Slashdot Mirror
FBI Doesn't Tell Courts About Bogus Evidence
dprovine writes "According to a joint investigation by The Washington Post and 60 Minutes, a forensic test used by the FBI for decades is known to be invalid. The National Academy of Science issued a report in 2004 that FBI investigators had given "problematic" testimony to juries. The FBI later stopped using "bullet lead analysis", but sent a letter to law enforcement officials saying that they still fully supported the science behind it. Hundreds of criminal
defendants — some already convicted in part on the testimony of FBI experts — were not informed about the problems with the evidence used against them in court."
why the evidence presented was not questioned as to its scientific veracity? From what I read and hear, there is valid reason for questioning every technology that is used against you, whether that is a radar gun, camera, eye witness, anything. Your lawyer should never accept anything as fact, and should attempt to prove in court that it is not valid. For instance, the breathalyzer tests have been shown to read higher than actual in more than 25% of tests. Can't remember the story lead or I'd link it.
It's things like this that ensure that the party with the most money will win...
Support NYCountryLawyer RIAA vs People
Lie detectors can be used in investigations but not in trials. There's a real, significant difference there.
The FBI also says it will begin monitoring the testimony of all lab experts to make sure it is based on sound scientific principles. FBI Assistant Director John Miller said, "We are going to the entire distance to see that justice is now served."
Evidence Of Injustice: FBI's Bullet Lead Analysis Used Flawed Science To Convict Hundreds Of Defendants
Chicken fried butter sticks? Do
Lead test; inconclusive article describes why; ballistics test; in spite of crime TV for 25 years: inconclusive; and in smaller calibers: unusable. Oh and to work at all the bullet cannot be at all deformed which happens if the bullet contacts a bone, wall or anything hard. Paraffin test for shooting a gun: inconclusive; you cannot tell the difference between shooting a gun, lighting a firework, shoveling manure, or shaking hand with a politician and since washing hands after any of these is a natural reaction, it will not work. Most "Scientific" crime tests rely more on the unlettered faith of the jurors in this heap big magic called "Science" then in the science itself...
- Minutus cantorum, minutus balorum, minutus carborata descendum pantorum.
"There are more things in heaven and earth, Horatio, than are dreamt of in your philosophy."
60 Minutes interviewed one person who was convicted with bullet lead analysis as the only physical evidence. All other evidence was circumstantial. No one knows yet how many cases used this as evidence, so there's no way to know how many cases relied heavily on it.
Developers: We can use your help.
I seem to recall reading that there is no scientific study validating the uniqueness of fingerprints.
Science doesn't prove, it disproves. You test your theory, and if it's not found to be invalid it's assumed to be valid, until further or different tests disprove it. If on the unlikley event the second law of thermodynamics is shown to be false, or false under some circumstances, it will be dropped or modified.
When it is shown that two people can, indeed, have identical fingerprints then fingerprints can no longer honestly be used as evidence.
mcgrew's razor: Never attribute to stupidity that which can be explained by greedy self-interest
Indeed. This isn't the first time William Tobin has blown the whistle by any stretch of the imagination. Tobin, who was formerly the chief metallurgist at the FBI, also called into question the investigative practices of individuals involved in the FBI's investigation of TWA Flight 800 and has testified as to various fraudulent activities that took place within various FBI investigations. Apparently this test was the same test used to convict Lee Harvey Oswald, and guess who called it into question? Yup. William A. Tobin.
Curiouser and curiouser, no?
My blog
There is no physical law that dictates that fingerprints are unique, but there is an enormous amount of statistical evidence to that effect. Of course, this is dependent on the fact that you've got a good print for comparison. Smudged prints, partial prints, etc, all significantly weaken the statistical case for fingerprint evidence.
*sigh* back to work...
