically participated in the murder. The prosecutor’s scenario contained a lot of details; it’s very likely that some of those details are wrong.
Also, I think that if P(defendant is guilty) less P(prosecutor’s theory is correct) is in the range of .05, that is strong evidence that the prosecutor is utterly incompetent
I disagree, depending on how you define “incompetent.” Let’s suppose you are the prosecutor—you are confident that Knox and Sollecito were involved in the murder but you don’t know exactly what happened. The police have brought you evidence suggesting that Knox stabbed Kercher. You know that in order to convict, you need to give the jury a plausible scenario for what happened. What choice do you have but to make your best guess and go with it?
Perhaps a better argument can be made that the prosecutors were unethical for advancing their speculations about sex games and such to the jury. It would be interesting to see a translation of the transcript of the prosecutor’s opening and closing arguments.
Generally speaking under American law, if Knox and Sollecito intended for Guede to commit a serious crime against Kercher (such as rape or robbery), and Guede killed Kercher in the process, then Knox and Sollecito would be guilty of murder.
Regardless of what Knox and Sollecito intended, they are guilty of none of Guede’s crimes unless they acted to participate in some way. But I agree that if Knox had committed conspiracy to rape (or rob, or whatever) Kercher with Guede, then she would be guilty of any murder Guede committed in furtherance or in consequence of that conspiracy. I’ll agree to a similar statement about aiding and abetting, but I don’t remember the technical language. (And all of this is American law, which is of limited relevance in Italy).
I disagree, depending on how you define “incompetent.” Let’s suppose you are the prosecutor—you are confident that Knox and Sollecito were involved in the murder but you don’t know exactly what happened. The police have brought you evidence suggesting that Knox stabbed Kercher. You know that in order to convict, you need to give the jury a plausible scenario for what happened. What choice do you have but to make your best guess and go with it?
A prosecutor’s theory is correct if it includes all the true, morally relevant details. Further, it contains no false, morally relevant details. Compare the a prosecutor saying:
(1) “The defendant shot the victim with a .44 caliber gun, but I don’t know the model of the gun.”
(2) “The defendant shot the victim, but I don’t know the caliber of the gun”
Assuming the statements are true descriptions of what actually occurred, the prosecutor saying (1) is correct and the prosecutor saying (2) is pushing the envelope. Further, a prosecutor who says, “The defendant shot the victim, and the killing was planned ahead of time,” when the defendant actually decided to kill during the meeting is incorrect.
As the last example shows, a defendant can be guilt of the crime charged. But I assert that prosecutors assert stories that contain morally relevant that are inconsistent with what actually happened far less than 5% of the time. And when inconsistencies happen, it suggests that the prosecutor is wrong on other morally relevant details.
The prosecutor’s scenario contained a lot of details; it’s very likely that some of those details are wrong.
From what I understand of the prosecutor’s theory of the case, it contained many additional morally relevant details beyond simply Knox and Sollecito’s physical involvement in Kercher’s murder.
P(Knox guilty) = .9
P(prosecutor correct) = .85
For the reasons I just discussed, your probability distribution seems inconsistent with the evidence. Specifically, you rescue Knox and Sollecito’s guilty from possible errors by the prosecutor by proposing a theory that even you think has only a 5% chance of being true.
Regardless of what Knox and Sollecito intended, they are guilty of none of Guede’s crimes unless they acted to participate in some way. But I agree that if Knox had committed conspiracy to rape (or rob, or whatever) Kercher with Guede, then she would be guilty of any murder Guede committed in furtherance or in consequence of that conspiracy. I’ll agree to a similar statement about aiding and abetting, but I don’t remember the technical language.
Agree.
A prosecutor’s theory is correct if it includes all the true, morally relevant details. Further, it contains no false, morally relevant details. Compare the a prosecutor saying:
I’m not sure what your point is here. Are we arguing about whether the prosecutor is competent? About whether he is unethical?
Do you agree that as a practical matter, in order to have a good chance of securing a conviction, a prosecutor needs to present a coherent narrative to the jury?
Specifically, you rescue Knox and Sollecito’s guilty from possible errors by the prosecutor by proposing a theory that even you think has only a 5% chance of being true.
I don’t understand this sentence. I’m reluctant to guess at what you mean, but please keep in mind that “prosecutor correct” can be taken different ways. Does it mean simply that he is correct about physical involvement; or does it mean that his entire narrative is substantially correct?
I don’t understand this sentence. I’m reluctant to guess at what you mean, but please keep in mind that “prosecutor correct” can be taken different ways. Does it mean simply that he is correct about physical involvement; or does it mean that his entire narrative is substantially correct?
My previous post was trying to define what I meant by correctness for the purpose of the probability analysis I did. In brief, a prosecutor is correct if the theory presented to the jury contains no morally relevant facts that are contradicted by what actually occurred. (To clarify what I mean by moral relevance: Did the burglar go in the unlocked front or unlocked back door? Legally relevant, but not morally relevant).
Thus, I think the answer to your question is that the prosecutor is incorrect if the narrative does not correctly describe what actually occurred. As I noted above, the prosecutor’s incorrectness is not inconsistent with the defendant’s legal guilt. But I assert that prosecutor’s incorrectness is suggestive of legal innocence because it shows the prosecutor doesn’t know what happened.
Do you agree that as a practical matter, in order to have a good chance of securing a conviction, a prosecutor needs to present a coherent narrative to the jury?
Sure, juries like stories. That isn’t precisely a rationalist feature of juries.
In the Kercher case, the prosecutor’s case was filled with morally relevant facts (e.g. the sex cult/orgy thing) that were wildly implausible and not supported by any evidence that I’m aware of. Then, more plausible assertions (i.e. it was impossible to enter the apartment via the fire escape) were contradicted. Since I independently believe that (P(defendant guilty of charged crime) & P(prosecutor incorrect)) is small, I’m left to estimate a very small likelihood that Knox is guilty.
Thus, I think the answer to your question is that the prosecutor is incorrect if the narrative does not correctly describe what actually occurred. As I noted above, the prosecutor’s incorrectness is not inconsistent with the defendant’s legal guilt. But I assert that prosecutor’s incorrectness is suggestive of legal innocence because it shows the prosecutor doesn’t know what happened.
Fine, but you don’t get to redefine the terms I used and then use your different definition to conclude that my probability estimate doesn’t make sense. That would be a fancy way of strawmanning me.
When I basically agreed that I believe that P (prosecutor correct) is 85%, I was careful to make clear what I meant by “prosecutor correct.” Please do not pretend that I meant (or that we were discussing) something different.
It was not my intent to strawman you. I seem to have misunderstood your position. Let’s return to basics.
Let’s define “no orgy, but guilty” as the state in which Knox and Sollecito are legally guilty of Kercher’s murder, but many of the prosecutor’s asserted morally relevant facts are false. By definition:
P(guilty) = P(prosecutor correct) + P(no orgy, but guilty)
P(no orgy, but guilty) = P(prosecutor incorrect) & P(guilty) [If I understand the terminology correctly]
Since you seem to agree that the prosecutor made incorrect assertions of morally relevant facts, your value of P(prosecutor correct) is negligibly small. Why do you think that the prosecutor made the assertions? Before you say “narrative,” let me assert that many of the prosecutor’s assertions could have been replaced with “I don’t know” with limited loss of coherence to the narrative.
Why do you think that the prosecutor made the assertions?
Can you give me a specific example?
Before you say “narrative,” let me assert that many of the prosecutor’s assertions could have been replaced with “I don’t know” with limited loss of coherence to the narrative
Ok, you’ve said it. And now my answer:
Narrative. I do agree that the prosecutor could have omitted a few things and still had a somewhat coherent narrative. But if he got rid of all the speculation, what he would be left without any motive or explanation for why Knox and Sollecito acted in concert with Guede.
Why do you think that the prosecutor made the assertions?
Can you give me a specific example?
Quoting wikipedia: “[The prosecutor alleged] that the murder was part of a Satanic ritual; that it was a sex game gone wrong; or that Kercher had refused to take part in an orgy.”
Honestly, a substantial part of my doubt of Knox and Sollecito’s guilt is the utter ridiculousness of those allegations. They were not necessary to provide narrative to the jury and could easily have been omitted, since there was apparently no evidence suggesting they were true.
Further, they are the type of allegations I would expect when the prosecutor is seeking to inflame, rather than rationally presenting the evidence.
Quoting wikipedia: “[The prosecutor alleged] that the murder was part of a Satanic ritual; that it was a sex game gone wrong; or that Kercher had refused to take part in an orgy.”
Ok, to answer your question, I believe that the prosecutor made these assertions because he felt pressure to answer the question of Knox and Sollecito’s motives.
Satanic ritual? The prosecutor could have just said, “The roommates had a big fight.” In terms of evidence of guilt, there are two major problems in the “satanic ritual” theory that are not present in a more mundane theory. The satanic ritual theory proves too much, in that there is almost no evidence that could possibly be inconsistent with it. Second, the theory functions to demonize the defendants, which is not consistent with strong actual evidence of guilt.
And those theoretical arguments ignore the practical fact that there was no evidence suggesting interest in satanic ritual by either defendant, which I consider important given the general implausibility of the theory given what I know about society (There are practically no murderous satanists in real life).
The prosecutor could have just said, “The roommates had a big fight.”
Apparently he did also propose more mundane motives. Here’s the full quote from Wikipedia:
The prosecutors proposed a number of possible motives for the murder, for which they produced no evidence, including that Kercher and Knox had fallen out over issues such as the cleaning roster in their home; that the murder was part of a Satanic ritual; that it was a sex game gone wrong; or that Kercher had refused to take part in an orgy.
When you quoted Wikipedia, why did you omit the part about the cleaning roster?
Because I’m interested in why the prosecutor kept talking after he articulated that theory.
Either theory is just speculation, agreed? I mean, if it’s unethical for the prosecutor to argue that there was a satanic ritual, it’s also unethical for him to argue that it’s a dispute over who should clean the apartment. Agreed?
I’m not trying to say the prosecutor is unethical. I’m saying that outlandish claims from the prosecutor are evidence of innocence. It’s just a Big Lie
And if the prosecutor really has no idea why defendant committed a crime, that’s also evidence of innocence, right?
One could rationalize the prosecutor’s behavior by saying that lower standards apply to statements to the media than statements in court. But the reason that it’s unethical to try cases in the media is that this reduces the likelihood that the trial will be correlated with the truth. I attribute knowledge of that principle to the prosecutor, so his decision to try the case in the media suggests that the defendants are innocent.
Ok, I thought we were discussing whether the prosecutor was ethical.
We were discussing whether the prosecutor is competent.
the decision to arrest Knox is evidence of guilt.
No, it isn’t. The decision to arrest is evidence that the authorities think they have evidence of guilt. But the arrest is not independent of the underlying evidence and does not make the underlying evidence stronger.
We were discussing whether the prosecutor is competent.
Ok, then why are you bringing up the question of whether the prosecutor’s assertions are evidentiary?
The decision to arrest is evidence that the authorities think they have evidence of guilt. But the arrest is not independent of the underlying evidence and does not make the underlying evidence stronger.
Similarly, the prosecutor’s proposal that there may have been a Satanic ritual involved is not independent of the underlying (lack of) evidence of motive and does not make the underlying evidence weaker.
The prosecutor’s theory is not evidence of the same type as DNA. But it tells us something about how strong the prosecutor thinks the legally relevant evidence is.
The reference to satanic ritual and sex games is an appeal to applause lights. It is political machinations, not an empirically based search for truth. The fact that the prosecutor made that appeal, rather than sticking with a plausible story, shows that the prosecutor doesn’t think the facts support guilt or doesn’t care.
As an observer, I’m allowed to give weight to the fact that one side of a dispute doesn’t seem to be trying to determine the facts.
The prosecutor’s theory is not evidence of the same type as DNA. But it tells us something about how strong the prosecutor thinks the legally relevant evidence is.
The decision to arrest the defendant is not the same type as DNA. But it tells us something about how strong the authorities think the legally relevant evidence is.
As an observer, I’m allowed to give weight to the fact that one side of a dispute doesn’t seem to be trying to determine the facts
I would say the amount of weight depends on the extent you are relying on unsupported assertions of that side. For example, if we knew nothing at all about the case except the prosecutor saying “Trust me, I have reviewed the evidence and I am satisfied that the Defendant is guilty, it would be very important that the prosecutor is proposing a seemingly outlandish hypothesis.
This is the expertise problem. You and I do not have access to all the evidence, so we look to the experts. If anyone is an expert on the strength of the case, it is the prosecutor. And a prosecutor referencing satanic rituals and sex cults is like a psychologist referencing the “Niceness Gene” to explain some behavior. It isn’t impossibly wrong, but it strongly suggests the expert doesn’t know what he is talking about.
This is the expertise problem. You and I do not have access to all the evidence, so we look to the experts. If anyone is an expert on the strength of the case, it is the prosecutor.
So in your view, any probability assessment of Knox’s guilt which does not put a lot of weight on the prosecutor’s credibility is deeply flawed. Agreed?
No. The physical evidence does not support Knox’s guilt. But the fact is that the prosecutor and other Italian authorities are engaging in a lot of motivated cognition to support the theory of guilt even when the evidence doesn’t support it.
The motivated cognition is so blatant that it persuaded me that Knox was probably not guilty even before I looked at the evidence. I suspect that if the world were more rational, my reasoning would be less likely to provide useful insight.
No. The physical evidence does not support Knox’s guilt.
How do you know this if you do not have access to all of the evidence and therefore must “look to the experts”?
It seems to me that you and I both have access to enough of the evidence to make a decent estimate of the probability of Knox’s guilt without relying on someone who has superior access to the evidence.
My estimate of how ethical it is for a prosecutor to propose theory X about a defendant’s motive increases with how common X is, and decreases with how sensational X is. Roommates arguing over cleaning an apartment is both more common and less sensational than satanic rituals, so my estimates of how ethical it is to propose those theories differs.
Of course, maybe there’s some particularly salient threshold that both examples exceed. If so, it’s worth articulating that threshold explicitly… when it comes to propounding theories about motive, it’s not clear to me what that threshold is or ought to be.
I’ve had no interaction with you on this site at all, but I have read your posts on the previous Amanda Knox threads, and while I believe I have a far greater aversion than wedrifid to making statements so likely to antagonize others, I have to say I find your judgment in this case in conjunction with your position as a lawyer downright frightening.
Well what exactly frightens you? I’m not a judge.
Judges are mostly selected from among lawyers, so that would be a lot more comforting if I were confident that the selection process were a genuinely good filter for people of exceptional judgment. But I would have a lot more trust in our justice system if I thought that lawyers tended to be people who would not readily become convinced of and argue strongly for positions in the absence of good reasons for believing them true.
At least in the United States, judges are mostly selected from among prosecutors. Defense attorneys, including public defenders, aren’t very well represented on the bench. General judgment aside, this a serious systemic bias of the system.
What mechanism makes it a problem? I can see why it would perhaps be most fair for an individual to have one of a former prosecutor and former public defender presiding over the individual’s original trial and appeal. The more judges make binding rules for the system, rather than in their courtroom, the more of a problem this would be (though I am under the impression this is not the case).
But why are individual differences in judicial bias OK?
I see it as a problem in the sense that prosecutors and defenders are conditioned to view the judicial process in somewhat different ways, and I don’t think that judges can simply wave away years of conditioning to see things from a prosecutorial point of view.
Individual bias isn’t great either, but at least there the bias isn’t necessarily going to be in the same direction at each step of the process. The fact that the trial judge and all the appeals court judges are all likely to be former prosecutors, on the other hand, cuts the same direction at every step.
I think this is a legitimate but small concern.
If half of all judges at each level were former prosecutors and half former defense attorneys, half of those who had a trial and one level of appeal would face the same bias at each step (assuming even promotion, and unless this was specifically corrected for).
A more fair system than evening out representation would be embracing a type of bias and having formal rules counteracting that, e.g. have all judges be former prosecutors and have a set of judicial ruels favoring defendants.
