If read your post correctly, you feel that you can discount as pretty much irrelevant the opinions of quite a lot of people (jurors, police, etc), on the simple basis that people can be spectacularly wrong on occasion. ( I’m really not sure about this.)
In fact, as far as I can tell, you start from “clean” priors and do all your updating based only on the “physical evidence”; no opinions entering your calculation.
This seems almost OK, but something’s nagging at me: how can you obtain thirty bits of confidence in your estimate using only evidence received from other people, via the Internet?
I’m also not sure about this, but your post seems to imply that a “good Bayesian” would be expected to assign that amount of confidence to his answer after only a couple of hours of surfing the Internet. I’m not saying that’s impossible, but it really sounds very unlikely to me.
I’d very much like to see a chain of numerical reasoning that reasonably puts a 1:1000 upper-bound on the likelihood that Guédé is innocent, without starting with implicit assumptions of 100% certainty about data read from the net.* If you think an hour on the Internet is enough to reach that kind of certainty, I don’t see why writing the calculations (for an upper bound, not the precise value) would take much longer, assuming that one would be gathering data and doing the calculations in one’s head during that hour.
(*: EDIT for clarification. By this I mean that, for instance, given claimed evidence E in support of theory T, you don’t update on the probability of T given E, but update on the probability of T given “I’ve read on the Internet that E”. Of course, many claims of E on the Internet have some weight, but I doubt two hours of Internet reading can add lots of weight on non-trivial subjects.)
The discounting of everyone and everything that implies guilt is the only way someone can make an argument that Knox and Sollecito are innocent. The computer shows no activity = experts are wrong. ISP shows no activity = ISP is wrong. Three different witnesses saw them = all wrong. Luminol shows footprints that match Knox and Sollecito = false positive for the Luminol. Expert says Knox shoe print in victim’s room = expert wrong. Expert says Sollecito’s bloody footprint in bathroom = expert is wrong. DNA = all wrong. This goes on and on.
Given the scenario of accepting that dozens of experts and witnesses are wrong or accepting that the evidence is accurate and they killed Meredith I think the logical choice is obvious.
It is easy to see why an hour on the internet beats a year in the courtroom is just foolish. The idea that a bunch of white knights on the internet could match a small army of experts is ludicrous
I find it much more ludicrous that a small army of experts would have so little disagreement if they weren’t privileging a hypothesis. How likely is it that interviews at 0145 and 0545 would be confused and contradictory if the suspect was innocent, as compared to if the suspect was guilty?
Personally, I think that a slightly confused interview history is slight Bayesian evidence of guilt, because someone with a prepared lie is less likely to appear confused about it than someone trying to tell the truth.
For one, because they do. The website which contains the claims of evidence has several small internal inconsistencies and notes several cases where the prosecution witnesses do not confirm identical beliefs.
One of the damning things about the DNA evidence is that the experts claim odds of “One in a trillion or two” and “Ten billion to one” that the DNA matches a random person. That requires that the odds of a given person having an identical twin about which they are unaware be less than that, and/or that DNA from a given crime scene be expected to match from about one-half of a living human to a miniscule fraction of all humans, if all humans were tested.
The standards for expert testimony are not very strict, either in the United States or Europe, and there remains significant internal controversy on some important matters within the field of DNA testing.
I’ve also seen direct conflicts between the claimed testimony on that page and on other reporting sources, although I lack the tools to evaluate which claim is correct at this time.
I’m a bit curious about something:
If read your post correctly, you feel that you can discount as pretty much irrelevant the opinions of quite a lot of people (jurors, police, etc), on the simple basis that people can be spectacularly wrong on occasion. ( I’m really not sure about this.)
In fact, as far as I can tell, you start from “clean” priors and do all your updating based only on the “physical evidence”; no opinions entering your calculation.
This seems almost OK, but something’s nagging at me: how can you obtain thirty bits of confidence in your estimate using only evidence received from other people, via the Internet?
I’m also not sure about this, but your post seems to imply that a “good Bayesian” would be expected to assign that amount of confidence to his answer after only a couple of hours of surfing the Internet. I’m not saying that’s impossible, but it really sounds very unlikely to me.
I’d very much like to see a chain of numerical reasoning that reasonably puts a 1:1000 upper-bound on the likelihood that Guédé is innocent, without starting with implicit assumptions of 100% certainty about data read from the net.* If you think an hour on the Internet is enough to reach that kind of certainty, I don’t see why writing the calculations (for an upper bound, not the precise value) would take much longer, assuming that one would be gathering data and doing the calculations in one’s head during that hour.
(*: EDIT for clarification. By this I mean that, for instance, given claimed evidence E in support of theory T, you don’t update on the probability of T given E, but update on the probability of T given “I’ve read on the Internet that E”. Of course, many claims of E on the Internet have some weight, but I doubt two hours of Internet reading can add lots of weight on non-trivial subjects.)
The discounting of everyone and everything that implies guilt is the only way someone can make an argument that Knox and Sollecito are innocent. The computer shows no activity = experts are wrong. ISP shows no activity = ISP is wrong. Three different witnesses saw them = all wrong. Luminol shows footprints that match Knox and Sollecito = false positive for the Luminol. Expert says Knox shoe print in victim’s room = expert wrong. Expert says Sollecito’s bloody footprint in bathroom = expert is wrong. DNA = all wrong. This goes on and on.
Given the scenario of accepting that dozens of experts and witnesses are wrong or accepting that the evidence is accurate and they killed Meredith I think the logical choice is obvious.
This site has all the transcripts and is in the process of translating them into English. http://themurderofmeredithkercher.com/
It is easy to see why an hour on the internet beats a year in the courtroom is just foolish. The idea that a bunch of white knights on the internet could match a small army of experts is ludicrous
I find it much more ludicrous that a small army of experts would have so little disagreement if they weren’t privileging a hypothesis. How likely is it that interviews at 0145 and 0545 would be confused and contradictory if the suspect was innocent, as compared to if the suspect was guilty?
Personally, I think that a slightly confused interview history is slight Bayesian evidence of guilt, because someone with a prepared lie is less likely to appear confused about it than someone trying to tell the truth.
I’m confused. Why would you expect dramatic expert disagreement in general on matters of fact?
For one, because they do. The website which contains the claims of evidence has several small internal inconsistencies and notes several cases where the prosecution witnesses do not confirm identical beliefs.
One of the damning things about the DNA evidence is that the experts claim odds of “One in a trillion or two” and “Ten billion to one” that the DNA matches a random person. That requires that the odds of a given person having an identical twin about which they are unaware be less than that, and/or that DNA from a given crime scene be expected to match from about one-half of a living human to a miniscule fraction of all humans, if all humans were tested.
The standards for expert testimony are not very strict, either in the United States or Europe, and there remains significant internal controversy on some important matters within the field of DNA testing.
I’ve also seen direct conflicts between the claimed testimony on that page and on other reporting sources, although I lack the tools to evaluate which claim is correct at this time.