An interesting piece of easily quantifiable Bayesian evidence could be phones being switched off overnight (dropping off the network) - how often did Knox do that? If she only did that once in many hundred days, on that night in particular, then that could be a very huge amount of evidence. Or she may have done that few times a week, in which case it’s irrelevant.
If she only did that once in many hundred days, on that night in particular, then that could be a very huge amount of evidence.
....No. Not even slightly. This line of questioning MIGHT be relevant if you didn’t already have the killer identified, with overwhelming physical evidence pointing towards them. You don’t need to explain why Knox turned her phone off, because you already have the killer and every single piece of physical evidence at the crime scene accounted for.
Well, for what it’s worth their wounds and bruises guy didn’t think it was a single killer. And when someone’s murdered at their own place in the dead middle of the night, often the cohabitants are involved.
The prosecutor claimed that from the number of stab wounds, it was unlikely that a single person could have inflicted them all. However, the number of stab wounds was by no means an outlier among murders known to have a single perpetrator (I do not have detailed statistics on this subject, but merely from my limited experience with case studies on the subject I have encountered quite a few cases which involved many more stab wounds from a single perpetrator.) Considering the pervasive incompetence their forensics teams demonstrated over the course of the case, I would assign very little weight to this.
The prosecution also presented pieces of “evidence” such as Knox placing extremely short phone calls to Kercher, too short to transmit any message. This and many other points raised by the prosecution fit the pattern of behavior that seems unusual, and so is presented as evidence for suspicion of murder, despite the fact that the behavior doesn’t make more sense if we suppose she was involved in the murder.
If Knox had a murder likelihood of 1/1000 after conditioning on the evidence that Kercher had been murdered, but before accounting for other evidence, and she’s then observed to have engaged in unusual actions with a 1/1000 probability, it makes no difference towards her likelihood as a culprit if they’re not actions which are more likely in the event that she’s actually guilty. We can come up with post-hoc explanations for why the unusual things might be related to involvement in the murder, and this kind of reasoning appears to have constituted a large part of the prosecution’s case, but if we don’t have any prior reason to suppose that guilt of murder is associated with such behaviors, then these explanations will tend only to be rationalizations of preexisting suspicion.
An interesting piece of easily quantifiable Bayesian evidence could be phones being switched off overnight (dropping off the network) - how often did Knox do that? If she only did that once in many hundred days, on that night in particular, then that could be a very huge amount of evidence. Or she may have done that few times a week, in which case it’s irrelevant.
....No. Not even slightly. This line of questioning MIGHT be relevant if you didn’t already have the killer identified, with overwhelming physical evidence pointing towards them. You don’t need to explain why Knox turned her phone off, because you already have the killer and every single piece of physical evidence at the crime scene accounted for.
Well, for what it’s worth their wounds and bruises guy didn’t think it was a single killer. And when someone’s murdered at their own place in the dead middle of the night, often the cohabitants are involved.
The prosecutor claimed that from the number of stab wounds, it was unlikely that a single person could have inflicted them all. However, the number of stab wounds was by no means an outlier among murders known to have a single perpetrator (I do not have detailed statistics on this subject, but merely from my limited experience with case studies on the subject I have encountered quite a few cases which involved many more stab wounds from a single perpetrator.) Considering the pervasive incompetence their forensics teams demonstrated over the course of the case, I would assign very little weight to this.
The prosecution also presented pieces of “evidence” such as Knox placing extremely short phone calls to Kercher, too short to transmit any message. This and many other points raised by the prosecution fit the pattern of behavior that seems unusual, and so is presented as evidence for suspicion of murder, despite the fact that the behavior doesn’t make more sense if we suppose she was involved in the murder.
If Knox had a murder likelihood of 1/1000 after conditioning on the evidence that Kercher had been murdered, but before accounting for other evidence, and she’s then observed to have engaged in unusual actions with a 1/1000 probability, it makes no difference towards her likelihood as a culprit if they’re not actions which are more likely in the event that she’s actually guilty. We can come up with post-hoc explanations for why the unusual things might be related to involvement in the murder, and this kind of reasoning appears to have constituted a large part of the prosecution’s case, but if we don’t have any prior reason to suppose that guilt of murder is associated with such behaviors, then these explanations will tend only to be rationalizations of preexisting suspicion.