I think it likely you are being blinded to the gross unfairness of having an individual tried under a former defense attorney when the rules are written with no regard for bias or assuming the average judicial bias is in favor of the prosecutor, or under a former prosecutor even if half of all judges are former defense attorneys, by the legitimate point that fairness would be increased if each defendant had at most one of each type between trial and appeal.
This is especially true considering how much less important appeal is than an original trial.
Only mostly? I had assumed it was an actual legal requirement. That’s interesting. Where you come from how many judges have ever not been lawyers and how on earth do they know what they are doing?
EDIT: From the looks of it some (40) states in the US allow non-lawyers to be low level judges, usually for small towns doing straightforward cases. From what I can tell in Australia (and most comparable countries) a law qualification of some sort is required.
I don’t know if there are any, but given the sheer number of judges, I would suspect that there have been judges who have never served as lawyers; there is no requirement that a federal judge have ever served as an attorney, and requirements for state judgeship vary by jurisdiction.
The Bureau of Labor Statistics states that a bachelor’s degree and work experience are the minimum requirements for judgeship or a magistrate position, but most workers have law degrees, so I’d take it as implied that some do not, and are thus exceedingly unlikely to have been lawyers.
It’s rare now, but used to be common, for prospective lawyers to pursue an apprenticeship rather than a law degree. You can still use either as a qualification to take the bar exam.
That depends on the jurisdiction, and also on how exactly you define “lawyer.” (Do you mean someone with a law degree, or a member of the local bar association or some equivalent guild? The former is usually, but not quite always, a requirement for the latter.)
If anything, in many state and local jurisdictions within the U.S., judges are elected by popular vote, and I would guess that in some of those there are no such requirements for candidates, at least in theory.
Interestingly, for a U.S. Supreme Court appointment, a law degree from a top 14 law school (and at least one academic degree from Harvard or Yale) has been a de facto requirement for decades, but as recently as the 1930s and 1940s, there have been occasional SCOTUS justices appointed without a law degree at all.
Apparently the last United States Supreme Court Justice without a law qualification was Robert H. Jackson—although he passed the bar exam without official training and was a prominent practicing lawyer. I haven’t found the last time someone was appointed to that role without passing the bar but research so far does seem to suggest it is an entirely political position, without qualifications required.
They’ve all been lawyers, it’s just not an official requirement.
Presumably for the same reason there’s technically no official requirement that they be human, either.
That is, those with the power to appoint such judges would look like tools if they voted in non-lawyers or non-humans so they will not do either.
(It occurs to me there is a mutual exclusion joke in there somewhere.)
It actually isn’t a requirement for Supreme Court justices—I’m not sure about other cases.
I see your point, but I suspect the problem is more in your own judgment than in mine. Consider that I have had the experience of being wrong on these sorts of issues—and having to face it—many many times.
In this case, knowing that you persist in your assignment of a high likelihood of guilt for Knox and Sollecito given the data that’s now available to you, I feel confident in saying that your ability to say oops is too poor.
After the advent of DNA testing, retrospective analysis of many crimes exonerated people who had previously been convicted. Had this case occurred before DNA testing was available, Knox and Sollecito would most likely have been convicted, but a DNA test should have been sufficient to exonerate them.
FWIW that assessment happens to be incorrect. I say “oops” regularly. And even in my professional life where there is a lot of money at stake, I occasionally have to withdraw a case and apologize. Or ask for leave to amend a paper.
Exactly what DNA evidence do you believe exonerates them? I am not aware of any, but I am happy to consider it in good faith. And yes, revise my probability estimates accordingly.
My understanding is that some doubt has been cast on the DNA evidence against the two. Which isn’t the same thing as being exonerated. But I really would like to hear about such evidence.
The DNA evidence showing that Guede was the killer.
This is exactly the sort of thing that gets people exonerated under the Innocence Project. The only difference is, in most cases, the authorities don’t usually go back to the scene desperately looking for new evidence to incriminate the original suspect.
I completely disagree. As far as I know, Guede’s fingerprints were found at the scene of the crime. So that if years later, his DNA was found to have been there, it should not have much of an effect of peoples’ assessment of the case. It would be a surprise if his DNA were NOT found at the crime scene.
Wait, what? You’re saying that the case against Guede is even stronger than I implied, and yet you’re disagreeing with me?
The evidence against Guede—of whatever type—is what exonerates Knox and Sollecito. The more evidence against him, the more exonerated they are.
If, years later, DNA and fingerprints were found from someone other than the person convicted, that would make it even more likely for the conviction to be overturned.
Only if you substantially rule out the possibility that all 3 were involved in the murder.
ETA: Besides, the claim on the table seems to be that if you look at the case without DNA evidence; and then ADD DNA evidence, it would substantially reduce, basically to zero, the assessed chances that Knox and Sollecito were involved in the murder. Agreed?
Why should we privilege that hypothesis?
No. Or at any rate, that’s not an interesting claim, so I don’t see why it should be on the table.
So you are saying that there are only two choices: (1) privileging the hypothesis; and (2) substantially ruling it out?
If it’s not interesting to you, then probably you should not have tried to answer my question. Here is the claim which was made:
I asked this:
And you said this:
Since Guede’s fingerprints were (apparently) found at the scene, extra DNA evidence which confirms his presence at the scene should not have much of an effect on anyone’s assessments of the likelihood of any conclusion.
In a word: yes.
The hypothesis in question has low prior probability, and hence is “substantially ruled out” by default, unless there is strong evidence pointing to it. In this case, there isn’t; people who think there is are, almost invariably, anchoring on the probability that they assigned (or that the investigators assigned) to Knox and Sollecito’s guilt before Guede was discovered.
On the contrary, I inferred from your question (correctly) that you didn’t understand the point about the evidence against Guede exonerating Knox and Sollecito.
Most likely, the person who made the above claim simply wasn’t aware of the fingerprint evidence (or, possibly, the difference between fingerprint evidence and DNA evidence). By the principle of charity, that shouldn’t have stopped you from understanding and addressing their real point, which was that Knox and Sollecito should have been dropped as suspects as soon as the physical evidence turned out to implicate Guede and not them.
I had, in fact, forgotten about the fingerprint evidence, but I stand by my claim that Knox and Sollecito would probably have been convicted prior to DNA testing, given that even with DNA testing they were only released on appeal, and with DNA testing we have a much higher burden of expectation of evidence to meet. To say that they probably would have been convicted does not and was not intended to mean that they should have been convicted given the state of evidence that would have been available in that case.
Prior to what? Prior to finding the evidence against Guede or prior to considering any evidence at all?
I do understand the point, I simply happen to disagree with you at the moment. Anyway, if you want to address a different issue from the one under discussion, it would be helpful if you made it clear from the beginning.
I think that’s too much of a leap. Charity must be balanced with the principle that it’s often a mistake to try to put words in peoples’ mouths.
Did you read the independent experts’ report that Komponisto linked to?
(Er, not to sound vain, but my collaborator and I did a bit more than link to the report! ;-) )
No I did not. As noted above, the main new evidence I have is the fact that the convictions were thrown out.
Would you mind quoting the part of the report which talks about the exonerating DNA evidence?
Each component of the report in which they retested the samples (knife and bra clasp) for biological evidence of Knox and Sollecito returned negative results, as detailed in the conclusion. Having found specific reasons to doubt the results of the Italian crime labs’ testing, the retest found no positive evidence associating Knox or Sollecito with the crime, and this is a case where absence of evidence is significant evidence of absence.
Let’s make sure I understand your argument: You seem to be saying that the (apparent) lack of DNA evidence on the knife and bra clasp is convincing evidence that Knox and Sollecito were NOT involved in the murder.
Do I understand you correctly?
There was an abundance of physical evidence of Rudy Guede on the scene. If Knox and Sollecito had any physical involvement with the crime, they should have left biological evidence on a similar order, but in fact tests for evidence of their involvement did not return positive results except when corrupted by serious mishandling and poor testing procedures. The knife and bra clasp were major pieces of the prosecution’s case because they alleged that they held genetic evidence implicating Knox and Sollecito, but in fact they did not, and even if they had carried evidence of Knox and Sollecito, it would have been a suspiciously small amount of evidence compared to what one would reasonably expect had they actually been involved.
Think of the allegation of Knox and Sollecito’s involvement in the murder as a claim that there is an elephant in a room. Another party examines the room and finds...
X: Well, there does seem to be an elephantish sort of smell in the room, but it’s otherwise devoid of large mammals.
Y: But you agree that it’s a distinctly elephantish sort of smell?
X: Yes, it does smell much more like an elephant than anything else I can think of.
Y: In that case, the presence of an elephant makes it far more likely for you to observe this smell than the non-presence of an elephant, therefore it meets the bayesian definition of evidence, so you must revise your belief in the presence of an elephant upwards!
X: But hiding an elephant in a room is very difficult. If there were actually an elephant in this room, I would expect a lot more evidence than that. I can’t think of any plausible way that you could get an elephant in here and hide every sign of its presence but its smell, only very implausible ones. Your claim was sufficient evidence to promote the hypothesis of an elephant in this room to my attention in the first place, so this is less evidence than I need to maintain my prior uncertainty. To maintain my prior uncertainty, I would need to observe evidence that gave significant reason for suspecting the presence of an elephant, such as this smell, while finding some sort of arrangement that could plausibly hide an elephant. This, on the other hand, is a small enough amount of evidence that I must revise my confidence to near certainty against.
It is reflective of a suspiciously poor case that the prosecution brought so little biological evidence against Knox and Sollecito compared to the one person we can be almost certain was actually involved. The pieces of evidence were mishandled, making them suspect to begin with. This provided an opportunity for us to make our beliefs pay rent. Given the state of the evidence against them, most of us on this site concluded that Knox and Sollecito were probably not involved in the murder, so the biological evidence brought forward by the police was probably not contingent on their presence at the crime, but due to subsequent corruption and mishandling of the evidence. Thus, we could predict that the independent experts’ review of the evidence would find that when properly processed, it did not point to Knox and Sollecito’s presence at the crime. If the evidence were contingent on their presence at the crime, we would not expect to observe this.
So do you concede that your high confidence in their involvement did not predict observations as well as our belief that they were probably not involved?
Well I’ll consider this argument, but let’s make sure we are on the same page about biological evidence. According to one web site I found, Guede’s DNA was found (1) on a swab of Kercher’s privates; (2) mixed with Kercher’s blood on Kercher’s handbag and the left cuff of her sweatshirt; and (3) on toilet paper in one of the bathrooms in the house.
Do you agree with that?
Not yet as I am skeptical of your argument that involvement by Knox or Sollecito would have most likely resulted in the same kind of biological evidence as there was implicating Guede. But I’m willing to consider your argument and I admit that I have not heard it before today.
Let’s start by making sure we agree about the DNA evidence against Guede.
That reflects my understanding of the state of the DNA evidence against Guede.
Keep in mind that that does not imply that this is the sum of all DNA evidence Guede left on the scene; when you have already found this much biological evidence, there is no reason to continue searching in order to obtain a comprehensive inventory of every trace left by the suspect.
Well, I’m not an expert on DNA evidence but just based on common sense, it seems to me that if playing the primary role in a murder and sexual assault reasonably leaves 2 DNA traces in a person’s blood stains a 1 trace in the person’s privates, it’s plausible that playing a secondary role might very well leave no DNA traces in the person’s blood and none in their privates.
Of course the DNA traces in other parts of the house are far less interesting since Knox and Sollecito can be expected to have left DNA biological evidence there regardless of whether they played some role in the murder.
So I remain skeptical of your argument but I am happy to consider an authoritative source which says that even someone who plays a secondary role in a murder with a knife is very likely to leave biological traces behind. Do you have such an authority? Or are you just going by your own general knowledge and common sense?
I disagree. The authorities had every reason to continue searching for DNA evidence since they suspected others besides Guede.
The prosecution claimed that more than one person must have been involved in the murder because, they alleged, Kercher was stabbed more times than was likely by a single assailant, and the wounds were consistent with the use of more than one knife. However, the latter part has been challenged, and I am myself aware of murders known to be committed by a single person in which the victim was stabbed more times than Kercher was. The allegation was a particularly ignorant one in the first place. However, assuming that it was true that more than one person was involved in stabbing Kercher, then even if he was the only one who sexually assaulted her, then from my own knowledge of forensics I can say that it is extremely unlikely that only one person would have left biological traces on her unless everyone else was wearing extensive protection. Keep in mind that the contrast is not 3 traces versus zero traces, but three distinct samples which have considerably more than the threshold needed to be detectable by the forensic processes versus no traces above threshold level; if anyone else was involved in physically stabbing Ms. Kercher to death, as the prosecution claimed, they would have to have done so while leaving behind no traces within orders of magnitude of what Guede left.
I have studied enough of forensics and biological labwork to make more than a common sense guess in this matter, but if my say so isn’t enough for you, I could email a forensics specialist, although you could do the same. I’m not aware of any website that explains this, although that doesn’t mean that that there are none.
The traces of Guede’s DNA were acquired by a simple procedure; find a spot likely to yield biological evidence, take a sample, test it. A swab from Kercher’s genitals, for example, contained DNA from Guede. They did not test to see “let’s see how many swabs we can take which register positive for Guede’s DNA,” because it’s unnecessary.
You seem to be treating the distinct samples as if they were individual hairs, where a person could plausibly leave behind a few or none at all, but that simply isn’t an accurate model.
None of that, of course, was the real reason. The real reason was that they had initially arrested Knox, Sollecito, and Lumumba on the basis of crackpot psychological theories, and had egg on their face after having to release Lumumba when he turned out to have an unimpeachable alibi; so they needed Knox and Sollecito to still be guilty in order not to look like the complete morons that they, in fact, were. The bottom line was already written, and it was motivated searching from that point on.
For what it may be worth, I don’t think this is a completely accurate summary of the prosecution’s argument. For example, it seems that the prosecution argued that the lack of defensive wounds suggests multiple attackers. Also, the fact that she was (allegedly) stabbed on both sides of her neck.
How much above the threshhold were each of the 3 samples which matched Guede?
Well what’s your job; educational background, etc?
I’m not sure I understand your point. Do you agree with me that this sampling would have been done in batches? Do you agree that Knox and Sollecito were suspects almost from the very beginning? Do you agree that therefore the authorities would have had reason to continue collecting DNA evidence even after they found crime scene samples which clearly matched Guede?
Had she been physically restrained by anyone else, we should strongly expect that person to have left biological traces. Besides, it’s easy to stab someone in both sides of the neck without switching hands. Like the claim that she was stabbed too many times for one person to account for, this sounds a great deal like reaching to rationalize a story.
I don’t know, although from the fact that all sources, including the Massei-Cristiani report, agree that there was an “abundance” of genetic evidence, I can infer that the traces were well above the limit of quantification, whereas the alleged genetic material from Sollecito, when tested by the police, was amplified by a procedure which increases sensitivity by more than an order of magnitude above the base level of detection. This places a ceiling for the amount of genetic material attributed to Sollecito of about one thirtieth of what was contained in any of the samples from Guede, but that’s assuming that Guede’s samples were only slightly over the limit of quantification. To be described as an “abundance”, there was probably many times more. And the independent experts’ report determined that the DNA from Sollecito, along with three other people on the sample, was probably due to corruption of the evidence well after the crime was committed.
Amanda Knox’s DNA was found in easily measurable levels on a knife that, independent report determined, had no traces of blood on it and could not have been used in the murder at all; it was improperly flagged as a piece of evidence in the first place.
I have a Bachelor’s in Environmental Science, currently pursuing credentials to become a high school chemistry teacher. In the meantime, I tutor.
