FBI created a forensics disaster; sent hundreds of innocent people to (1 Viewer)

[apalmer]

Seems really odd that some folks want to defend the FBI in this instance.

Who did that? Big difference between defending the FBI and pointing out the flawed assumptions of the OP. Or did you miss that part?

 

[GroveDiesel]

Seems really odd that some folks want to defend the FBI in this instance.

Who did that? Big difference between defending the FBI and pointing out the flawed assumptions of the OP. Or did you miss that part?

While I did make an assumption in my OP, there's more evidence that it's a good one than a flawed one at this point.

 

[Slapdash]

The FBI is shady as hell. Look at how they killed that guy they were questioning about the Boston bombing. They basically executed him and he wasn't even armed. Then they did an investigation into it, and said they did nothing wrong...

Oh yeah, forgot about that one.

 

[Tennessee_ATO]

It sounds like some terrible stuff happened here and there ought to be a full investigation. Grove Diesel is right in suggestimg that the death penalty should perhaps be put on hold while we check the validity of these verdicts; I've made that argument for years.

Yet I also sense a typical reaction here which is to assign deliberate evil intention to the FBI and to government in general. As usual, I suspect mistakes and incompetence are the true culprits rather than design.

Every other time this has been revealed, it's always been a detective leaning on a forensics lab to come up with a particular conclusion. That's the only way this ever happens.

"Listen, Doc, this guy did it. We all know it. But his lawyer is getting the search thrown out on a technicality. We know he's the guy, we just need you to do whatever you can to make sure it's clear scientifically that he's the guy. We can't have you talking in probabilities, we need absolutes. We need you to be saying the same thing all the other testimony is going to say."

This isn't an accident or incompetence. It's a guy with his thumb on the scale. It's 100% deliberate.

Not really what I was referring to. I'm sure that what you described happens a lot. I just don't believe it's top down, ordered by above, which seems to be the implication of at least some people here. I probably shouldn't have used the words "mistakes or incompetence"- what I really meant was that I doubt it's a centralized effort.

Who said it was "top down ordered from above"?

You did, in the title of your thread: "FBI sent hundreds of people" etc. As if it was a directed effort, and not certain individuals who screwed with the evidence.

That makes no sense. Whether it was malice or incompetence, virtually all of the FBIs forensic examiners were giving false/misleading testimony for 30 years. I don't know how you can possibly argue that my thread title is incorrect.Furthermore, they trained up to 1000 other examiners to use the same false/misleading standards.

Whether it is malice or gross incompetence, the fact remains that they are responsible.

Where in the article does it say hundreds of innocent people were sent to prison? I think I missed that part.

I took a page out of the FBI's handbook on how I applied statistics. Probably should have included the phrase "may have".But I don't think it's a stretch. Only one jurisdiction has actually examined all their cases and 5 of the 7 ended up being exonerated. The percentage that are actually innocent is likely much lower than that, but with 3500 cases directly involving FBI testimony plus who knows how many cases involving examiners that those FBI examiners trained, I don't think hundreds is all that unbelievable.

Hundreds is almost a statistical certainty.

I'm sorry you don't seem to understand how math works.

Let's start with what we know for sure: We have 2500 cases in which federal forensic examiners declared a hair match. After a review of ~ 1300*, we have a known fail rate of 90%. That's ~2250 cases in which forensic examiners found conclusive matches when, in fact, there was no scientific basis to make that assertion. We know that in the only jurisdiction in which they have concluded investigations into all convictions using the false testimony (D.C.), 5 of 7 convicts were exonerated. That's a whopping ~71.43%. We know the fail rate for trials is ~95.9%.

We don't know how many cases went to trial out of that 2500 sample, but the article at least implies that there were 268 actual trials out of that ~1300* investigated thus far. That's a rate of ~20.62%, or a total of 515 trials total. Applying our known fail rate for trials tells us that we have 494 trials in which the federal forensic examiners provided substantively false testimony to convict the accused.

Applying the DC exoneration rate (71.43%) to those 494 trials involving false testimony to convict gives us ~353.

And then you have to drill down to the 500-1000 state forensic examiners who were taught by those federal guys "to testify the same way". How many trials/conviction do you think those guys have testified in? We have 28 federal examiners giving us ~515 trials in which they provided false testimony. Say they just trained 500 state examiners, and those 500 state examiners only had a fail rate of half of their trainers, and those state examiners testified as often as their federal counterparts (all assumptions that understate reality: there is no reason to believe that only 500 were trained, no reason to believe they would be twice as good as their federal trainers, and state examiners testify far more frequently than their federal counterparts).

That's ~9196 trials with testimony of hair matches from state examiners trained in the faulty (at best) or intentionally deceptive (at worst) techniques of the "elite" federal guys. Halve the federal fail rate to get a 47.95% fail rate for the inferior state guys. That's 4409 state trials in which bad testimony was used to convict.