Perhaps, but they certainly had an incentive *not* to check too closely into the science either. First, science is not their area of expertise which brings up the second point. Why question potentially valuable scientific evidence that helps your (the prosecution's) case if the scientists say that it is good or at least do not say that it has any problems? As other posters have said, the penalty for getting your conviction overturned someday is small and unlikely while a failure to secure a conviction could potentially have more short term career consequences for an up and coming prosecutor.
Oswald was convicted?! That must have been a speedy trial.
There is a lot of research in both forensics and statistics about fingerprint uniqueness. A classic reference for calculating these numbers is Stoney and Thornton - Stoney, David A., and John I. Thornton. 1986. A critical analysis of quantitative fingerprint individuality models. Journal of Forensic Sciences 31 (4): 1187-1216.
Fingerprint identification is done by comparing the location and orientation of "minutiae," small defects in a fingerprint pattern. Typically, it takes around 12 minutiae to be considered a good match (by the FBI). Now, IF ALL OF THESE MATCH, the probability of misidentification is small (10^-8 or so assuming perfect procedure, I don't remember actual numbers, but it's a small, small chance). However, it's even money that the FBI has two fingerprints that match up to six minutiae.
There have been some high-profile misidentifications (take this case) but these seem to come from a couple of errors:
1) true screwups in the lab (only double-checking and real responsibility will help)
2) people believing that six or fewer minutiae matching is conclusive (education and honesty needed)
Incidentally, a somewhat convincing anecdote is that identical twins have correlated*, but clearly distinct fingerprints!
*i.e. if one has a whorl-pattern, the other is more likely than average to also have a whorl - this doesn't mean the minutiae are correlated, I don't know about that.
Given the context of your post (which I did not quote) I realize this statement was probably tongue in cheek, but I felt the need to respond anyway.
CCTV systems are used to investigate crimes. But they rarely actually catch the criminal act themselves. Instead it's used to connect people, places, and times. Would you like to be linked to criminal activity just because you happened to be in two wrong places at the wrong times? If a linked crime happens in 2 areas, the probability goes up that all the people recorded in that area are suspects. This also increases your probability to be charged with a crime you did not convict. Even if you are acquitted in the end, the mere charge of a crime in today's societies comes with negative consequences. Even if you are proven innocent, you suffer reputation damages, probably wife and kids, and most assuredly your job.
Perish the thought that protecting the very processes which are there to protect your fundamental freedoms would ever triumph over the ease of emotion laden controversy.
The scientific report:
Forensic Analysis: Weighing Bullet Lead Evidence, The National Academies Press, 2004.
URL: http://www.nap.edu/openbook.php?isbn=0309090792
An extract from "Appendix K: Statistical Analysis of Bullet Lead Data by Karen Kafadar and Clifford Spiegelman (169-214)" follows.
Section 3.1 FBI Calculation of False-Positive Probability (FPP):
"The FBI reported an apparent FPP that was based on the 1,837-bullet data set (Ref. 11). The authors repeated the method on which the FBI's estimate was based as follows. [...] The FBI summarized the results by claiming an apparent FPP of 693/1,686,366, or 1 in 2,433.4 ('about 1 in 2,500'). ***That estimated FPP is probably too small, in as much as this 1,837-bullet data set is not a random sample of any population and may well contain bullets that tend to be further apart than one would expect in a random sample of bullets.***"
Section 3.2 Simulating False-Positive Probability:
"We simulate the probability that the 2-SD interval (or range interval) for one bullet's concentration of one element overlaps with the 2-SD interval (or range interval) for another bullet's concentration of that element. The simulation is described below. [...] Thus, the FPP could be estimated here as roughly 47/91, or 0.516. [...] Because homogeneous batches of lead, manufactured at different times, could by chance have the same chemical concentrations (within measurement error), the actual FPP could be even higher."
Section 4.2 Individual Equivalence t Tests:
"[...] Probabilities such as the FBI's claim of '1 in 2,500' are inappropriate when based on a data set such as the 1,837-bullet data set; as noted in Section 3.2, it is not a random collection of bullets from the population of all bullets, or even from the complete 71,000+ bullet data set from which it was extracted."