Yes, yes, no. The sampling was done in batches, but not in a comprehensive “let’s test everything in the room at once” sweep. Knox and Sollecito were suspects from near the beginning, but it was on the basis of a scenario that was quickly proven to be wrong. Once it was proven that the person who they concluded that Knox’s testimony had implicated could not possibly have been involved, they had practically no case for suspecting Knox and Sollecito at all. What they did have was a great deal of motivation to find evidence implicating Knox and Sollecito, because otherwise they would look and feel like idiots.
When they did the initial sampling and found multiple traces of evidence from Guede, they had already reached a state where it was unlikely that if someone else had been physically involved, they would have missed the evidence of them. They should have revised downwards from the prior likelihood that any randomly selected murder would be committed by multiple people. They had obtained evidence such that the hypothesis of Knox and Sollecito’s involvement should have been discarded, given that under interrogation they had failed to implicate the one person who was definitely involved, the other person “implicated” by their testimony had been conclusively proven not to be there, and they had not found biological evidence of their involvement under an initial search that would probably have found it if there were any. At this point, their initial scenario was effectively disproven beyond reasonable doubt, but still they insisted on continuing to search for evidence implicating two people they now had every reason to believe were innocent.
So you agree then, that you mischaracterized the prosecution’s argument?
I’m not sure I understand your point. You seemed to be arguing that the amount of DNA traces left by Guede were “orders of magnitude” above the detection threshhold, i.e. at least a hundred times higher. And that if Knox or Sollecito had been in physical contact with Kercher, they would have left at least a hundredth of the DNA left by Guede—enough to have been unambiguously detected.
Did I misunderstand your argument?
Ok, and assuming that’s true, they would have had a strong motive to do more swabs right? And if they in fact did so, then under your hypothesis, they should have found additional Guede DNA material, right?
And what classes did you take which gave you your expertise in forensics?
Not in any way that makes their case seem weaker than it actually was.
The traces left by Guede were all above the limit of quantification. The only piece of evidence on the scene that contained traces above the limit of detection for anyone other than Kercher or Guede was the bra clasp, which contained DNA from Kercher as the major contributor, and DNA above the limit of detection but below the limit of quantification for several other people, according to the independent experts’ report, DNA which they claim was most likely due to mishandling and contamination. That leaves no traces above the limit of detection from any uncontaminated source for any potential perpetrator other than Guede, so if we are to posit that anyone else was involved, we must suppose that they did it while shedding orders of magnitude less biological material.
By the time they went to retrieve further evidence, Meredith Kercher had already been buried. Her body, along with the objects on which they had found traces of Guede’s DNA, had already been removed. They did not re-swab her body to see if they could get more of Guede’s DNA off of it, nor did they attempt to see if they could get more traces from any of the other articles on which they had already found his DNA. Had they scoured the scene for more traces of Guede, I would be surprised if they did not find any, but they weren’t looking for more traces of Guede, they were looking for traces of Knox and Sollecito. Further traces of Guede found on the scene 47 days later might not even have been kept due to their lesser value as evidence since there was already plenty of opportunity for corruption and the case against him was rock solid; if they found, say, small traces of his DNA on the floor on the reexamination, they’d have no particular reason to make note of it.
Two semesters of biology labs covered forensics. My labs in atmospheric and environmental chemistry also dealt with collection and analysis of trace chemicals from a scene.
Umm, does that mean yes or no?
Umm, does that mean yes or no? And what are the numerical limits for quantification and detection?
Umm, does that mean yes or no?
But when you “scour the scene,” there’s no way to narrow your search to DNA of a particular individual. Because the matching is determined back at the lab. Agree?
If you claim that there is likely other DNA evidence against Guede which was discarded, then I will need a cite. I have an extremely hard time believing that the lab would have tested a sample; realized it was Guede DNA; and not made a note of it in the record which would have come into the trial materials.
I don’t think that’s enough for your opinion to simply be accepted on the basis of authority. Is there no reference online for your claim that secondary participation by Knox or Sollecito—such as grabbing or holding—would have left plenty of DNA material which the Italian authorities would have been very likely to find?
At a minimum, if what you are is saying is true, then surely one of the defense experts would have testified at trial that Knox could not have been a participant because none of her DNA traces were on or about Kercher’s person. Agreed?
For practical purposes, no.
It means that I do not strongly expect that there would be anything left on the scene which they would file as further evidence against Guede.
Yes and no. If they already have already established previously that person X has come into contact with objects X Y and Z, further testing of objects X Y and Z is likely to reveal further traces.
For the latter last two points, I do not have a strong expectation that further evidence against Guede would have been brought to court given that later retrieval would have been less reliable, and the case against him was already watertight, nor do I have a strong expectation that the defense would have claimed that Amanda Knox could not have been involved based on the weakness of the evidence of her involvement.
If we’re going to go on though, let’s review.
Meredith Kercher is killed. Knox and Sollecito are waiting outside when the police arrive to investigate the residence, and Edgargo Giobbi tags Amanda Knox as a suspect within hours of the crime’s report because he thought the way she swiveled her hips while putting on her protective shoes at the crime scene was unusual. Edgardo Giobbi claimed in a later interview
Having established suspicion, Giobbi now had motivation to find more reasons to suspect her. On the basis of her mannerisms being judged unusual, she was brought in for interrogation, along with Sollecito, who she had claimed to be with that night. She alleges that she was badly mistreated, which the police deny. She was not told her legal rights, provided with a lawyer, or given an official interpreter. The police told Knox that Sollecito had changed his story and no longer said that he was with her that night at the presumed time of the murder. She was persuaded to sign a statement implicating her boss Lumumba as the killer.
After giving her statement implicating Lumumba, Amanda Knox was allowed a break to receive coffee and food, and then retracted her statement within hours, saying that she had been with Sollecito as she originally claimed, and had no knowledge of the killing.
Lumumba was arrested and held for two weeks, but produced a solid alibi. Crime labs matched the print found at the scene to Rudy Guede, police substitute Guede for Lumumba in their crime scenario, while retaining Knox and Sollecito as suspects, still purely on the basis of psychological profiling (keep in mind that they did not observe everybody else Kercher had known to see if anyone else also had suspicious mannerisms, with blinding to the hypothesis so the investigating police did not know whom they were supposed to find suspicious.)
Meanwhile, Guede, the only person with hard physical evidence placing him at the scene of the crime, fails to implicate either Knox or Sollecito in his testimony.
The police continued to search for evidence against Knox and Sollecito (and I will emphasize again, did not search to see if a comparable degree of evidence could be raised against anyone else Kercher had known) and produced a knife found in Sollecito’s kitchen, which they claimed contained traces of Kercher’s DNA despite her never having been in his kitchen. The evidence was improperly tested, and the independent experts’ report found that the knife could not have been used in a murder at all.
The police returned to the scene 47 days after the murder, and searched for items which could produce evidence against Knox and Sollecito. They claimed to have found evidence on a bra clasp which had been cut from Kercher in the attack. Since the date of the first investigation, the investigators had moved the clasp four feet across the room and left it under a rug with a collection of other items. The independent experts’ report found that the clasp predominantly contained traces from Kercher, but had small traces of DNA from four other people, one of whom was Sollecito. The report found that the evidence had been mishandled and probably corrupted and that the police did not follow proper testing procedures.
At this point, what probability do you assign to the proposition that Knox and Sollecito were involved in Kercher’s murder? Do you still think that focusing a search for evidence against Knox and Sollecito in particular, beyond everyone else who could potentially have been involved in the crime scene, after they and the only person with forensic evidence placing him at the scene had mutually failed to implicate each other, was reasonable?
I’d like to hear people’s probabilities for Brazil revising his estimate down below 50% in the next year.
This is interesting. But I can see someone seeing it as rude to speculate on the chance that someone else is going to change their mind. In some contexts this could come across as condescending and can easily make worse feelings that a discussion involves others who are opponents rather than people mutually trying to get a better understanding of reality.
With that disclaimer out of the way: I’m estimate an 8% chance that Brazil will reduce his estimate below 50% in the next year and publicly say so on Less Wrong.
Mine is 2%.
Hmm, from the look of things, it might be time to update the model...
Whoa! Indeed :-(
Edit: In fact not my prediction stands.
I think you are the one who has a bit of a problem saying “oops.” Please just admit that you misstated the prosecution’s argument. I promise you the sky won’t fall as a result.
I do not understand what your point is here. Maybe it will help if we quantify your argument:
First, what are the numerical thresholds involved for the amount of DNA required for detection and identification.
Second, how far above that threshold was Guede. Please answer the question with a rough percentage.
Third, what is the minimum amount of DNA one would reasonably expect to be left by Knox had she participated secondarily in the murder?
Further traces of Person X, right?
Anyway, please give me a cite for your (apparent) claim that other DNA evidence against Guede is likely to have been discarded. Or are you just speculating?
Again, does this mean yes or no?
Look, you are presenting what is clearly a very solid argument for Knox’s innocence if your assumptions are correct.
Your argument is as follows:
(1) If Knox had been involved in the murder, there would have been DNA traces of Knox on or about Kercher’s person which the authorities would have been very likely to find.
(2) There were no such traces found.
(3) Therefore Knox was very likely NOT involved in the murder.
If (1) and (2) are true, (3) clearly follows. And there doesn’t seem to be any dispute about (2). Further, you assert (1) based on your expertise with forensic DNA testing. Surely Knox’s experts are at least as expert as you. So if (1) is true, then Knox’s experts would have almost certainly made the this point..
Agreed?
I agree that the prosecution made claims that I did not relay. I also did not relay that Giobbi originally flagged Amanda Knox as a suspect based on what he considered to be unusual mannerisms while putting on her safety shoes. I am happy to admit that I misstated the prosecution’s argument, absent the implication that I omitted only information that strengthened their case. I do not agree that I understated the strength of the prosecution’s case.
I do not know how far above the threshold for detection of DNA Guede’s samples were; I can make inferences from the characterizations of the evidence, both from the prosecution and defense, that they were above the limit of quantification (which as a general rule is about three times higher than the limit of detection,) without using the same amplification procedure that was used on the bra clasp, which would further increase the sensitivity of that test. However, the reports which would give the quantitative values for the traces he left relative to the limit of detection, if they are available to the public at all, are in Italian.
I am prepared to accept if you do not consider my judgment in this case sufficiently authoritative; I can respect your suspicion on that count. We could suspend this discussion in order to contact a forensics expert and ask for an estimate of the likelihood that a person could be involved in physically grappling and stabbing a victim with a knife without leaving identifiable biological traces on the scene.
However, I think you would be guilty of motivated continuing, demanding evidence far in excess of what you need to alter your decision.
I affirm that I agree with statements 1, 2 and 3. However, I am not sure of the level of forensics expertise of Knox’s defense; the independent commision certainly possesses greater expertise in forensics than I, and they affirmed that there was no reason based on the forensic data to suspect Knox and Sollecito’s involvement. As for her defense, I know that they were able to take issue with the handling of the evidence by the police, but they could have done this by investigating “what are the procedures that a forensics specialist is supposed to take in such a situation, and did the police follow them? The case of the prosecution sets a low floor on how much familiarity I must assume that a trial team in a case dealing with forensics would have to have, so I have a high degree of uncertainty regarding the expertise of the defense. Further, despite being simply wrong, “absence of evidence is not evidence of absence” is a sufficiently pervasive truism (what do you call it when people treat something false as a truism?) that I do not have a very strong expectation that, even given a high level of expertise in forensics, the defense would present an absence of evidence as evidence of absence, rather than simply arguing that there was no reason to believe in their guilt (which the defense did.)
Will you provide your current judgment of the case based on the information I provided in my previous post? That information I can provide citations for, so we won’t have to stall waiting on further expert input.
I have considered your argument in good faith and I reject it. You have not been able to answer simple, reasonable questions to let me understand and evaluate your argument. You have also evaded and ignored simple, reasonable questions.
Most importantly, you have not been able to provide an authoritative cite for your implicit claim that secondary involvement by Knox would have very likely resulted in Knox leaving DNA evidence which would very likely have been found. It’s also strange to me that I have not found prominent mention of your argument anywhere, especially since it would have been available to Knox’s defense (more specifically her experts) from the get-go.
In short, your argument has not stood up to mild scrutiny. However, if you can provide answers to my questions I will reconsider.
It is my general experience that people provide arguments of the form “if X, we should strongly expect evidence Y, we do not observe Y, therefore we should strongly disbelieve X” far less often than they are applicable, which is why I do not have a strong expectation that it would have been used in this case contingent on its applicability.
I do not evade the question of how much more of Guede’s DNA was found than is necessary to meet the detection limit, I affirm that I do not know, although I maintain that I have enough information to infer that it was much more than any other potential participants must have left.
At this point, I consider you to be evading the questions of whether the evidence you already have available to you supports a belief in Knox and Sollecito’s guilt, and whether you still feel that the police were justified in holding them as suspects after they received the evidence implicating Guede. I see three ways this discussion can go from here. We can suspend it, and write to an independent expert in forensics, and thus establish whether a person with acknowledged expertise in forensics would agree with my judgment. We can continue based on the information for which we already have adequate citations, and you can provide your judgment based on that information, or we can abandon it as fruitless.
That’s not my experience. In fact, the pro-Knox page linked to in the very first thread explicitly makes such an argument in general terms (not specific to DNA).
Can you please quote the question I evaded?
It’s up to you. I’ve considered your argument in good faith and rejected it. But the door is still open if you want to provide the information and citations I requested.
From this comment
If you do not trust that I have sufficient knowledge to support the argument that the state of the biological evidence in this case more or less completely rules out the possibility of Knox and Sollecito’s involvement, I am prepared to accept your discounting that argument (although I find your belief that the defense would necessarily have used it if it were valid naive, if arguments of that form were used as commonly as they merit, “absence of evidence is not evidence of absence” would not have become a popular saying in the first place.) We can of course try to contact an independent expert, but is your openness to the likelihood of Knox and Sollecito’s innocence contingent on that particular proof?
I think it’s pretty clear that my assessment of the probabilities has not changed after having considered your argument, but let me make it perfectly clear: my assessment has not changed.
Yes.
I’ve already stated that your argument is very solid if your assumption is correct. So I am conceding in advance that if you can support your assertion, my probability assessment will change a lot.
Can you give a numerical probability? You claimed that your assessment had shifted somewhat in light of the information that Knox and Sollecito had been found not guilty on appeal.
I’m going to do a run down of my probability estimates for Knox, Sollecito, and Guede, given the evidence available to me excluding the argument I’ve already given for the unlikelihood of Knox and Sollecito’s presence at the scene given the state of the forensic evidence. Could you explain what, if any of it, you disagree with, and how your own analysis would differ?
Prior probabilities. Common wisdom is that most people are killed by someone they know. I’m going to go with 80% likelihood of murder by an acquaintance, although I can’t actually find a single figure all sources agree on. So we’ll weight the prior probability of guilt such that people Kercher knows account for 80%, and everyone else accounts for 20%. Unfortunately, I can’t find a good source estimating how many people most people know, and it depends how close a standard of acquaintance you’re measuring. I’m going to give a ballpark estimate of about 1000, and thus apportion an average of .00125 prior likelihood of guilt between them.
For Knox, let’s adjust up for closeness of acquaintance (we’ll call it a factor of 10:1) for a likelihood of .0125. She had no prior arrest record; I’ll adjust down by 70 percent to .00875, and she was a woman who was neither a close family member nor an intimate partner, so I’ll adjust it down by another 90%, to .000875. I don’t feel that the prosecution established any good reason to suspect a motive, but I’ll be charitable to their case and not revise further down for that.
I’ll call Sollecito’s probability .0007, adjusting upwards for his increased likelihood as a male, but downwards based on his much lesser closeness to the victim; I have to take probability mass away from slight acquaintances if I’m adding it to close ones.
Guede was male, only slightly acquainted with Kercher, had been involved in breaking and entering prior to the case, and also did not have a clear motive. People with prior criminal records account for a majority of violent crimes despite being a minority of the population. The population without a criminal record is giving up a majority of their probability mass to the minority with, so this is a pretty significant boost. I don’t know the proportion of Italy’s population that has a record, but if we assume similar rates to the U.S. this will increase Guede’s likelihood by about 11 times. So let’s call his prior probability about .0077.
