Continuation Part 12: Amanda Knox/Raffaele Sollecito

[Rolfe]

 

[anglolawyer]

But, as we know, science is different in Italy. As Grinder used to say, 'three compatibles equals a match' and I used to add ' and nobody may be accused of lying' to which Grinder would then say 'it's an opera' and I would reply, 'panto'.

 

[Numbers]

Alex_K on IIP/IAF provided the following quote from the ruling by the UK (Northern Ireland) judge who acquitted Sean Hoey of bombing charges in 2007.



I will point out the ECHR may also go to rulings by domestic courts to look for precedents regarding standards for reliability. Of course, they offer the ECHR position after consideration of other information and precedents.

ETA: In the quote, note: At every stage of handling, the prosecution must demonstrate that for each item of the evidence (item with potential DNA material) "that the item has been preserved at every stage free from the suspicion of interference or contamination"." ...it is for the prosecution to establish the integrity and freedom from possible contamination of each item throughout the entirety of the period between seizure and any examination relied upon."

This was not done by the Italian police and prosecution in this case, in particular for the bra clasp, but also for the knife.

Bumped down to point out the quote is from Judge Weir.

The prosecution's responsibility to assure the reliability, integrity, and freedom from contamination of evidence is also the standard in the US.

 

[anglolawyer]

Bumped down to point out the quote is from Judge Weir.

The prosecution's responsibility to assure the reliability, integrity, and freedom from contamination of evidence is also the standard in the US.

It's time I repeated my pi analogy. I am not sure you were around the last time I trotted this out. Say a teacher sets his maths class the task of calculating pi to 10 places and producing the result along with the methodology. Nobody manages to do it except one student who produces the correct result but is unable to show any workings. Does the student get a pass or a fail? Clearly, a fail. She might have looked it up on the internet, as I just did:

3.1415926535

and I promise you I have not the faintest idea how to calculate pi nor even if it is the result of a calculation at all.

Stefanoni is in a closely analogous position. She has to show not only a profile, correct in all 16 loci, but also her methods. That she has not done so is the reason Conti-Vechiotti say this:

Taking into account that none of the recommendations of the international scientific community relative to the treatment of Low Copy Number (LCN) samples were followed, we do not accept the conclusions regarding the certain attribution of the profile found on trace B (blade of knife) to the victim Meredith Susanna Cara Kercher, since the genetic profile, as obtained, appears unreliable insofar as it is not supported by scientifically validated analysis;

Note, this has nothing whatever to do with contamination. It is an entirely separate, free-standing and sufficient by itself ground for rejecting her evidence. It is exactly the same as the student who cannot show her methodology to the teacher.

 

[Anode]

It beats me why they don't just agree the rock was thrown from outside but the burglary was still staged.

Probably because no one believes that Amanda could throw a big rock that far. And it's doubtful that Raf could either. But Rudy could...

 

[anglolawyer]

Probably because no one believes that Amanda could throw a big rock that far. And it's doubtful that Raf could either. But Rudy could...

I don't think that's why. If the rock was thrown fom inside then it's definitely a staging, which is a key tenet of guilter thinking, too precious to abandon (even though there are plausible, or no less implausible, scenarios in which Guede by himself did the staging) and thus to be defended at all costs, even of losing one's reputation for unbiased thinking.

 

[Numbers]

It's time I repeated my pi analogy. I am not sure you were around the last time I trotted this out. Say a teacher sets his maths class the task of calculating pi to 10 places and producing the result along with the methodology. Nobody manages to do it except one student who produces the correct result but is unable to show any workings. Does the student get a pass or a fail? Clearly, a fail. She might have looked it up on the internet, as I just did:

3.1415926535

and I promise you I have not the faintest idea how to calculate pi nor even if it is the result of a calculation at all.