Now, pick whatever rate you want for exoneration. The small sample we have says 5/7, or ~71.43%. Say that sample overstates the state rate by a factor of 5, resulting in an ~14.29% exoneration rate. That's 630 innocent folks convicted.

630 + 352 = 982. And that's underestimating what the data tells us at every step for which we don't have reliable data.

So, yeah, "hundreds" is almost a statistical certainty.

*And this is also likely understating the percentage. The article isn't clear about how many have been reviewed. It says that 2500 were identified, 1200 are yet to be reviewed, and 900 lab reports "are nearly compete, with about 1200 cases remaining". The implication is that they have either reviewed 1300 or 900, so I went with the larger number to depress the results. If the trial rate is actually 268 out of 900 instead of 1300, then the projected number of trials jumps to 744, with 714 involving bad testimony instead of 494.

 

[George Jefferson Airplane]

I'm sorry you don't seem to understand how math works.

Let's start with what we know for sure: We have 2500 cases in which federal forensic examiners declared a hair match. After a review of ~ 1300*, we have a known fail rate of 90%. That's ~2250 cases in which forensic examiners found conclusive matches when, in fact, there was no scientific basis to make that assertion. We know that in the only jurisdiction in which they have concluded investigations into all convictions using the false testimony (D.C.), 5 of 7 convicts were exonerated. That's a whopping ~71.43%. We know the fail rate for trials is ~95.9%.

We don't know how many cases went to trial out of that 2500 sample, but the article at least implies that there were 268 actual trials out of that ~1300* investigated thus far. That's a rate of ~20.62%, or a total of 515 trials total. Applying our known fail rate for trials tells us that we have 494 trials in which the federal forensic examiners provided substantively false testimony to convict the accused.

Applying the DC exoneration rate (71.43%) to those 494 trials involving false testimony to convict gives us ~353.

And then you have to drill down to the 500-1000 state forensic examiners who were taught by those federal guys "to testify the same way". How many trials/conviction do you think those guys have testified in? We have 28 federal examiners giving us ~515 trials in which they provided false testimony. Say they just trained 500 state examiners, and those 500 state examiners only had a fail rate of half of their trainers, and those state examiners testified as often as their federal counterparts (all assumptions that understate reality: there is no reason to believe that only 500 were trained, no reason to believe they would be twice as good as their federal trainers, and state examiners testify far more frequently than their federal counterparts).

That's ~9196 trials with testimony of hair matches from state examiners trained in the faulty (at best) or intentionally deceptive (at worst) techniques of the "elite" federal guys. Halve the federal fail rate to get a 47.95% fail rate for the inferior state guys. That's 4409 state trials in which bad testimony was used to convict.

Now, pick whatever rate you want for exoneration. The small sample we have says 5/7, or ~71.43%. Say that sample overstates the state rate by a factor of 5, resulting in an ~14.29% exoneration rate. That's 630 innocent folks convicted.

630 + 352 = 982. And that's underestimating what the data tells us at every step for which we don't have reliable data.

So, yeah, "hundreds" is almost a statistical certainty.

*And this is also likely understating the percentage. The article isn't clear about how many have been reviewed. It says that 2500 were identified, 1200 are yet to be reviewed, and 900 lab reports "are nearly compete, with about 1200 cases remaining". The implication is that they have either reviewed 1300 or 900, so I went with the larger number to depress the results. If the trial rate is actually 268 out of 900 instead of 1300, then the projected number of trials jumps to 744, with 714 involving bad testimony instead of 494.

Innocent<>falsely convicted

 

[Tennessee_ATO]

I'm sorry you don't seem to understand how math works. Let's start with what we know for sure: We have 2500 cases in which federal forensic examiners declared a hair match. After a review of ~ 1300*, we have a known fail rate of 90%. That's ~2250 cases in which forensic examiners found conclusive matches when, in fact, there was no scientific basis to make that assertion. We know that in the only jurisdiction in which they have concluded investigations into all convictions using the false testimony (D.C.), 5 of 7 convicts were exonerated. That's a whopping ~71.43%. We know the fail rate for trials is ~95.9%.

We don't know how many cases went to trial out of that 2500 sample, but the article at least implies that there were 268 actual trials out of that ~1300* investigated thus far. That's a rate of ~20.62%, or a total of 515 trials total. Applying our known fail rate for trials tells us that we have 494 trials in which the federal forensic examiners provided substantively false testimony to convict the accused.

Applying the DC exoneration rate (71.43%) to those 494 trials involving false testimony to convict gives us ~353.