There’s a considerable amount of fudge factor in these numbers, and you are of course free to take issue with them, but I think they give us a reasonable ballpark order to start with.
First, we’ll deal with Guede. His handprint in blood places him conclusively at the scene of the crime (I doubt that anyone has ever left a handprint in blood at a scene at which they were not present) and he does not deny having been at the scene, so we’ll call the odds of his presence infinitesimally below 1. He changed his story repeatedly over the course of his interrogation, beginning with “another anonymous black man did it while I was in the bathroom after having sex with her.” The prior likelihood of this is very low, and Guede left DNA traces he was unlikely to have left if he was not involved in killing her. I’d adjust Guede up from the prior by an order of about 10^7, which would leave a likelihood of guilt that approximates to 1-(10^-5).
Let’s move on to Amanda Knox. Giobbi first flagged her as a suspect based on the belief that her mannerisms at the crime scene were odd. I expect that a person is significantly more likely to behave oddly at a crime scene (relative to the norm for an uninvolved person) if they were involved than if they were uninvolved. I’d put an odds ratio of maybe 15 to 1 on it, except that Knox was already known to have odd mannerisms (such as doing yoga stretches in places considered inappropriate,) so I’d revise that down to maybe 5 to 1. So that gives us a probability of involvement for Knox of about .0044. All the evidence brought forward by the police after this point is filtered evidence, because they focused their investigations on Knox and Sollecito, and put more effort into finding reasons to suspect them than anyone else who might have been involved.
In the United States in 1976-1996, 16% of murders were committed by more than one person; since I don’t have any better figures I’ll assume similar rates for present day Italy. Ignoring the failure of forensics to turn up evidence of anyone else, I find the arguments posed by the prosecution for believing that the murder had multiple perpetrators extremely uncompelling. Arguments such as the number of stab wounds and the fact that Kercher had wounds on both sides of her neck are extremely weak evidence; particularly taken alongside claims such as one which they offered for the break-in being staged, that the pattern of the broken glass was inconsistent with a break from outside, which was tested and found to be false, these arguments sound to me like scrambling for anything that sounds like evidence. Ignoring the failure of anyone else to leave biological traces, I’d be willing to extend perhaps a 3 to 1 odds ratio on their evidence for multiple attackers, so together with the prior unlikelihood of a murder having multiple perpetrators, that brings Knox’s likelihood of guilt to about .0022.
Knox and Sollecito failed to implicate Guede, and Guede failed to implicate Knox and Sollecito; for their mutual failure to implicate each other, I’d reduce Knox and Sollecito’s likelihood about 20 times, bringing Knox to about .00011.
I do not find any evidence presented by the prosecution compelling with regards to establishing a motive or likelihood of presence at the crime for Knox or Sollecito, their case seems to me to be a collection of behaviors and actions attributed to Knox with the implication “this seems more unusual for a person who was not involved than a person who was.” Since this evidence is filtered, I only find it reasonable to update on it to the extent that I believe the body of evidence is stronger than what I should expect for anyone with my current assignment of guilt subject to a similar police investigation. I think if you really looked and sincerely believed you had reason to find anything, you could find a substantial body of reasons to be suspicious of anyone. One of the most common ways that people err is in believing that a false proposition must have no valid evidence, and every argument for increasing your confidence in it must be refutable. I would have to assume that people I know are generally significantly weirder and more suspicious than most of the population to assign an odds ratio of even 10 to 1 on the prosecution’s evidence, but my social circle probably is fairly odd, so I’ll give them that, but I think I’m being generous.
All in all, that gives approximately .001 odds of guilt for Knox. Sollecito was tagged as a suspect mainly through association with her, so I would say that he is at most as likely as she is to have been involved.
Of course, there is plenty of uncertainty in these figures, and this does not portray the entirety of the relevant evidence even ignoring the improbability of other people involved in the murder not leaving biological traces (one particular piece of evidence I even left out deliberately, you can try and guess what,) but I think it reflects a reasonable probability estimate given that omission for an amount of evidence that a person can reasonably be expected to process without a substantial time commitment.
Perhaps about 30%.
I don’t have the energy to do a point by point response, but I will point out what I think is the central flaw in your reasoning.
The flaw, it seems to me, is that you seem to dismiss the circumstantial evidence against Knox (and Sollecito) as “filtered evidence.” There are two problems with this.
First, the evidence is not filtered in the sense that the only evidence we (or the jury) saw is that which was cherry-picked by the prosecution. Knox and Sollecito both had the opportunity to gather and present evidence which would tend to exculpate them. To be sure, they did not have the same ability to get cooperation and compel the production of evidence as the authorities, but they surely had quite a lot. And they had the advantage over the authorities of being firsthand witnesses to their own whereabouts and actions.
Second, even filtered evidence can be very probative. To take Eliezer’s hypothetical a step further, imagine that a particular coin is flipped exactly 1000 times and our hypothetical advocate tells us that between 641 and 1000 of the flips resulted in heads. Even though we do not know how the actual data was filtered, and even though our advocate probably has some kind of agenda, we can be pretty confident that the coin is biased.
If we know (or can make a good guess at) the advocate’s agenda, we can get an even better handle on the situation. For example, if somebody tells you that he attends a Top 6 law school, it’s pretty likely that he goes to Columbia, NYU, or Chicago. And not Harvard Yale or Stanford.
Wait a second. At the top of the thread you said your opinion hadn’t changed. 30% from 90% is a big drop.
No, what I am saying is that if Desrtopa’s assertion is correct, then my estimate will drop from roughly 90% to roughly 30%.
How is your probability affected by what former FBI agent Steve Moore has to say?
It’s not, since Moore obviously is trying to spin the case in favor of Knox’s position. For example, he falsely asserts that there was not evidence of Knox’s presence at or near the crime scene. So I don’t put a lot of stock in his authority.
Since he is a qualified expert on forensics making essentially the same claim that I was about the state of forensic evidence, I’m curious to know what you would consider to be adequate corroboration for that point.
Why do you condition strongly on the prosecution’s arguments, when their evidence has been found in many points to be false, mishandled, or exaggerated, but not condition at all on the arguments of Moore?
I’m not sure what you mean by “condition on,” but I would be similarly skeptical (probably even more skeptical) if the prosecution assured us that it’s perfectly plausible for the secondary participant in a murder such as this not to have had DNA traces found on or about the victim’s person. Even if the assertion were made on behalf of the prosecution by a board certified forensic investigator.
By “condition on,” I mean “update your probability in light of.” To refuse to update at all, you must suppose that it is no more likely that Moore would claim that the state of the evidence effectively rules out Knox and Sollecito’s involvement if they were not involved than if they were.
So it seems that you are claiming that it is much more likely that the prosecution would produce the evidence and arguments that they did if Knox and Guede were guilty than if they were innocent (enough to multiply the likelihood of their guilt by orders of magnitude,) but no more likely that Moore, who has neither pay nor reputation riding on the outcome of the case, would claim that there is dramatically less evidence than we would expect to see in the case of Knox and Sollecito’s involvement, if they were not involved than if they were.
Would you agree with that?
You gave a probability estimate of 90% back in Komponisto’s first thread, and early in this one you said that you had revised your confidence on the fact that Knox and Sollecito had been found innocent on appeal, but now you are asserting exactly the same probability that you did back then. Why is that?
Also, I would appreciate it if you would answer my question about what you would consider to be adequate corroboration for my argument that if Knox and Sollecito had been involved we would strongly expect forensic evidence that we did not see.
Assuming that the various evidence I am considering comes from the prosecution, yes.
Here’s what I said early in the thread:
My probability estimate remains roughly 90% and I think that’s pretty clear.
Ok, now a question for you: Why are you so eager to find a contradiction in my posts?
First of all, please do not weasel, i.e. change your position without acknowledging it. Your original point was about DNA evidence.
Anyway, what would support your argument is if you can produce a quote from a textbook, police manual, or other authoritative source which says, more or less, and in general terms, that even physically restraining another person will almost always leave biological traces which can be traced back to the person who did the grabbing or holding.
I’m not sure it’s the same claim. As I recall, your claim was limited to traces of Knox and Sollecito on or about Kercher’s person. Moore’s point is more general.
But anyway, what I had in mind was a statement which is not specific to the case. For example, a quote from a forensics textbook or a police manual.
Yes, Moore’s point is more general than mine, he points to additional evidence that it had not occurred to me that we would expect. Moore’s point is stronger than mine.
I do not have a forensics textbook or police manual, but Moore probably does. If I ask him for a citation, and he provides one, would you consider that adequate?
Moore’s point is stronger if his assumptions are correct.
If it says more or less, and in general terms, that even physically restraining another person will almost always leave biological traces which can be traced back to the person who did the grabbing or holding.
And by the way, if you really plan to research this, what would also be helpful to me would be a cite for the proposition that the Italian forensics labs may have (or did) find additional material attributable to Guede which did not make it into the record.
What sort of odds ratio would you assign to the evidence presented by the prosecution given the knowledge that it’s filtered? If your other figures are not similar to mine, it would need to be more than two orders of magnitude higher than what I assign, or about 3,000 to 1.
My low weighting of the prosecution’s evidence hinges, not simply on the fact that it is filtered, but that it is not much stronger than the evidence that I would expect them to find against anyone else given a similar level of investigation. So it is not analogous to dismissal of the evidence that “coin flips between 641 and 1000 were heads” on the basis that while true, it is filtered and so can be discounted. It is more analogous to receiving the evidence “at least six of ten coin flips came up heads,” and deciding “If you had stronger evidence than that, you probably would have told me, and that is not much more evidence than I would expect to see from an average unbiased coin.”
The defense extensively pointed out the weakness of the evidence brought forward by the prosecution, they did not put similar resources to the prosecution in order to raise similar evidence against anyone else to show how easy it would be. If people sincerely want to produce reasons for believing something, they will tend to both find and overweight them. I would bet at 1:1 odds that had the prosecution investigated me with a similar level of prior suspicion, they would have found as much or more reason to suspect me, and I may be atypical, but I do not think I am 1 in 3,000 suspicious. Conditioned on the facts that Knox was already known to have odd mannerisms, and was a foreigner from the perspective of the police (one of the original pieces of evidence that the police used to narrow her down as a suspect, for instance, was that she texted “see you later” to her boss, whom she was not about to see,) I think you will need to provide some very strong arguments to convince me that you are not dramatically overweighting.
I’m not sure what you mean by “odds ratio” and I don’t know exactly what evidence was presented by the prosecution, so it’s hard to say. My assessment of 90% is based on the evidence I am aware of.
I disagree. I shared some of my arguments about my interpretation of the evidence in the last couple threads.
Are you saying that you now only estimate a 30% chance of guilt?
No, what I am saying is that if Desrtopa’s assertion is correct, then my estimate will drop from roughly 90% to roughly 30%.
If you accepted my assertion that an appropriate assignment of guilt not accounting for the lack of forensic evidence is about 0.1%, you would adjust your estimate to 30%?
No, if I accepted your assertion that even grabbing or holding someone is very likely to leave biological traces which can be identified by the authorites, then I would adjust my estimate down to 30%.
Can you explain why? That sounds like a claim that makes no sense in terms of probability theory, and I suspect that you either do not know enough about calculating probabilities to understand why it doesn’t make sense, or you are failing to apply that understanding.
We often make inferences like that implicitly, even here, revising our beliefs much less in light of new information than actually makes sense. One of our weaknesses as humans is that we will tend to stay much closer to the first number that comes into our heads than makes sense on further information. That’s why will will sometimes sanity-check our estimates by going through a numerical breakdown of what we know, like I did earlier. It gives a chance to look and see “Hey, I assigned a probability of 20%, but when I plug in the most reasonable numbers I can come up with and try and calculate a probability, I come up with .03%! To believe that the probability is really 20%, I would have to guess that this piece of evidence is a million times more probable if the belief is true than if it’s false, and that seems like way too much, or that the odds of this particular other thing happening are orders of magnitude off from what this study says, or some other combination of things that all look really improbable in light of the information I have! It looks like my first instinct was not just wrong, but very wrong, so I’d better change it by a lot.”
Before I put time and effort into providing more evidence, I think it’s important to establish that you actually know how to process evidence in a way that makes sense mathematically, not just intuitively. You can you go through a rundown of how you arrived at your numbers? I know you said before that you didn’t have the energy, but I think it will save a lot more time and energy in this conversation if you do it than if I go on trying to retrieve information for you while expecting you to update not at all or far less than makes mathematical sense.
You sure? 90% to 30% isn’t really an orders of magnitude difference. Of course, taking that into account, “beyond a reasonable doubt” should be something like “at least 99.9% probability of guilt” (or whatever incidence of innocent people you’re willing to throw in prison / execute), in which case a shift from 90% to 30% would be a shift from “not guilty” to “slightly more not guilty”.
These should really be different standards.
I don’t understand what you’re asking if I’m sure about. I’m saying that given the premise that brazil84 accepts the forensic argument against Knox and Sollecito’s guilt, it does not make sense to update his probability by so little.
Aha—I’d misread you as saying that the update was too big, rather than too little.
He could accept it at less than full confidence.
He already stated that given that he accepts the argument, he would update to 30% confidence, which conveniently frees us from the need to consider the probability outside the argument.
It’s possible that he really meant “given that I accept the argument at low confidence, as opposed to not at all, as I currently do,” but that would make his statement awfully misleading, and brazil84 has demonstrated too little familiarity with Bayesian reasoning for me to read his statements with that much charity.
He should mean “given that I assign the argument a comparable credence to other things I would state that I accept”—which shouldn’t be 1. I think this probably shouldn’t be a big enough difference to matter (I missed the “given...” in the context, in my response), but it would be interesting to update the numbers to verify.
I could try, but first please explain to me what does not make sense about my statement. If I start at 90% based on other evidence, where should I end up (in your view) if your assertion about biological traces is established, and why?
ETA: Better yet, why don’t you provide your calculation which starts at 90% and ends up at some number very different from 30%.
It depends on the weight of the evidence
For easy calculation, let’s convert the probabilities to odds. 30% is represented by odds of .43 (3:7) and 90% is represented by odds of 9 (9:1)
9/.43 is 20.9. So if you have proposition A, which you believe with 90% confidence, but then you get some piece of evidence which Not-X makes 20.9 times more likely than X does, you should reduce your confidence to 30%. If you had evidence that Not-X made a thousand times more likely than X, then that should reduce your odds to .009, which is equivalent to a probability of slightly over .009, or .9%
So we can work backwards and figure that, given the premise that you accepted the piece of evidence and adjusted your probability from 90% to 30%, then if you did the right sort of calculations you must have decided that it was a bit more than twenty times more likely that Knox and Sollecito would have failed to leave behind the forensic evidence that would implicate them given their innocence than given their guilt.
Considering the number of pieces of evidence that Moore claims that they would almost certainly have left if they were guilty, which they did not leave, I don’t think that passes a basic sanity check.
Converting to log-odds, for the practice reading log-odds units...
In bits, we have:
90% initial confidence = lg(90%/10%), which is about 3.2 bits likely 30% final confidence = lg(30%/70%), which is about −1.2 bits likely, or 1.2 bits unlikely total evidence needed to shift our view that far: 4.4 bits
In decibels:
90% initial confidence = 10 log(90%/10%), which is about 9.5 decibels likely 30% initial confidence = 10 log(30%/70%), which is about −3.7 decibels likely, or 3.7 decibels unlikely total evidence needed to shift our view that far: 13.2 decibels of evidence
Could you please just answer my question? I think it’s very simple and reasonable.
(1) In your view, if I start at 90% based on the other evidence; and your assertion about biological traces is established, then where should I end up? 5%? 1%? 88%? 0.01%? It’s a very simple question.
(2) Please show me your calculation.