Stefanoni is in a closely analogous position. She has to show not only a profile, correct in all 16 loci, but also her methods. That she has not done so is the reason Conti-Vechiotti say this:

Taking into account that none of the recommendations of the international scientific community relative to the treatment of Low Copy Number (LCN) samples were followed, we do not accept the conclusions regarding the certain attribution of the profile found on trace B (blade of knife) to the victim Meredith Susanna Cara Kercher, since the genetic profile, as obtained, appears unreliable insofar as it is not supported by scientifically validated analysis;

Note, this has nothing whatever to do with contamination. It is an entirely separate, free-standing and sufficient by itself ground for rejecting her evidence. It is exactly the same as the student who cannot show her methodology to the teacher.

Yes. The demonstration of reliability requires complete documentation of methodology, including validation of methods and quality control/ assurance methods and records.

LCN DNA methodology without testing of multiple samples from the same item of evidence is suspect, because of well-known statistical DNA artifacts that occur in LCN profiling. Obtaining a LCN profile without finding any statistical artifacts on a single sample is suspiciously lucky, in my opinion.

Your analogy is imperfect, as you may know. The student is being asked to demonstrate knowledge of a method to calculate pi, of which there are many [see: http://en.wikipedia.org/wiki/Pi; http://mathworld.wolfram.com/PiFormulas.html]. In forensic testing and scientific discovery, the expected result is truly unknown.

ETA: The V&C statement you quoted is a polite way of saying, "we think you cheated."

 

[anglolawyer]

Yes. The demonstration of reliability requires complete documentation of methodology, including validation of methods and quality control/ assurance methods and records.

LCN DNA methodology without testing of multiple samples from the same item of evidence is suspect, because of well-known statistical DNA artifacts that occur in LCN profiling. Obtaining a LCN profile without finding any statistical artifacts on a single sample is suspiciously lucky, in my opinion.

Your analogy is imperfect, as you may know. The student is being asked to demonstrate knowledge of a method to calculate pi, of which there are many [see: http://en.wikipedia.org/wiki/Pi; http://mathworld.wolfram.com/PiFormulas.html]. In forensic testing and scientific discovery, the expected result is truly unknown.

I did not know and now you have spoiled my weekend

ETA In idiot's language, what does 'validation of methods' mean?

 

[Bill Williams]

I have yet to figure out exactly what Machiavellian's position either.
Assuming that it is that the rock was thrown from outside, I am unclear the difference between staging and not staging.

One of the things which is confusing about Machiavelli's position is this:

When it's pointed out to him that there is no forensic evidence of Knox in the murder room, he comes back with:

"You do not have to be found to be in the room to be convicted of murder."

That one is one of his best rhetorical escape routes.... it's just hard to figure out, then, what he's arguing.

 

[Numbers]

I did not know and now you have spoiled my weekend

ETA In idiot's language, what does 'validation of methods' mean?

I had explained validation to a small extent in a post lost due to my clumsy skills at the keyboard.

Validation, in general, is a statistically valid exercise - a study that involves the actual physical processes - that demonstrates a process or method is fit for use.

For example, in DNA work, one should do repeated analyses de novo (from a starting point known to be validated) on known DNA samples, using a statistical model to determine the appropriate number of repetitions and of various types of DNA materials, and check that the DNA method produces the expected result and noting any statistical variation (statistical error), with quantification of the statistical error rate and assurance that such error rate is less than or equal to some predetermined acceptable limit.

The point is that physical processes - such as lab tests or manufacturing methods - have some potential for variation, or even producing fully erroneous results or products. One needs to assure that what results from the process meets the expectations of use. For forensics in a criminal case, the standard must be essentially certainty that the DNA profile is indeed what was present in the sample. There are many artifacts that can occur in DNA profiling, and those artifacts are larger in number and more severe in effect on interpretation for LCN DNA.

 

[Bill Williams]

Yes. The demonstration of reliability requires complete documentation of methodology, including validation of methods and quality control/ assurance methods and records.

I did not know and now you have spoiled my weekend

ETA In idiot's language, what does 'validation of methods' mean?

This is what I was trying to get at, when referring to the photoshopped pic Machiavelli claimed he made of the bedsheet-stain, kitchen-knife equivalence.

Unless he could take us through the steps of how that picture was produced, where the faux-overlay of the knife on the stain (in the second pic) just so happens to be taken from the same vantage point as the stain....