And then you have to drill down to the 500-1000 state forensic examiners who were taught by those federal guys "to testify the same way". How many trials/conviction do you think those guys have testified in? We have 28 federal examiners giving us ~515 trials in which they provided false testimony. Say they just trained 500 state examiners, and those 500 state examiners only had a fail rate of half of their trainers, and those state examiners testified as often as their federal counterparts (all assumptions that understate reality: there is no reason to believe that only 500 were trained, no reason to believe they would be twice as good as their federal trainers, and state examiners testify far more frequently than their federal counterparts).

That's ~9196 trials with testimony of hair matches from state examiners trained in the faulty (at best) or intentionally deceptive (at worst) techniques of the "elite" federal guys. Halve the federal fail rate to get a 47.95% fail rate for the inferior state guys. That's 4409 state trials in which bad testimony was used to convict.

Now, pick whatever rate you want for exoneration. The small sample we have says 5/7, or ~71.43%. Say that sample overstates the state rate by a factor of 5, resulting in an ~14.29% exoneration rate. That's 630 innocent folks convicted.

630 + 352 = 982. And that's underestimating what the data tells us at every step for which we don't have reliable data.

So, yeah, "hundreds" is almost a statistical certainty.

*And this is also likely understating the percentage. The article isn't clear about how many have been reviewed. It says that 2500 were identified, 1200 are yet to be reviewed, and 900 lab reports "are nearly compete, with about 1200 cases remaining". The implication is that they have either reviewed 1300 or 900, so I went with the larger number to depress the results. If the trial rate is actually 268 out of 900 instead of 1300, then the projected number of trials jumps to 744, with 714 involving bad testimony instead of 494.

Innocent<>falsely convicted

Exonerated means "freed from guilt or blame". So if want to argue that exoneration (the standard I applied) doesn't mean "innocent", have at it.

 

[George Jefferson Airplane]

I'm sorry you don't seem to understand how math works. Let's start with what we know for sure: We have 2500 cases in which federal forensic examiners declared a hair match. After a review of ~ 1300*, we have a known fail rate of 90%. That's ~2250 cases in which forensic examiners found conclusive matches when, in fact, there was no scientific basis to make that assertion. We know that in the only jurisdiction in which they have concluded investigations into all convictions using the false testimony (D.C.), 5 of 7 convicts were exonerated. That's a whopping ~71.43%. We know the fail rate for trials is ~95.9%.

We don't know how many cases went to trial out of that 2500 sample, but the article at least implies that there were 268 actual trials out of that ~1300* investigated thus far. That's a rate of ~20.62%, or a total of 515 trials total. Applying our known fail rate for trials tells us that we have 494 trials in which the federal forensic examiners provided substantively false testimony to convict the accused.

Applying the DC exoneration rate (71.43%) to those 494 trials involving false testimony to convict gives us ~353.

And then you have to drill down to the 500-1000 state forensic examiners who were taught by those federal guys "to testify the same way". How many trials/conviction do you think those guys have testified in? We have 28 federal examiners giving us ~515 trials in which they provided false testimony. Say they just trained 500 state examiners, and those 500 state examiners only had a fail rate of half of their trainers, and those state examiners testified as often as their federal counterparts (all assumptions that understate reality: there is no reason to believe that only 500 were trained, no reason to believe they would be twice as good as their federal trainers, and state examiners testify far more frequently than their federal counterparts).

That's ~9196 trials with testimony of hair matches from state examiners trained in the faulty (at best) or intentionally deceptive (at worst) techniques of the "elite" federal guys. Halve the federal fail rate to get a 47.95% fail rate for the inferior state guys. That's 4409 state trials in which bad testimony was used to convict.

Now, pick whatever rate you want for exoneration. The small sample we have says 5/7, or ~71.43%. Say that sample overstates the state rate by a factor of 5, resulting in an ~14.29% exoneration rate. That's 630 innocent folks convicted.

630 + 352 = 982. And that's underestimating what the data tells us at every step for which we don't have reliable data.

So, yeah, "hundreds" is almost a statistical certainty.

*And this is also likely understating the percentage. The article isn't clear about how many have been reviewed. It says that 2500 were identified, 1200 are yet to be reviewed, and 900 lab reports "are nearly compete, with about 1200 cases remaining". The implication is that they have either reviewed 1300 or 900, so I went with the larger number to depress the results. If the trial rate is actually 268 out of 900 instead of 1300, then the projected number of trials jumps to 744, with 714 involving bad testimony instead of 494.

Innocent<>falsely convicted

Exonerated means "freed from guilt or blame". So if want to argue that exoneration (the standard I applied) doesn't mean "innocent", have at it.

You're the one throwing around the word "innocent."

There are hundreds if not thousands of variables in criminal cases, they might have to exclude the hair match but that doesn't mean the murder weapon and an eye witness weren't also present. Using some statistical formula to try to predict the outcomes of all the cases seems like wasted energy IMO, it's really going to be a case by case evaluation.