I opened my response box before you edited your comment to include the question (I actually left the house in the meantime.) I think that you do not know the necessary math well enough to believe that you can defend your assertion, and so are trying to evade the question by trying to make me answer further questions and hoping that you will find something you can take issue with.
Nevertheless, I will show you a calculation. I will use what I believe represents a reasonable lower limit on the weight one could assign to the evidence. I will not respond to requests to cite unless you agree to provide citation indicating that the weight implicit in your own estimate is more reasonable.
We start with a probability of .90. I will assume that murderers leave forensic evidence with 99.9% frequency, and non-murderers have false positives with .1% frequency.
Out of a population of 10,000 people at this confidence level of guilt, we will expect 9000 murderers and 1000 innocent people. We’ll assign a 99.9% confidence that a person who is not guilty will not leave forensic evidence implicating them (with .1% chance of false positive) and a 0.1% chance that a person who is guilty will not leave forensic evidence implicating them.
Out of 9000 murderers, we will expect 9 to be false negatives (not leave forensic evidence) Out of 1000 non-murderers, we will expect 999 not to leave forensic evidence, with one false positive.
That leaves 1008 people who have not left forensic evidence, 999 of whom are innocent. So if a person has not left forensic evidence, their odds of guilt are 1-(999/1008), or about 0.9%
Edit: I actually made an embarrassing mistake in my calculations the first time around, and overestimated the posterior probability. Results now corrected.
To make myself clear, this is not the weight I assign to the evidence, I am saying that it would be unreasonable to assign less weight than this.
As I recall, the original part of my post specifically asked you to give me the probability I should end up with. The addition merely asked you to provide your calculation. You provided neither.
Anyway, the main flaw in your reasoning consists in conflating the propositions that (1) Knox (and Sollecito) were physically involved in the murder (i.e. holding, grabbing, and or stabbing); and (2) that Knox and Sollecito were guilty. For example, it’s possible that Knox and Sollecito encouraged Guede from a few feet away to attack Kercher but never actually touched her. And perhaps did not even intend for him to kill her.
Doing the same calculation as you, but substituting 95% for 99.9% (actually you seem to have used 99% in your calculation) gives 450 false negatives and 999 correct negatives. So out of 1449 people who have not left biological traces on or about Kercher’s person, 450 are actually guilty. ETA: 31%.
Anyway, I still look forward to your cites that (1) physically participating in a knife murder (i.e. holding, grabbing or stabbing) is extremely likely to result in biological traces being left on or about the victim’s person; and (2) the Italian authorities may have (or did) find additional traces from Guede on or about Kercher’s person which did not end up in the record.
Yes, I already stated before that an odds ratio of approximately 20:1 would result in a posterior probability of 30%. I have also asserted that this weight does not pass a basic sanity check. You have stated that if you accept the argument from forensic evidence, you would assign a probability of 30%. Moore, who is a forensics expert, judges that for Knox and Sollecito to have been involved in the crime, based on the state of the evidence, is nigh impossible. 5% likelihood is not a sane estimation.
I am extremely tired of looking forward to how you will attempt to evade each of my arguments and ask me to do more work in exchange for promises of “good faith consideration” which amount to looking for any possible excuse not to update. As I stated in my previous comment, I will contact an expert for citations if you agree to provide a citation for the claim that 5% likelihood of failure to leave forensic evidence is a credible assignment. One of us is asserting a confidence many orders of magnitude apart from a qualified expert,and one of us has been repeatedly demanding further evidence for the other’s confidence assertion, and in a reasonable argument, these would not be the actions of two separate people.
I’m not sure what you mean by “the argument from forensic evidence,” but let me ask you this: Do you understand that the probability that Knox (and Sollecito) were physically involved in the murder (i.e. holding, grabbing or stabbing) is not necessarily the same as the probability that they were involved in the murder?
Sorry, but you are the one who is being evasive. I explained to you EXACTLY what the flaw was in your argument, and you ignored it.
The probability that Knox (and Sollecito) were involved in the murder is necessarily greater that the probability that they were physically involved. Agreed?
And do you agree that if they were involved in the murder but did not hold, grab, or stab Kercher, then there is a good chance that they would not have left biological traces on or about Kercher’s person?
These are reasonable, yes or no questions. Please do not evade them.
Yes, but Moore also points out forensic evidence that we should expect to see if they were simply in the room at the time of the murder which we do not see. I agree that if they were involved in the murder, did not physically engage with Kercher, there is a good chance they would not have left biological traces on her person. However, if they were physically in the room at the time of the murder, there is still forensic evidence which we should expect but do not see. If we assume that they were conspirators in her murder, the forensic evidence doesn’t rule it out, but the prosecution did not allege that Knox and Sollecito conspired with Guede but were not physically present, because they do not have the evidence of communication between them to support it.
You are consistently refusing to stick your neck out and engage with questions that will expose your analysis of the case to criticism. I have just answered your questions, yes I agree that if Knox and Sollecito did not physically engage with Kercher, we should not expect traces on her, but we still have an expectation of evidence around the room, and yes they could have been involved through conspiracy without being present and leaving physical evidence, but we don’t have the evidence to support that either.
Now I am going to ask you to provide a rundown of your numbers, as I did myself, and I asked you to earlier, but you avoided with further questions. If you ask further questions without providing one, I am going to have to write you off as too logically rude to engage with.
That’s a different issue. If you want to abandon your original claim, fine but you should acknowledge it. The question I answered was how my estimation of the chances of Knox’s (and Sollecito’s) guilt would change if I were very confident that physical participation in the murder would have left DNA traces of Knox (or Sollecito) on or about Kercher’s person.
That’s the question I answered. If you want to ask me a different question, fine, but please stop pretending that we are talking about a different question.
I have my own rules of debate, and I don’t engage with people who strawman me, i.e. people who insist on attributing to me positions which I did not take. So please knock it off.
Anyway, Moore’s assumption is that Knox and/or Sollecito were physically involved in the murder. Re-read his piece.
Exactly what? If Knox and Sollecito were standing just inside the doorway, encouraging Guede to attack Kercher, and then turned and ran out when he started stabbing her, what evidence would we expect to see? Certainly not bloody footprints of Knox or Sollecito.
I already provided my calculation, changing your 99% to 95% to reflect the possibility that Knox and/or Sollecito were guilty but did not do anything like hold grab or stab Kercher. I think that there is a 5 to 10% chance that Knox and Sollecito were guilty but did not physically participate.
And again my question:
The probability that Knox (and Sollecito) were involved in the murder is necessarily greater that the probability that they were physically involved.
Please quote 3 questions which you think I have avoided and I will try to answer them.
And please do not ignore or evade my question above:
If Knox and Sollecito were standing just inside the doorway, encouraging Guede to attack Kercher, and then turned and ran out when he started stabbing her, what evidence would we expect to see?
ETA: By the way, I totally concede that if Moore’s assumptions are correct then the probability that the prosecution’s scenario is correct is close to zero. I also concede that if your assertion about biological traces is correct, then the probability that the prosecution’s scenario is correct is close to zero.
What I asked for is a run down of your numbers leading to your estimation of your probability of guilt, such as I provided here.
I have asked for
A step by step analysis of your probability estimate, as I linked above.
I did not ask for your calculation of the weight of the forensic evidence, simply substituting in another number for mine, I calculated for you what weight you would have to assign the evidence, and I asked you why you thought that was a reasonable probability, and you responded with more questions instead of attempting to provide an answer.
A source for your probability assignment of the strength of the forensic evidence.
An explanation for why you revise your probability estimate strongly on the arguments of the prosecution, who have a vested interest for arguing Knox and Sollecito’s guilt, when their evidence has been found in many points to be false, exaggerated or mishandled, but do not revise your probability estimate at all on the arguments of Moore, who in addition to being a qualified expert, does not have money or reputation riding on the outcome of the case, and stands to lose reputation by making assertions about it that would reflect poorly on his expertise. You have asserted that this is an accurate representation of your analysis of the evidence, but not, as I requested, given any explanation for it.
As I stated in my previous comment, I will answer further questions from you if you provide a rundown of your numbers that form the basis for your probability estimate. This would be a simple request for you to fulfill, and would save time and energy over the course of this conversation. Take as long as you want, it’s not going to take as long as this conversation will if you hedge around.
It didn’t seem to me that you were doing that. My impression was that you were asking how I got from 90% to 30%; not how I got to 90% in the first place. The topic under discussion was whether my probability estimate had changed in light of various information and it looks to me like you are trying to play a gotcha game.
Anyway, I will give you a kind of run down with the understanding that if you ignore or evade any of my questions from now on, we are done.
The numbers I am giving you represent a rough approximation of the intuition in my thinking. At this point, 90% is not the result of an explicit mathematical calculation. Anyway, in my opinion, the prior probability that Knox is guilty is pretty low (i.e before one looks at the various circumstantial and forensic evidence). On the one hand, it’s extremely rare for someone of her demographic to take part in a murder; on the other hand it’s pretty common that when somebody is murdered it was someone in their household. On balance, perhaps about 1 in 500 is my prior estimate of the probability of Knox’s guilt. At the same time, the chances that an innocent person would have this kind of circumstantial evidence against them (changing alibis, staged break-in, forgotten phone calls, and so forth) is much lower. Perhaps 1 in 5000. (This is including the exculpatory aspects of the evidence as well.) Of course a guilty person would be very likely to have this kind of evidence against them. So out of 100,000 hypothetical Knoxes, 200 are guilty and roughly 190 have this kind of circumstantial evidence against them. At the same time, 95,000 are innocent and about 20 of those innocent Knoxes have this kind of circumstantial evidence against them. So of 210 people who have this kind of circumstantial and forensic evidence against them, roughly 90% are guilty.
I’m not sure what you are asking for here, but the source of my estimate of the strength of the forensic evidence (taking Moore’s assertions as true) is the observation that Knox (and Sollecito) may have been involved in the murder but not in a way which would have put them close enough to Kercher to leave the sort of evidence Moore discusses. I have estimated this probability at about 5 to 10%. i.e. I think there is a 5 to 10% chance that IF Knox and Sollecito were involved in the murder, they were involved from at least a few feet away.
For example, it’s possible that Knox and Sollecito were encouraging Guede from just inside Kercher’s door, and turned and ran away shortly after he started stabbing her.
So here’s a couple questions for you. (And if you ignore or evade them, we are done.)
Do you agree that in the scenario I described where Knox and Sollecito encourage Guede and then run, it’s pretty likely that they would not have left any DNA traces on or about her [ETA: Kercher’s] person?
Do you agree that in this scenario, it’s pretty likely that Knox and Sollecito would not have left the kind of evidence described by Moore in his article?
That’s not what Moore says. His reasoning is based on the assumption that if they were involved, they were physically involved. Read his article again.
I am not sure what you mean by “arguments,” since my probability estimate is based mainly on the evidence. Evidence is not the same thing as arguments. And neither is the same thing as an opinion. I am not basing my probability estimate on the prosecution’s arguments.
Moore’s actual argument is very solid. The problem is his opinion. I am skeptical of his opinion since he has an agenda. I am also skeptical of the prosecution’s opinions.
And another two questions:
The probability that Knox (and Sollecito) were involved in the murder is necessarily greater that the probability that they were physically involved. Agreed?
And do you agree that if they were involved in the murder but did not hold, grab, or stab Kercher, then there is a good chance that they would not have left biological traces on or about Kercher’s person?
I wouldn’t call it a gotcha game, but I was of course expecting to find something in your probability analysis that I would object to. My primary dispute has all along been that I think your probability estimate of Knox and Sollecito’s guilt is much too high. If our posteriors disagree, then we have an implicit disagreement about what our priors should be and/or how much we should update on the available evidence. If we do a breakdown of our probability analyses, then we can determine the source of our disagreement.
The “staged break in” claims have been more or less discredited; all the evidence thus far has been found consistent with a real break-in. Unless you’re referring to some argument presented by the prosecution that I haven’t been able to find, the “forgotten phone calls” refers to a three second phone call from Knox to Kercher, which I do not think accounts for any weight as evidence at all, as I have argued here. The “changing alibis” was done,according to Knox, under extreme duress; she claims that she was treated abusively during interrogation, and encouraged to implicate Lumumba, as the police suspected him on the basis of her text message to him, until she began to imagine that perhaps she and Lumumba really had been involved after all. After she signed the statement implicating him, had a meal and some time to gather her thoughts, she went back to saying that she had no recollection of having been involved, and that the statement was collected under coercion.
A prior probability of .002 for guilt strikes me as a fairly credible starting point (I assigned a higher prior, but in retrospect I think I should have adjusted down considerably based on the fact that Knox and Sollecito meet none of the FBI dangerous individuals criteria,) but it seems that you are adjusting strongly upwards on evidence that had already been discredited, or is already very weak due to high likelihood of coercion.
In addition to the forensic evidence, Moore also lays out sixteen FBI criteria that serve as indicators of violent inclinations, out of which Knox meets zero. For a person to cheer on the attacker of a close acquaintance they had no motive to kill themselves, I would strongly expect them to show signs of a violent or sociopathic personality, which Knox and Sollecito do not.
Yes, I agree. However, I think that this scenario has a dramatically low prior probability. The calculations I’ve given there describe the upper limits of my confidence interval on the likelihood I would assign to such an event; I believe that even one person, who has no motive to kill the victim, is not a sociopath, and meets none of the FBI dangerous individuals criteria, witnessing a violent attack by a person they do not have strong affiliations with (if the person were, say, a fellow gang member or mobster it would be a different matter,) and cheering on or otherwise encouraging the attacker, rather than say, intervening to help the victim, calling the police, or panicking and running away, would be an event completely unprecedented in the history of criminology. For two people to do so together goes beyond incredible into downright ludicrous.
I agree that the forensic evidence does not, as far as I am aware, rule out such involvement, but the evidence of Knox and Sollecito’s failure to meet any of the FBI dangerous individuals criteria does weigh heavily against it, and in my prior probability distribution for Knox and Sollecito’s involvement in the murder, I would only assign such a scenario a fraction of the likelihood that I would assign to the already unlikely “Bizarre sex game goes out of control” scenario, which the forensic evidence does rule out.
Agreed and agreed, but i would assign most of the prior probability mass for their being involved in the murder, but not physically, to the possibility that they were involved through conspiracy, and the lack of evidence of communication between Knox and Sollecito and Guede weighs heavily against that. The prior probability of their being involved via happening to witness the attack, and encouraging the attacker, represents such a miniscule part of my probability distribution that I would need a great deal more evidence to raise it to the point of being worthy of attention. It took me a while to realize that you were positing such a scenario because its initial plausibility was so low that it didn’t occur to me that you would suggest such a thing.
Given that many of the prosecution’s claims have been discredited, or simply do not constitute evidence in favor of Knox and Sollecito’s involvement, I think that a 5000:1 odds ratio is much too much weight to assign to their case. However, even if we do assign that much weight, I don’t think that it comes anywhere near to overcoming the prior improbability of the sort of involvement you’re suggesting.
I’m not sure what you mean by this. The fact that a piece of evidence is “consistent with” a real break-in does not mean that such a piece of evidence is not indicative of a staged break-in. For example, when Sollecito called the police, he allegedly advised them that there had been a break-in but nothing had been taken even though there was no way he could have known this if it had been a real break-in. This evidence is “consistent with” a real break-in in the sense that it’s possible to come up with an innocent explanation for Sollecito’s statements. But it doesn’t change the fact that it’s a piece of evidence which supports the staged break-in hypothesis.
No, I was mainly referring to Knox’s trial testimony where she claimed to have forgotten about a call she made to her mother on the day the body was discovered. The call was made extremely early in the morning, Seattle time.
So what? Look, I agree that one can look at each piece of circumstantial evidence against Knox and come up with an innocent explanation for each piece. It doesn’t change the fact that when you look at all of it together, it’s extremely unlikely that an innocent person would have all of that evidence against him.
I’m not sure what your point is here. I agree that Knox is not a likely murderer. That’s why my estimate of the probability of her guilt is only 90%. If she had been Kercher’s ex-lover; or if she came from a more violent demographic, my estimate would be higher.