There are a whole host of photographic issues he'd have to document before anyone could even take the photoshop seriously. Him saying, "I measured" as his only quality-control statement is absurdly laughable!

But then again, Machiavelli has already laid out the principle. When an expert testifies in court, they do not need to bring evidence. The court assumes they are telling the truth, even without evidence. It then becomes the defence's job to prove them wrong.....

..... which, in lieu of evidence before anyone, becomes a tad difficult! (Even before considering the reverse of burden of proof.)

 

[Bill Williams]

It's time I repeated my pi analogy. I am not sure you were around the last time I trotted this out. Say a teacher sets his maths class the task of calculating pi to 10 places and producing the result along with the methodology. Nobody manages to do it except one student who produces the correct result but is unable to show any workings. Does the student get a pass or a fail? Clearly, a fail. She might have looked it up on the internet, as I just did:

3.1415926535

and I promise you I have not the faintest idea how to calculate pi nor even if it is the result of a calculation at all.

Stefanoni is in a closely analogous position. She has to show not only a profile, correct in all 16 loci, but also her methods. That she has not done so is the reason Conti-Vechiotti say this:

Taking into account that none of the recommendations of the international scientific community relative to the treatment of Low Copy Number (LCN) samples were followed, we do not accept the conclusions regarding the certain attribution of the profile found on trace B (blade of knife) to the victim Meredith Susanna Cara Kercher, since the genetic profile, as obtained, appears unreliable insofar as it is not supported by scientifically validated analysis;

Note, this has nothing whatever to do with contamination. It is an entirely separate, free-standing and sufficient by itself ground for rejecting her evidence. It is exactly the same as the student who cannot show her methodology to the teacher.

(1/4)pi=4tan^(-1)(1/5)-tan^(-1)(1/(239)).

You are welcome.

 

[Bill Williams]

But, as we know, science is different in Italy. As Grinder used to say, 'three compatibles equals a match' and I used to add ' and nobody may be accused of lying' to which Grinder would then say 'it's an opera' and I would reply, 'panto'.

You guys are well on your way to inventing a religion.

 

[NotEvenWrong]

Did Grinder actually say three compatibles equals a match? What the heck does that even mean?

 

[Numbers]

(1/4)pi=4tan^(-1)(1/5)-tan^(-1)(1/(239)).

You are welcome.

There are also many infinite-series approximations, valid to any desired place. They are exact in the (infinite) limit (which can't be reached, for pi, because it is an irrational - can't be expressed as a ratio or fraction).

 

[Numbers]

Did Grinder actually say three compatibles equals a match? What the heck does that even mean?

It's a joke about the bizarre Italian judicial lack of standards. They call things that have a questionable or speculative relationship "compatible", then "sum" these "compatibles" to claim a proof.

 

[LondonJohn]

I had explained validation to a small extent in a post lost due to my clumsy skills at the keyboard.

Validation, in general, is a statistically valid exercise - a study that involves the actual physical processes - that demonstrates a process or method is fit for use.

For example, in DNA work, one should do repeated analyses de novo (from a starting point known to be validated) on known DNA samples, using a statistical model to determine the appropriate number of repetitions and of various types of DNA materials, and check that the DNA method produces the expected result and noting any statistical variation (statistical error), with quantification of the statistical error rate and assurance that such error rate is less than or equal to some predetermined acceptable limit.

The point is that physical processes - such as lab tests or manufacturing methods - have some potential for variation, or even producing fully erroneous results or products. One needs to assure that what results from the process meets the expectations of use. For forensics in a criminal case, the standard must be essentially certainty that the DNA profile is indeed what was present in the sample. There are many artifacts that can occur in DNA profiling, and those artifacts are larger in number and more severe in effect on interpretation for LCN DNA.

Exactly - and I think that courts (particularly, it would appear, in Italy....) simply have not yet come to grasp the concept that the problems of reliability are magnified disproportionately greatly in low template DNA analysis compared with "standard" PCR DNA work.