 

[Tennessee_ATO]

I'm sorry you don't seem to understand how math works. Let's start with what we know for sure: We have 2500 cases in which federal forensic examiners declared a hair match. After a review of ~ 1300*, we have a known fail rate of 90%. That's ~2250 cases in which forensic examiners found conclusive matches when, in fact, there was no scientific basis to make that assertion. We know that in the only jurisdiction in which they have concluded investigations into all convictions using the false testimony (D.C.), 5 of 7 convicts were exonerated. That's a whopping ~71.43%. We know the fail rate for trials is ~95.9%.

We don't know how many cases went to trial out of that 2500 sample, but the article at least implies that there were 268 actual trials out of that ~1300* investigated thus far. That's a rate of ~20.62%, or a total of 515 trials total. Applying our known fail rate for trials tells us that we have 494 trials in which the federal forensic examiners provided substantively false testimony to convict the accused.

Applying the DC exoneration rate (71.43%) to those 494 trials involving false testimony to convict gives us ~353.

And then you have to drill down to the 500-1000 state forensic examiners who were taught by those federal guys "to testify the same way". How many trials/conviction do you think those guys have testified in? We have 28 federal examiners giving us ~515 trials in which they provided false testimony. Say they just trained 500 state examiners, and those 500 state examiners only had a fail rate of half of their trainers, and those state examiners testified as often as their federal counterparts (all assumptions that understate reality: there is no reason to believe that only 500 were trained, no reason to believe they would be twice as good as their federal trainers, and state examiners testify far more frequently than their federal counterparts).

That's ~9196 trials with testimony of hair matches from state examiners trained in the faulty (at best) or intentionally deceptive (at worst) techniques of the "elite" federal guys. Halve the federal fail rate to get a 47.95% fail rate for the inferior state guys. That's 4409 state trials in which bad testimony was used to convict.

Now, pick whatever rate you want for exoneration. The small sample we have says 5/7, or ~71.43%. Say that sample overstates the state rate by a factor of 5, resulting in an ~14.29% exoneration rate. That's 630 innocent folks convicted.

630 + 352 = 982. And that's underestimating what the data tells us at every step for which we don't have reliable data.

So, yeah, "hundreds" is almost a statistical certainty.

*And this is also likely understating the percentage. The article isn't clear about how many have been reviewed. It says that 2500 were identified, 1200 are yet to be reviewed, and 900 lab reports "are nearly compete, with about 1200 cases remaining". The implication is that they have either reviewed 1300 or 900, so I went with the larger number to depress the results. If the trial rate is actually 268 out of 900 instead of 1300, then the projected number of trials jumps to 744, with 714 involving bad testimony instead of 494.

Innocent<>falsely convicted

Exonerated means "freed from guilt or blame". So if want to argue that exoneration (the standard I applied) doesn't mean "innocent", have at it.

You're the one throwing around the word "innocent."

There are hundreds if not thousands of variables in criminal cases, they might have to exclude the hair match but that doesn't mean the murder weapon and an eye witness weren't also present. Using some statistical formula to try to predict the outcomes of all the cases seems like wasted energy IMO, it's really going to be a case by case evaluation.

Absolutely, because if someone is exonerated of a crime they are ... wait for it ... "innocent".

Innocent is a term that has a specific meaning in a criminal context. It doesn't mean "free from sin", it means:

An absence of facts required for conviction under a criminal statute.

So, if there's is a lack of evidence for conviction, that person is "innocent", quite literally by definition.

The old standard is "innocent until proven guilty" for a reason. We all start out as innocent and are only guilty if convicted.

As far as criticizing my statistical analysis (such as it was), go back to what started this. I commented that "hundreds" of innocent people being convicted is almost a statistical certainty. Jonesed fired off a little ROTFLMAO emoticon in response. I simply showed a very quick explanation for why my statement is accurate.

Sure, every case will stand on its own. But to pretend that we aren't talking about many thousands of trials in which false evidence was used to convict is naive. And to pretend that without that e"objective" and "conclusive" evidenced wasn't what the jury relied on in a huge percentage of those cases is equally naive. Remember, hair analysis was the DNA testing of the 90's. It was unassailable, 100% iron-clad proof. At least that's what juries were told. Turns out not only was that not the case, but there was an added layer of active deceit sprinkled on top. We know for a fact that 5/7ths of convictions in which the false testimony was used in DC resulted in the wrongful conviction of innocent people. Do we really think that there won't be significant percentages of wrongfully convicted people outside DC?

And do we really think the 700 files police and prosecutors aren't turning over to the feds to be examined really represent cases in which there were iron-clad convictions without the false hair matching testimony?

 

[peaces]

Whoops

http://news.yahoo.com/man-served-34-years-murder-charges-dismissed-220944135.html