Well let me ask you this: Assuming that Knox was involved in the murder, what do you think the chances are that she was not physically involved?
Knox had already returned to the residence she shared with Kercher, and witnessed the signs of the break-in, but not, according to her own testimony, entered Kercher’s room and seen the murder scene (if she had entered the room, she would probably have left footprints and/or gotten blood on her shoes, which the police would have been able to discover, so we can assume that she probably did not.) She returned to Sollecito’s house after this, had breakfast, and made her calls. If their house had been robbed, it is probable that they would have taken things that she would not have to enter Kercher’s room to find missing, so “nothing was taken” would be a reasonable inference based on the information she had available to her. I agree that the claim that nothing was taken shifts the probability in favor of a staged break-in, but I don’t think that it shifts it very much.
The prosecution’s case for a staged break-in also rested on arguments such as the pattern of broken glass being inconsistent with a break from outside, and it being impossible to get in through the fire escape, both of which were simply false and represent no weight as evidence at all.
Well, this is not what I thought you had in mind, but I do not think that this is something Knox is much more likely to testify that she forgot if she was involved in the murder than if she was not. Whether she was innocent or guilty, we can assume that she was under a fair amount of stress, and the contents of the phone call weren’t incriminating anyway.
I strongly disagree. I would bet in favor of police finding a body of evidence of at least equal strength against me, were I to be investigated for a murder I was uninvolved in. I forget phone calls frequently, make statements based on inference which I do not have highly certain knowledge of on a regular basis, would probably change my alibi if I were abused and exposed to strong suggestion by the police, and have unusual mannerisms. This is why I originally brought up the point about the evidence being filtered; I am unconvinced that the evidence brought forward by the prosecution is more than I would expect them to find for the average person to whom I assigned the same prior probability of guilt.
I’d call 99.99999% a fair estimate. But I’d estimate about 0.00000000000001% likelihood of her being physically involved, based on my previous likelihood calculation, plus the evidence that she doesn’t meet any of the FBI dangerous individuals criteria, and the lack of evidence of conspiracy with Guede, both of which I didn’t account for before, and the evidence of her failure to leave forensic evidence. So if you multiply the chance that she was involved physically by the number of times more likely I think it was that she was involved non-physically, you get about 0.0000001% likelihood. Most of my probability mass for that would be that she somehow managed to conspire with Guede without producing evidence that the police would discover, rather than the even more unlikely hypothesis that she happened upon the murder and encouraged it.
My probability estimate that she and Sollecito were both involved is, of course, lower than my probability estimate that Knox alone was.
At the time Sollecito reported that there was a break-in and nothing was taken, he was aware of the broken window in Romanelli’s room but Romanelli had not returned and inspected her room. Agreed?
I don’t recall the prosecution making this claim. In fact, I don’t recall hearing anything about a fire escape. Can you give me a link?
I disagree. The contents of the call were very inculpatory and it was extremely convenient for her to forget about it. Anyway, even if she was under stress, it seems pretty unlikely to me that she would forget the call and continue to forget the call even after she was reminded of both the call and its contents. Especially given that it was an unusual call; made at a time when she was wide awake but knew her mother would be sleeping.
I doubt you would forget an unusual phone call made to your mother at a time when you knew you would be waking her up. Especially if you were reminded about the time and contents of the call.
Well in your view, what are the strongest pieces of circumstantial evidence against Knox?
Ok, then you should not object if my probability of Knox’s guilt doesn’t drop much if I add the assumption that Moore’s assertions are substantiated.
By the way, I am STILL waiting for cites backing up your claims about (1) trace DNA ending up on or about Kercher’s person if Knox (or Sollecito) grabbed, held, or stabbed her; and (2) further evidence of Guede DNA traces being discarded and not making it into the record.
Agreed.
She was a college student without a lot of experience living on her own. Her home is broken into, she’s in a panic, she calls mom and asks “My home has been broken into, what do I do?”
Forgetting such a call, even repeatedly, would be entirely characteristic of me.
I had remembered komponisto, who translated the Massei-Cristiani report, making such a statement, but applying a search function over his posts, I can’t find any such claim. I suspect that someone else made a statement along those lines and I misremembered, but I don’t know that they would be a credible information source, so I’m prepared to drop that claim.
Unusual mannerisms, and the dryer.
That’s my estimate for the relative probability after the forensic evidence is accounted for. My likelihood estimate of Knox and Sollecito’s guilt after accounting for the forensic evidence is approximately equal to my probability estimate prior to it minus all the probability mass I assigned to their physical involvement. You already acknowledged that if you accept Moore’s arguments, it reduces the likelihood of Knox and Sollecito’s physical involvement to near zero. If your prior estimate assigns 90% probability to Knox and Sollecito being guilty, and 85% probability to any of the prosecution’s scenarios being correct, then that leaves at most 5% of your probability mass assigned to the possibility that Knox and Sollecito were involved non-physically, since the prosecution did not put forward any scenarios in which Knox and Sollecito were involved non-physically. So if you subtract nearly all of the probability mass that you assigned to Knox and Sollecito being involved physically, that should reduce your probability estimate of Knox and Sollecito’s guilt to just over 5% at most, less by any overlap between “Knox and Sollecito physically involved in the murder” and “Prosecution’s scenarios wrong.”
For reasons I have already explained, this would be difficult for me to produce. I would need to contact an expert. I will agree to put in that effort if you also attempt to find a cite for your implicit premise, that 1:2 relative odds of non-physical to physical involvement is a credible distribution of probability mass in a murder with no evidence of conspiracy. You may agree to do this, and I will attempt to get a citation from an expert, you may offer some other assertion of yours that you are willing to go to the effort to acquire a citation for, and if I agree that it’s a reasonable substitution, I will make the effort to get the citation. If you do neither, I will not make the effort to get the citation, and I will have to take issue with your making further requests.
Ok, then I still don’t understand what you mean by your claim that the evidence is “consistent with” a real break-in. This is one piece of evidence which you seem to concede supports the staged break-in hypothesis.
Assuming you are accurately summarizing the contents of the call, it’s suggestive of guilt. The fact that you (seem to) have no idea why is telling.
You should also be prepared to confront the reality that you have very poor understanding of the case against Knox and Sollecito. This is the second time you have confidently mischaracterized the prosecution’s case. You don’t understand the significance of the forgotten phone call. And you seriously believe that Knox’s mannerisms are among the strongest pieces of circumstantial evidence against her.
It’s obvious that you are suffering from confirmation bias, i.e. you aren’t paying serious attention to evidence (or argument) which contradicts your views about the case.
Ok, then tell me your probability estimate ignoring the “forensic evidence.”
Actually the odds I have been using are roughly 1:20. You seriously think that’s way off-base? (Please don’t evade this question.)
Anyway, your (apparent) claim about Italian police and legal procedure is one that even you need to concede you do not have the expertise to make.
I too have no idea. Yes, it’s something a person who has something to hide and doesn’t want to involve the police might do. It’s also something a confused college kid in a foreign place might do on encountering any hiccup. Completely consistent with either guilt or innocence means it sways me very little.
Then I suggest you educate yourself about the case.
That was a remarkably unhelpful comment. I have read quite a bit about the case and still don’t have any idea why that would be suggestive of guilt.
Do you? Any hints, or are these trade secrets?
You asserted before that you yourself do not have a great deal of familiarity with the arguments made by the prosecution, and yet you find it adequate to support 90% confidence of guilt.
I agree that this statement was probably a mischaracterization of the prosecution’s argument, but can you name any other piece of evidence regarding the break in which is more probable in light of a staged break-in than a real one?
I think you are dramatically underestimating the frequency with which college students unused to living alone and exposed to frightening and unfamiliar situations will default to the action of “panic and call mom.”
I am probably subject to some degree of confirmation bias in this case, but I would need to weight the evidence in favor of Knox and Sollecito’s involvement many, many times more heavily for it to become a credible assertion. I think that you are yourself subject to an extremely strong confirmation bias here; weighting Knox and Sollecito’s changing of their alibis under probable abuse more heavily than Knox, Sollecito and Guede’s mutual failure to implicate each other, for instance, seems purely nonsensical.
Some quick googling doesn’t give me any results relevant to the question of what proportion of murders involve conspiracy, but I would suggest a prior of maybe 1:100. However, the lack of evidence to support a conspiracy would weigh heavily against that, so I’d shift down to maybe 1:10,000.
“Suspects witness victim, who they have no motive to kill, being attacked, and cheer on the attacker,” would approximate to zero percent of my probability mass.
If you can point to any evidence that such an event (non sociopathic, FBI zero threat rating individuals witnessing an attack on someone they have no motive to kill, by someone they do not have strong affiliation with, and cheering on the attacker,) has ever happened before, I would agree that I am underestimating the probability. Can you point to any such evidence?
I think it’s somewhat off base as a prior, and very very off base in light of the lack of evidence of conspiracy, and if you assign a significant fraction of that to “witnessed attack and cheered it on,” I think it’s positively ludicrous.
Can you explain what this apparent claim is?
Nonsense, I clearly stated that my probability assessment was based mainly on the actual facts as opposed to arguments.
Your second mischaracterization. Not only that, you seriously believe that one of the best pieces of circumstantial evidence against Knox is her “mannerisms.” And not only that, you do not understand the significance of the forgotten phone call. Just admit that you don’t really understand the case against Knox (or Sollecito). I promise you the sky won’t fall.
Given that you clearly don’t know the strength of that evidence, you shouldn’t have an opinion one way or another.
Please show me where I made such a weighing. Please QUOTE me. Failing that, please admit that I made no such weighing and apologize.
What exactly do you mean by “conspiracy”? It seems pretty clear that if Knox and Sollecito were involved in the crime, they acted in concert with Guede.
In the scenario I describe where Knox and Sollecito encourage Guede from a few feet away and then run, would you count that as a “conspiracy”?
I think I probably could. But let’s make sure I understand what you are saying.
You are claiming that Scenario A (non sociopathic, FBI zero threat rating individuals witnessing an attack on someone they have no motive to kill, by someone they do not have strong affiliation with, and cheering on the attacker) is at least a thousand times less likely than scenario B ((non sociopathic, FBI zero threat rating individuals witnessing an attack on someone they have no motive to kill, by someone they do not have strong affiliation with, and physically participating in the attack).
You are also claiming that Knox and Sollecito are known to be non-sociopathic and known not to have had a motive to harm Kercher.
Do I understand you correctly?
That if additional samples taken from on or about Kercher’s person matched Guede, it would not have become part of the record in the case.
ETA:
A single piece of evidence? Sollecito’s phone call comes close, but probably not.
No, I don’t concede that. If you can explain to me why it’s so much more likely that Knox would make that phone call if she were guilty than if she were innocent, perhaps it will be something I failed to consider, and I will change my mind. But as-is, it seems to me like such a weak enough piece of evidence that had the police not already promoted Knox and Sollecito to attention, they would probably not have treated it as worthy of notice.
I inferred that you made such a weighting from this statement
If I was incorrect, then I apologize for that. But I don’t see how you’re getting an odds ratio of 1:5000 in favor of guilt for changed alibis (very high likelihood of coercion,) a forgotten phone call, a phone call stating that nothing had been taken in the break-in (which they could reasonably infer,) versus mutual failure to implicate, absence of evidence of motive, absence of evidence of dangerous personalities or inclinations, absence of evidence of conspiracy, and absence of evidence of physical presence at the scene.
By conspiracy, I mean that they intended for Kercher to be murdered, and made plans to that effect in advance with her murderer.
No. I am saying that Scenario A is much less likely than the possibilities that dominate the probability mass for “Knox and Sollecito physically involved in Kercher’s murder.” This would include possibilities such as “violent argument over cleaning roster gone downhill,” “sex game gone wrong,” etc. The prior probability for “Knox and Sollecito witness Guede attacking Kercher, and proceed to assist Guede in the attack,” as with the probability that they would cheer him on, is infinitesimal.
In retrospect, I think that the relative prior probability that I assigned to their involvement through conspiracy (1:100) was too low, but it’s still very low in light of the absence of evidence that Knox and Sollecito conspired with Guede, and the likelihood that they would observe an attack in progress and support the attacker is infinitesimal.
Also note that if Knox and Sollecito came to the residence Knox shared with Kercher, and witnessed an attack in progress, they cannot have let Guede in. Any evidence in favor of a staged break-in is evidence against this hypothesis.
The prosecution tried quite hard to find a motive for the killing, and was unable to establish one, unless you account “crazy sex game gone wrong” or “satanic ritual” as motives, neither of which they were able to present evidence for. At this point, the notion that Knox and Sollecito had a motive to murder Kercher deserves a very low probability.
Amanda Knox was known by her acquaintances for having an low tolerance for conflict, and being uncomfortable if she and people around her were in disagreement, behavior which is highly inconsistent with Antisocial Personality Disorder. Further, if the prosecution was able to find evidence that she or Sollecito had APD, they had strong motive to make record of it, but the Massei-Cristiani report contains no such evidence. So the probability that Knox and Sollecito would witness an attack on Kercher and support the attacker may be dominated by the possibility that they were secretly sociopaths and had reason to want her dead, but we have more evidence than for the average person to suspect that neither was the case.
I never said that I expected that if they found additional samples implicating Guede, it would not have become part of the case, I said that I did not have strong expectation that if further evidence been found against Guede, it would have been raised in the case (or to be more precise, I do not have a strong expectation that they would have brought the evidence to the case and I would have found out about it; Guede’s case was not controversial and so was not heavily reported on, and he opted for fast track trial in which he waived the right to contest any evidence against him, and all the primary documents are in Italian.)
Because I do not have a great deal of expertise regarding the Italian legal system, my model of reality would not have taken much of a hit either way.
It’s not really my job to explain it to you. It’s your job to do your homework to find out what the best and most important evidence is on both sides before you make your probability assessment. I have done that with both sides. On the other hand, you obviously have a very poor understanding of what the case against Knox is. As far as I can tell, you haven’t even tried to defend your claim that Knox’s mannerisms was among the best circumstantial evidence against her.
Yes, you are incorrect. My statement was clearly looking at the body of evidence as a whole.
Ok, then I don’t see how the concept of a “conspiracy” adds much to the discussion. If there was a conspiracy, Knox and Sollecito plausibly may still have been physically involved. If there was not a conspiracy, Knox and Sollecito may plausibly not have been physically involved.
In that case you are comparing apples and oranges, since you are comparing one narrow scenario I proposed with much bigger set.
You are ignoring other possibilities here. A more likely scenario is that Knox and Sollecito conspired with Guede to harm Kercher in some way but did not intend for her to die.
For example, consider the murder of Janet Chandler which took place back in 1979. Her roommate and co-worker—a girl named Laurie Swank—apparently helped to organize a sexual attack on her. Swank admitted to cheering on the attackers who ultimately killed Chandler. There was no DNA testing back then, but if there had been, there would not have been any of Swank’s DNA on or about Chandler’s person.
I agree but so what? I’m not proposing that Knox and Sollecito walked in on Guede assaulting Kercher and spontaneously decided to help him. I’m proposing that they acted in concert for some unlawful purpose (which by hypothesis pretty clearly happened) and that for whatever reason, Knox and Sollecito were not close to Kercher when Guede stabbed her to death.
So basically there are two possibilities, assuming that Knox and Sollecito were involved in the murder: (1) They acted in concert and were all physically involved; and (2) They acted in concert but only Guede was physically involved. You seem to be saying that (1) is far more likely than (2). I don’t see why, since both scenarios are (a priori) pretty remote. I don’t see any (a priori) reason to think that one is 100 or 1000 times more likely than the other.
But let’s be specific: What do you think is the probability ratio of (1) and (2)?