I think that courts (particularly, it would appear, in Italy....) have become comfortable with the issues surrounding "standard" DNA evidence, and erroneously think that the same rules and standards apply to low template DNA evidence - or if not exactly the same rules and standards, then only a small tweak. In reality, an entire new set of much, much more stringent rules and standards is required in order to allow the admission of low template DNA evidence.

Unfortunately, I once again lay a certain amount of blame in this specific case at the feet of the defence lawyers in the Massei trial. They could (and easily should) have produced an expert in DNA analysis to show the court unequivocally that 1) the standards required for low template analysis are dramatically different from those required for "standard" DNA analysis, and 2) the low template evidence in this case manifestly and definitively came nowhere even close to meeting those standards.

PS: Pi to 10 dp is 3.1415926536 - one of my strange party tricks is an ability to recite Pi to 50 dp (3.14159265358979323846264338327950288419716939937511 since you ask )

PPS: for non-mathematicians who might be slightly confused, the reason why the last digit of Pi to 10dp is 6 rather than 5 is that the following dp is >=5 (it is in fact 5) - this means that to 10dp one needs to round up the 5 to a 6.....

 

[Bill Williams]

Nuovo presidente della Repubblica, un voto anche per Giuliano Mignini. Umbri tra tweet e scatti incerti......

Is it true that Mignini is running for president of Italy?

Say it ain't so!

 

[carbonjam72]

Exactly - and I think that courts (particularly, it would appear, in Italy....) simply have not yet come to grasp the concept that the problems of reliability are magnified disproportionately greatly in low template DNA analysis compared with "standard" PCR DNA work.

I think that courts (particularly, it would appear, in Italy....) have become comfortable with the issues surrounding "standard" DNA evidence, and erroneously think that the same rules and standards apply to low template DNA evidence - or if not exactly the same rules and standards, then only a small tweak. In reality, an entire new set of much, much more stringent rules and standards is required in order to allow the admission of low template DNA evidence.
Unfortunately, I once again lay a certain amount of blame in this specific case at the feet of the defence lawyers in the Massei trial. They could (and easily should) have produced an expert in DNA analysis to show the court unequivocally that 1) the standards required for low template analysis are dramatically different from those required for "standard" DNA analysis, and 2) the low template evidence in this case manifestly and definitively came nowhere even close to meeting those standards.
PS: Pi to 10 dp is 3.1415926536 - one of my strange party tricks is an ability to recite Pi to 50 dp (3.14159265358979323846264338327950288419716939937511 since you ask )

PPS: for non-mathematicians who might be slightly confused, the reason why the last digit of Pi to 10dp is 6 rather than 5 is that the following dp is >=5 (it is in fact 5) - this means that to 10dp one needs to round up the 5 to a 6.....

When did the defense become aware that Stef was working with Low template DNA? only very late in the first trial, no? They did move to invalidate the trial though, because the DNA evidence was misrepresented before Micheli, din't they, or have I confused things?

I just think its flawed reasoning to suggest there is anything the defense could have done to get a fair verdict from a biased process.

 

[LondonJohn]

When did the defense become aware that Stef was working with Low template DNA? only very late in the first trial, no? They did move to invalidate the trial though, because the DNA evidence was misrepresented before Micheli, din't they, or have I confused things?

I just think its flawed reasoning to suggest there is anything the defense could have done to get a fair verdict from a biased process.

No, they knew from well before the trial that the "evidence" was low template level. What they didn't know was exactly how minuscule it was, and nor of course did they have the EDFs to properly understand the magnitudes, amplifications and noise levels/artifacts.

In my view, in the Massei trial they should have put an Italian DNA scientist with an international reputation on the stand to point all this out: 1) we categorically do not have the data to allow us to conduct a proper defence; 2) regardless of that, we know that the quanta of DNA being presented are well within the low-template range; 3) that being the case, a whole different - and much more stringent - set of rules applies to the admissibility and reliability of such DNA evidence; 4) we can show conclusively to the court that these rules have categorically not been followed in this case; and therefore 5) the courts should not be considering any of the DNA evidence reliable, and in fact should consider it inadmissible.