I don’t think that’s what you said, but anyway it seems to me that Knox and Sollecito (or their supporters) would have had incentive to raise this issue. If there were 100 traces of Guede on Kercher’s person and none of Knox, there’s a decent—perhaps compelling—argument that the prosecution was wrong. Certainly Moore did not make this argument even though it would have fit in nicely with the rest of his essay.
Downvoted for this remark and most of the next paragraph. In this context, Derstopa is clearly aware of the phonecall and doesn’t understand why one would think it is evidence of guilt. Explaining why one would think that it is non-trivial evidence of guilt (if one thinks it is) is what one should do if one trying to have an actual discussion. Refusing to explain that and telling someone to just do their homework is rude, logically rude, and generally not productive for people trying to understand your viewpoint.
I would say it depends on what is being discussed. My point in that post was not so much that Knox is probably guilty as that Derstopa is ignorant of the case against Knox. Note also that Derstopa is aware of the phone call only because I mentioned it to him.
I’m not Desrtopa, but I had heard about the call way back in 2009. I dismissed it as essentially irrelevant, because no one who has mentioned it has ever said anything convincing about how it was inculpatory. Apparently it was supposed to be obvious? But it wasn’t, not to me, nor apparently to Desrtopa. At best I can make strawman arguments which I find easy to knock down. Having you respond to those would be a waste of everyone’s time. So please, just tell me why you found this call suspicious.
I don’t speak Italian. I have been limited to secondary sources, rather than the prosecutor’s own arguments directly, but I really have looked closely at arguments by those who think Knox and Sollecito are guilty. At the time, they did not articulate any reason for me to find the phone call suspicious. I still haven’t found anything after looking again.
I don’t think it’s obvious at all.
Which sites did you look at?
The first sentence is irrelevant to my observation. The second claim (about only knowing about the phonecall from your remark) seems inaccurate to my reading of the conversation. From my skimming the thread again the only bit I see that could give you the impression that Derstopa wasn’t aware of the phonecall is this exchange where Derstopa doesn’t say that. In that context, Derstopa thinks you are talking about a different phonecall. Nothing in that is a useful indication that Derstopa didn’t know about this phone call. Moreover, if you thought that that was the problem then telling Derstopa that he should read up on the prosecutions case would have made minimal sense as a reply at that point, not a reply three days later to where Derstopa asks about how one thinks the phonecall is relevant, especially when they’ve already explained why it seems only marginally relevant.
Here’s what he said:
It seems pretty clear to me that he was not aware of the phone call I was referring to.
As I recall, at that point, I was only just beginning to realize just how profoundly ignorant Derstopa was about the case against Knox.
JoshuaZ’s comment reflects everything I have to say about this.
Then why is there no evidence of prior communications between Knox, Sollecito and Guede to support this? Knox had only met Sollecito a week before Kercher’s murder, and had just started dating him, and there’s no evidence that Sollecito and Guede had ever even met as of the time of the murder. The prosecution has no record of phone calls between Knox and Guede, and has no evidence or testimony of any sort to suggest that the two were more than slightly acquainted.
I agree that this scenario had greater prior likelihood, but an absence of evidence that Sollecito and Guede had ever even met is a heavy weight of evidence against it.
You think that there’s enough evidence to think that Knox and Sollecito, for whom there is an absence of evidence of any motive, and who had just recently met, deliberately collaborated with Guede, who there is no evidence that Sollecito had ever met, or that Knox had ever called or met with privately, conspired with Guede to harm Kercher? When did they do the conspiring, after 9:10 PM on the night that Kercher was killed?
The prosecution never alleged conspiracy between Knox, Sollecito and Kercher; the information they had available to them didn’t render it tenable.
If Knox and Sollecito were complicit in the crime, but not participants in the crime scene, they would most likely have to have premeditated some sort of harm against Kercher, and the prosecution found no evidence at all to suggest they would have any motive for such. If they were physically involved, they could have premeditated harm against Kercher, but they could also have become involved in the heat of the moment. Most murders are not premeditated, particularly in cases with no apparent motive prior to the crime.
I would say that my previous assignment of 1:100 may have been excessive; a prior ratio of 1:20, as you assigned, might be more reasonable. But the evidence that renders any before-the-event conspiracy between the three highly unlikely weighs against it.
Just how little evidence do you expect prosecutors with a heavy dose of motivated cognition to find for an innocent person, that you think what the prosecution found weighs so heavily against Knox and Sollecito? The data from the Innocence Project suggests an error rate higher than this for average cases prior to the use of DNA evidence, and the evidence against Knox and Sollecito is far weaker than for the average suspect, let alone convict, and the evidence pointing to their innocence far stronger.
As I said to JoshuaZ, the main point of that passage was that you are ignorant of the case against Knox. Which you do not seem to deny anymore.
Not all communications result in a phone or other record. Particularly if some of those communications are between drug users and elements of the criminal underworld.
Ok, then it was reasonable for me to go from 90% to 30%. Unless you think there is an error in my math?
I would say a lot less than here. Maybe one or two pieces of inculpatory evidence at most.
By the way, can I take it you are conceding that Knox (and/or her supporters) would have had an incentive to introduce and publicize evidence that there were more traces of Guede on or about Kercher than the two I mentioned earlier?
I do deny this. You accuse me of not defending my claim that Knox’s mannerisms are among the strongest evidence against her. I have defended at several points my reasons for assigning very low weight to pieces of evidence put forward by the prosecution. You, on the other hand, while asserting that the prosecution’s arguments collectively deserve high weight, have not defended your reasons for assigning high weight to any of them. In response to requests for a defense of such, you’ve told me it’s not your job to do my homework.
Only if you assign zero weight to the evidence rendering conspiracy between them unlikely.
Not all communications require physical meeting or result in a phone record. But when you have no evidence of private meeting, no phone records, no email records, and a very short time frame in which any sort of communication could have taken place, and you don’t adjust your likelihood of communication downwards at all, then it’s clear that you’re desperate to come to a particular conclusion and don’t want to perform any sort of update which would render it more unlikely.
Then why are so many primary suspects and even convicts later proven to be innocent? Your estimation of the amount of evidence that it is reasonable to expect against an innocent person seems strongly contradicted given our information on false conviction rates.
I have stopped trying to argue that point because, in the face of a qualified expert without any professional stake in the case claiming that it is enormously unlikely for Knox and Sollecito to have left an absence of evidence if they were physically involved in the case, you insist on demanding a particular piece of evidence which I have already provided repeated and abundant explanations as to why it would be difficult for me to produce.
In fact, when I put it in words, I have to wonder I’m still arguing about any of this. Thus far you have shown a tremendous unwillingness to update in favor of any exculpatory evidence (when I think evidence from the prosecution is weak, I explain why I do not think it is much more likely if Knox and Sollecito are guilty than if they are innocent. You refuse to update on evidence entirely with an explanation of how it does not disprove all possible guilt scenarios.) You have been rude, logically rude, and I suspect you could go on arguing for Knox and Sollecito’s guilt indefinitely, increasing your confidence in your probability assignment no matter how much evidence mounts against you.
You have given me every reason to believe that if exposed to further evidence against your position, you will refuse to change your assessment at all. This simply isn’t worth my time.
Indeed, for some time I’ve been tempted to caution you against feeding the trolls.
If brazil84 weren’t a lawyer, I would have written this debate off as pointless a long time ago. It’s raising a negative affect around all the time I spend on this site, but simply walking away knowing that he’s going to continue practicing law without realizing that he’s been making some big mistakes of judgment that demand commensurately big corrections has up till this point been even more frustrating.
Then I commend you for your efforts.
brazil84 being a lawyer, though, reduces my confidence that this will pay off. That is a person optimized for debating rather than truth-seeking.
As a lawyer, I feel obligated to defend the profession a little. Lawyers exist to maximize the utility of the client. But a lawyer who has no grip on what is true cannot tell the client when it makes sense to throw in the towel. And that lack of judgment is not rewarded.
That said, lawyers are relatively well trained in behavior that is orthogonal to truth but correlated with winning. But for the most part, the legal system (i.e. the judge) is trying to determine truth within the limits of its empirical system.
Well said. But note that the behavior is not entirely orthogonal to truth-seeking. Being trained to discuss a dispute through the techniques of debating will surely decrease one’s ability to say “oops”. (not in the sense of it being a necessary decrease, but in the sense of it being a tendency to decrease such that my prior reduced confidence is justified)
This is actually something I struggle with, being trained as a polemic philosopher.
You are talking about the evidence, right?
In response to your request regarding the phone call, yes. The issue there is not that you have considered the argument on this point and rejected it; the issue is that you are completely unaware of the significance of the forgotten phone call. I identified the phone call merely as an example of the body of evidence against Knox.
Well, assuming for the sake of argument that (1) Knox and Sollecito were involved in the murder; and (2) there was absolutely no conspiracy as you have defined that term, are you saying that your a priori probability estimate of physical involvement by Knox and Sollecito is 1000 times higher than no physical involvement?
If not 1000, then what?
Are you saying that I have done so?
I dispute that “so many” primary suspects and even convicts are later proven to be innocent. How often do you think this happens? What is the percentage of convicts who are later completely exonerated? Surely it is far less than 10% and I have allowed a 10% chance that Knox and Sollecito are innocent.
The only exculpatory evidence you have offered (which I have not already considered) is based on your non-expert opinion. I have patiently explained to you the effect it would have on my probability estimate if you were able to document your opinion.
Or is there some other piece of exculpatory evidence you think I am not considering in my probability estimate?
Look, you are the one who is proposing that additional traces of Guede on Kercher would not have made it into the public record. I am asking for you to back up that claim.
Seriously, if you want to continue this debate, show me some evidence that you will actually update in the face of more evidence against your position. In the beginning of this debate, you said that you were aware that the DNA evidence against Knox and Sollecito had been challenged. In the face of the independent experts’ report indicating that there is no valid DNA evidence against them, and Moore’s expert claims for all the evidence we should expect to see in the case of the sort of involvement alleged by the prosecution that we do not see, you update your probability by… nothing. You say you would update your probability were I to provide the right references, but when I offer to make the effort to obtain those references in exchange for a comparable effort on your behalf to produce evidence for your position (I offered you to substitute evidence for any of your assertions if you suggested the substitute first and I agreed that it was appropriate,) you evade. Why should I think this conversation is worth my time? You’ve given me every reason think that if I go to further efforts, you’ll just find more excuses not to update.
You two agree on something!
I already conceded that my probability estimate would drop from roughly 90% to roughly 30% if you could substantiate your claim about transfer of DNA traces.
But anyway, I am not particularly interested in debating Knox’s actual innocence with you given your ignorance of the case against Knox.
That’s not what the independent expert report said, as far as I know.
Nonsense, you are the one who is evading; since it’s not even clear that you dispute my “position” on the point and you refuse to clarify your position. You demanded evidence for a strawman position you set up.
I will ask you one last time, if you evade or ignore we are done:
Assuming for the sake of argument that (1) Knox and Sollecito were involved in the murder; and (2) there was absolutely no conspiracy as you have defined that term, are you saying that your a priori probability estimate of physical involvement by Knox and Sollecito is 1000 times higher than no physical involvement?
If not 1000, then what?
ETA:
I don’t know, what exactly are you trying to accomplish?
We are done. You’ve given me no more reason to think there’s any point in having a conversation with you.
I don’t want to get sucked in by replying to more of brazil84′s comments, but in the spirit of this quote (edit, link fixed,) in case it will help me to do better in future...
Did it look to anyone else like I was bobbing and weaving with my responses to these questions? My perception was that I answered these questions in a straightforward way, but I may have been suffering from illusions of transparency.
You definitely got somewhat sucked into “debate mode”, and you often had far too many points lobbed at you to have caught all of them. It would seem to be possible for brazil84 to honestly have thought you were evading something, but I don’t think an outside observer should have thought so.
That you seem to have kept your cool throughout this ridiculous exercise is a credit.
You might want to check the link. EDIT: works now.
On the substance, all I can say is dayenu.
No. You’re fine. That’s just how brazil argues.
Ok bye.
By the way, if any lurkers are reading this, I will point out the fatal flaw in Desrtopa’s reasoning on the 90% to 30% issue. See, the prior probability of Knox and Sollecito being involved in the murder but not physically is quite low. But the prior probability of Knox and Sollecito being involved in the murder physically is also low. What matters for my calculation is the ratio of the two probabilities.
I suspect that’s why Desrtopa had to bob and weave with my questions on this issue. Because he knew deep down that there is no basis to think that this ratio is wildly lopsided. And if the ratio is not lopsided, then it’s perfectly reasonable to go from 90% to 30%.
What exactly are you trying to accomplish?
I enjoy debating and I am fascinated by situations where the popular view is wrong. Also, I like debating where there isn’t anything important at stake.
Once in a while I learn something from debating with someone who disagrees with me.
What about you?
I strongly suspect that you are incorrect about which side of this debate has more adherents. Neither LW nor Wikipedia are likely to be representative.
I hope to learn something in every discussion I have. Debate is for changing people’s minds. Debate for its own sake is pointless (aka—I don’t debate Flat-Earthers).
You don’t seem interested in saying something that I can glean insight from, and you don’t seem open to changing your mind.
Actually, I’ve done that before. It helps a surprisingly large amount in getting one to realize how astronomy can easily pay rent. And it really tests how well one understands how much historical scientists did really careful, clever observations. That said, I can’t plausibly argue that this is a good use of one’s time, merely that is probably marginally better use than say playing Farmville.
On this discussion board, the popular view seems to be that Knox is most likely innocent. I make no claims about what is believed in the world in general.
The second part is incorrect—I am open to changing my mind. As far as the first goes, I don’t know where you are at in terms of insight so I couldn’t say.
ETA:
Agree denotatively but disagree connotatively.
On this he is clearly correct.
In any case I don’t think this particular comment deserved to be as down voted as it was.
First of all, just because they aren’t completely exonerated does not mean they are not innocent. A better question is what the percentage chance of innocence is.
Second of all, this assumes that letting a guilty person go free and sending an innocent person to get punished is of equal weight, which it is not.
The discussion is about the probability of actual guilt, not the expected utility of punishment, so that’s not relevant.
ETA: Relatedly, I’ve commented elsewhere that a 90% probability of guilt for murder should be well within the “not guilty” verdict range.
Oh right, good point.
Err that should be “what do you think is the A PRIOR ratio.”
Pardon for the digression, but isn’t Knox’s physical involvement in the murder the prosecutor’s theory of what happened? Otherwise, why should we be interested in Knox DNA on the victim’s bra?
I admit that it is conceptually possible that P(Knox is guilty) != P(prosecutor’s theory is correct), but in practice they tend to converge.
ETA: I’ll even agree that P(Knox is guilty) >= P(prosecutor’s theory is correct). Nonetheless, P(Knox is guilty) less P(prosecutor’s theory is correct) is a small number.
Yes, as far as I know.
Actually I think the allegation here is that Sollecito got his DNA on the bra while cleaning up the crime scene.
I agree, but in this case there is less of a convergence than usual since the prosecution had no choice but to speculate. In any event, I’m allowing a pretty modest probability—like 5 or 10% -- that Knox and Sollecito were guilty despite a lack of physical involvement as alleged by the prosecution.
ETA: For example, the scenario I described where Knox and Sollecito encourage Guede from just inside the doorway and then turn and run when the blood starts to spill.
This scenario you describe pushes very hard on the distinction between moral guilt and legal guilt, at least in American law. Are we debating Knox’s legal culpability or moral culpability?
So, from your point of view:
P(Knox guilty) = .9
P(prosecutor correct) = .85
And to clarify: “Knox guilty” = knowing Knox’s relevant behavior and thoughts from that time, every rational person would agree that she had committed a violation of the law that proximately caused the victim’s death.
Also, I think that if P(defendant is guilty) less P(prosecutor’s theory is correct) is in the range of .05, that is strong evidence that the prosecutor is utterly incompetent. After all, .05 probability is one in twenty. I confess unwillingness to believe that the prosecutor in Knox’s case is that incompetent. (More rigorously, my belief in that level of incompetence approaches .001, and even that seems high).
Well, it would still be illegal, although I admit that I interpreted brazil84′s meaning of “involved in the murder” more narrowly. I would have to make a different probability calculation for the the odds that they did not plan or participate in, but discovered and encouraged, the attack on Meredith Kercher. It would still contain much of the same evidence that I discussed before, and given that I have never even heard of such a thing happening, I would assign a much lower prior, particularly for someone who, as Moore points out, meets none of the FBI criteria for a dangerous personality.
It’s even more unlikely when you consider that we’re dealing with two people rather than one. If Knox and Sollecito did not plan to kill Kercher, and then found her being attacked in her home, the odds that one of them would be crazy enough to hang back and encourage the attacker, incredibly low as they are, are much higher than the odds that both would.
That’s why I specifically reference American law, which contains no general duty to rescue. That is one of the differences between English-descended Common Law, and French-descended Civil Code (which the common model in most of continental Europe).
I’m going to do a probability estimate for Knox and Sollecito having discovered and encouraged the attack. I’d do it in response to one of brazil84′s comments, but I’m already waiting on him to respond to some other questions and don’t want to sidetrack him.
Let’s start with the prior probability that Knox and Sollecito would witness the murder. Knox and Kercher lived in the same apartment, and Knox and Sollecito spent a lot of time together. Let’s call it .2
Neither Knox nor Sollecito meet the criteria for Antisocial Personality Disorder or any of the FBI dangerous individuals criteria. What should we call the probability that such a person, upon witnessing a violent attack on an acquaintance they have no motive to harm (the prosecution having established no motive) would stand back and encourage the attacker? I’ll call it >.0001. So if you have two individuals, the chance that they’d both be the sort of person who would hang back and encourage the attacker is >10^-8. But both of them would probably expect the other person to consider this behavior crazy and horrible, much like a person is less likely to commit actual murder in front of another person who they don’t know to be sympathetic, so I’ll assign a probability of .1 that if both were the sort of person to encourage the attacker, either would make the first move to start doing so. So at this point we have a probability of about 2x10^-10.
Factor in the mutual failure of Knox and Sollecito and Guede to implicate each other; I’ll assign this the same .05 probability that I did before. So now we have a probability of >8x10^-12.
If brazil84 assigns .85 probability to the prosecution’s account being correct, and .9 probability to their involvement, (I believe he may have been simply giving this as an example rather than actually asserting this,) then let’s suppose that the remainder of the probability lies in Knox and Sollecito being physically present and encouraging the attacker, but not entering the room or participating themselves. How much weight would we have to give to the prosecution’s evidence against them to get from our current probability to 5%? Their evidence would have to be more than 6.2x10^9 times more likely given Knox and Sollecito’s guilt than given their innocence.
As I stated in my earlier comment, I think that crediting them with a likelihood ratio of 50 times, including Knox’s mannerisms at the scene, is generous to their case.
Some additional thoughts on the arguments of the prosecution...
As I said in my first probability analysis, the arguments the prosecution put forward are mostly in the form of “This action seems more suspicious for an innocent person than a guilty person.” But for much of their case, I think even that is giving them too much credit. Let’s look at some examples.
The three-second phone call from Knox to Kercher. This is unusual behavior, right? Well, I’ve done this more than a few times myself, but I agree, it is pretty unusual. So how much weight would I assign that evidence? None. The question is not how unusual the behavior is, but how much more unusual if she’s innocent than if she was involved in the murder. If I know that Amanda Knox was going, later that day, to come across Meredith Kercher being attacked, and encourage or participate in the attack, the three second phone call will not be any less confusing to me. Stalking and incessant phone calls are certainly more easily explained in light of a murder, but I know of no evidence whatsoever to suggest that murderous inclinations and unusually short phone calls are associated, nor can I think of any particular reason they would motivate one.
The time at which Amanda Knox claims to have woken up. The Massei-Cristiani report finds suspicion in the fact that Knox claimed to have gotten up at 10:00-10:30 AM that day, while her roommates characterized her as an early riser. For a person to violate their established sleep habits is unusual, although still normal enough that their say-so is generally enough to establish it as probably true absent additional reason to doubt them. I am personally not an early riser, but this morning I got up and left the house for a walk at 6:00 AM, while the stars were still bright out (which is in fact what got me thinking about the prosecution’s report in the first place.) But Amanda Knox was a murder suspect, doesn’t that give us additional reason to doubt her claim? Well, no. Whether she was up at 10:30 or hours earlier has no bearing on the strength of her alibi. We know that the murder didn’t happen in the morning, and we know that she was in other places later in that day prior to the murder; it’s not as if she failed to establish that she ever left the house. Being involved in the murder doesn’t give her any additional reason to claim that she got up at 10:30.
Combined with misrepresentations and mistreatments of forensic evidence, including claims like the pattern of glass being inconsistent with a break-in, which was found to be simply false when tested, a large part of the prosecution’s body of arguments does not represent any weight as evidence at all.
Generally speaking under American law, if Knox and Sollecito intended for Guede to commit a serious crime against Kercher (such as rape or robbery), and Guede killed Kercher in the process, then Knox and Sollecito would be guilty of murder.
Basically yes, if by “prosecutor correct” you mean the prosecutor is correct that Knox and Sollecito physically participated in the murder. The prosecutor’s scenario contained a lot of details; it’s very likely that some of those details are wrong.
I disagree, depending on how you define “incompetent.” Let’s suppose you are the prosecutor—you are confident that Knox and Sollecito were involved in the murder but you don’t know exactly what happened. The police have brought you evidence suggesting that Knox stabbed Kercher. You know that in order to convict, you need to give the jury a plausible scenario for what happened. What choice do you have but to make your best guess and go with it?
Perhaps a better argument can be made that the prosecutors were unethical for advancing their speculations about sex games and such to the jury. It would be interesting to see a translation of the transcript of the prosecutor’s opening and closing arguments.
Regardless of what Knox and Sollecito intended, they are guilty of none of Guede’s crimes unless they acted to participate in some way. But I agree that if Knox had committed conspiracy to rape (or rob, or whatever) Kercher with Guede, then she would be guilty of any murder Guede committed in furtherance or in consequence of that conspiracy. I’ll agree to a similar statement about aiding and abetting, but I don’t remember the technical language. (And all of this is American law, which is of limited relevance in Italy).
A prosecutor’s theory is correct if it includes all the true, morally relevant details. Further, it contains no false, morally relevant details. Compare the a prosecutor saying:
(1) “The defendant shot the victim with a .44 caliber gun, but I don’t know the model of the gun.”
(2) “The defendant shot the victim, but I don’t know the caliber of the gun”
Assuming the statements are true descriptions of what actually occurred, the prosecutor saying (1) is correct and the prosecutor saying (2) is pushing the envelope. Further, a prosecutor who says, “The defendant shot the victim, and the killing was planned ahead of time,” when the defendant actually decided to kill during the meeting is incorrect.
As the last example shows, a defendant can be guilt of the crime charged. But I assert that prosecutors assert stories that contain morally relevant that are inconsistent with what actually happened far less than 5% of the time. And when inconsistencies happen, it suggests that the prosecutor is wrong on other morally relevant details.
From what I understand of the prosecutor’s theory of the case, it contained many additional morally relevant details beyond simply Knox and Sollecito’s physical involvement in Kercher’s murder.
For the reasons I just discussed, your probability distribution seems inconsistent with the evidence. Specifically, you rescue Knox and Sollecito’s guilty from possible errors by the prosecutor by proposing a theory that even you think has only a 5% chance of being true.
Agree.
I’m not sure what your point is here. Are we arguing about whether the prosecutor is competent? About whether he is unethical?
Do you agree that as a practical matter, in order to have a good chance of securing a conviction, a prosecutor needs to present a coherent narrative to the jury?
I don’t understand this sentence. I’m reluctant to guess at what you mean, but please keep in mind that “prosecutor correct” can be taken different ways. Does it mean simply that he is correct about physical involvement; or does it mean that his entire narrative is substantially correct?
My previous post was trying to define what I meant by correctness for the purpose of the probability analysis I did. In brief, a prosecutor is correct if the theory presented to the jury contains no morally relevant facts that are contradicted by what actually occurred. (To clarify what I mean by moral relevance: Did the burglar go in the unlocked front or unlocked back door? Legally relevant, but not morally relevant).
Thus, I think the answer to your question is that the prosecutor is incorrect if the narrative does not correctly describe what actually occurred. As I noted above, the prosecutor’s incorrectness is not inconsistent with the defendant’s legal guilt. But I assert that prosecutor’s incorrectness is suggestive of legal innocence because it shows the prosecutor doesn’t know what happened.
Sure, juries like stories. That isn’t precisely a rationalist feature of juries.
In the Kercher case, the prosecutor’s case was filled with morally relevant facts (e.g. the sex cult/orgy thing) that were wildly implausible and not supported by any evidence that I’m aware of. Then, more plausible assertions (i.e. it was impossible to enter the apartment via the fire escape) were contradicted. Since I independently believe that (P(defendant guilty of charged crime) & P(prosecutor incorrect)) is small, I’m left to estimate a very small likelihood that Knox is guilty.
Fine, but you don’t get to redefine the terms I used and then use your different definition to conclude that my probability estimate doesn’t make sense. That would be a fancy way of strawmanning me.
When I basically agreed that I believe that P (prosecutor correct) is 85%, I was careful to make clear what I meant by “prosecutor correct.” Please do not pretend that I meant (or that we were discussing) something different.
It was not my intent to strawman you. I seem to have misunderstood your position. Let’s return to basics.
Let’s define “no orgy, but guilty” as the state in which Knox and Sollecito are legally guilty of Kercher’s murder, but many of the prosecutor’s asserted morally relevant facts are false. By definition:
P(guilty) = P(prosecutor correct) + P(no orgy, but guilty)
P(no orgy, but guilty) = P(prosecutor incorrect) & P(guilty) [If I understand the terminology correctly]
Since you seem to agree that the prosecutor made incorrect assertions of morally relevant facts, your value of P(prosecutor correct) is negligibly small. Why do you think that the prosecutor made the assertions? Before you say “narrative,” let me assert that many of the prosecutor’s assertions could have been replaced with “I don’t know” with limited loss of coherence to the narrative.
Can you give me a specific example?
Ok, you’ve said it. And now my answer:
Narrative. I do agree that the prosecutor could have omitted a few things and still had a somewhat coherent narrative. But if he got rid of all the speculation, what he would be left without any motive or explanation for why Knox and Sollecito acted in concert with Guede.
Quoting wikipedia: “[The prosecutor alleged] that the murder was part of a Satanic ritual; that it was a sex game gone wrong; or that Kercher had refused to take part in an orgy.”
Honestly, a substantial part of my doubt of Knox and Sollecito’s guilt is the utter ridiculousness of those allegations. They were not necessary to provide narrative to the jury and could easily have been omitted, since there was apparently no evidence suggesting they were true.
Further, they are the type of allegations I would expect when the prosecutor is seeking to inflame, rather than rationally presenting the evidence.
Ok, to answer your question, I believe that the prosecutor made these assertions because he felt pressure to answer the question of Knox and Sollecito’s motives.
Satanic ritual? The prosecutor could have just said, “The roommates had a big fight.” In terms of evidence of guilt, there are two major problems in the “satanic ritual” theory that are not present in a more mundane theory. The satanic ritual theory proves too much, in that there is almost no evidence that could possibly be inconsistent with it. Second, the theory functions to demonize the defendants, which is not consistent with strong actual evidence of guilt.
And those theoretical arguments ignore the practical fact that there was no evidence suggesting interest in satanic ritual by either defendant, which I consider important given the general implausibility of the theory given what I know about society (There are practically no murderous satanists in real life).
Apparently he did also propose more mundane motives. Here’s the full quote from Wikipedia:
When you quoted Wikipedia, why did you omit the part about the cleaning roster?
Because I’m interested in why the prosecutor kept talking after he articulated that theory.
Either theory is just speculation, agreed? I mean, if it’s unethical for the prosecutor to argue that there was a satanic ritual, it’s also unethical for him to argue that it’s a dispute over who should clean the apartment. Agreed?
I’m not trying to say the prosecutor is unethical. I’m saying that outlandish claims from the prosecutor are evidence of innocence. It’s just a Big Lie
And if the prosecutor really has no idea why defendant committed a crime, that’s also evidence of innocence, right?
One could rationalize the prosecutor’s behavior by saying that lower standards apply to statements to the media than statements in court. But the reason that it’s unethical to try cases in the media is that this reduces the likelihood that the trial will be correlated with the truth. I attribute knowledge of that principle to the prosecutor, so his decision to try the case in the media suggests that the defendants are innocent.
Ok, I thought we were discussing whether the prosecutor was ethical.
I agree. And the decision to arrest Knox is evidence of guilt.
We were discussing whether the prosecutor is competent.
No, it isn’t. The decision to arrest is evidence that the authorities think they have evidence of guilt. But the arrest is not independent of the underlying evidence and does not make the underlying evidence stronger.
Ok, then why are you bringing up the question of whether the prosecutor’s assertions are evidentiary?
Similarly, the prosecutor’s proposal that there may have been a Satanic ritual involved is not independent of the underlying (lack of) evidence of motive and does not make the underlying evidence weaker.
The prosecutor’s theory is not evidence of the same type as DNA. But it tells us something about how strong the prosecutor thinks the legally relevant evidence is.
The reference to satanic ritual and sex games is an appeal to applause lights. It is political machinations, not an empirically based search for truth. The fact that the prosecutor made that appeal, rather than sticking with a plausible story, shows that the prosecutor doesn’t think the facts support guilt or doesn’t care.
As an observer, I’m allowed to give weight to the fact that one side of a dispute doesn’t seem to be trying to determine the facts.
A prosecutor will always make the strongest case they think they can, and juries do not comprise perfect rationalists.
The decision to arrest the defendant is not the same type as DNA. But it tells us something about how strong the authorities think the legally relevant evidence is.
I would say the amount of weight depends on the extent you are relying on unsupported assertions of that side. For example, if we knew nothing at all about the case except the prosecutor saying “Trust me, I have reviewed the evidence and I am satisfied that the Defendant is guilty, it would be very important that the prosecutor is proposing a seemingly outlandish hypothesis.
This is the expertise problem. You and I do not have access to all the evidence, so we look to the experts. If anyone is an expert on the strength of the case, it is the prosecutor. And a prosecutor referencing satanic rituals and sex cults is like a psychologist referencing the “Niceness Gene” to explain some behavior. It isn’t impossibly wrong, but it strongly suggests the expert doesn’t know what he is talking about.
So in your view, any probability assessment of Knox’s guilt which does not put a lot of weight on the prosecutor’s credibility is deeply flawed. Agreed?
No. The physical evidence does not support Knox’s guilt. But the fact is that the prosecutor and other Italian authorities are engaging in a lot of motivated cognition to support the theory of guilt even when the evidence doesn’t support it.
The motivated cognition is so blatant that it persuaded me that Knox was probably not guilty even before I looked at the evidence. I suspect that if the world were more rational, my reasoning would be less likely to provide useful insight.
How do you know this if you do not have access to all of the evidence and therefore must “look to the experts”?
It seems to me that you and I both have access to enough of the evidence to make a decent estimate of the probability of Knox’s guilt without relying on someone who has superior access to the evidence.
I’m not sure I would agree with this, FWIW.
My estimate of how ethical it is for a prosecutor to propose theory X about a defendant’s motive increases with how common X is, and decreases with how sensational X is. Roommates arguing over cleaning an apartment is both more common and less sensational than satanic rituals, so my estimates of how ethical it is to propose those theories differs.
Of course, maybe there’s some particularly salient threshold that both examples exceed. If so, it’s worth articulating that threshold explicitly… when it comes to propounding theories about motive, it’s not clear to me what that threshold is or ought to be.
Err, that should be “very likely to leave biological traces behind in sufficient amounts that investigators are likely to find and match them.”