FBI admits to giving flawed testimony for decades


I think this could delegitimize large swaths of forensic evidence. If I’m a defense attorney and the forensic evidence looks bad for my client. I’d l use the fact that 95% of the time, the witnesses lied. M


Christ what murdering assholes!


I have been trying to figure out a post that either conveys my gobsmacked-ness or at least be funny. But alas, this is all you get.


There have been stories like this before, about various forensic sciences not working. The FBI used to do some kind of bullshit metallurgical analysis of bullets, and I’m sure I’ve seen others and continued to do it after they knew it didn’t work.

But even when it’s not systemic, there are always individuals involved. Those outside canada may not know about Charles Smith, an “expert” in shaken baby syndrome whose testimony was used to convict over 200 people of killing children when the children had almost all died naturally or accidentally.

If you are a prosecutor, you really don’t want me on the jury. I just don’t think it even makes sense to trust what police or experts say more than I trust a coin toss that has a vested interested in coming up heads.


Does “over 95 percent” seem like a euphemism for 100%?


IIRC the whole science of “bite marks” is hooey. It can rule out some people, or if one has say missing teeth, indicate a pattern. But unlike finger prints, bite marks can’t be conclusively linked to a specific person in most cases. Best uses of it would be to rule people out.

IIRC on bullet forensics, that it too is said to be more of an exact science than it is. For sure things like the rifling pattern on a bullet can help narrow down which type of gun it came from, but if you had two identical guns from the same manufacturing lot, the rifling marks are going to be extremely similar and at some point you are making judgement calls and professional opinions that they aren’t the same vs having 100% proof.


Yeah, the evidence that fingerprints are unique isn’t even there. We know they are usually pretty different, but when databases get big you have to start thinking about actual uniqueness. They say the chance of a DNA sample randomly matching another person is less than one in a million, but they’ve got more than a million DNA samples in the database. So DNA is still pretty good, but if you get a DNA match for someone who doesn’t appear to have anything to do with the situation and you can’t find any links, it’s not unreasonable to think of randomness.

Of course, that’s ignoring human error. When someone says something has a one in a million chance of being wrong, I want to say, “So you’ve done this over a million times and no one ever made a mistake?”


I am a bit more comfortable with fingerprints, though as you said it isn’t conclusive they are unique, or that even small variations won’t be overlooked, or how prints can warp or look different based on how the print is put down.

Though I am sure you have heard about the CSI effect, where jurors now want rock solid scientific evidence, which we don’t always have. I think it is a cautionary tale that we shouldn’t convict on one piece of evidence or circumstantial evidence. But at the same time, something inconclusive shouldn’t rule out a conviction if several other things add up.


I’ve heard people talk about this, but conviction rates have gone from 85% in 1992 to 93% in 2012, so I don’t think it’s a meaningful hindrance on the justice system.


Hmm, good point. Or are they putting too much faith in forensics testimony, even though as this article shows, it isn’t always accurate?


Yeah, if there is a CSI effect then it must go both ways. On one hand, the juror wants absolute proof like they have on CSI. On the other hand, they juror thinks that forensic scientists are wizards and their word is infallible, like on CSI.

Anyway, all of that only matters in very high profile and complex cases. The majority of criminal cases are pretty simple, and the increased conviction rate probably reflects underfunded public defenders more than anything else.


I’ll just add this to my file.


Pfft, he’s got Evidence to Respond to. Parking correctly would have taken an extra 3 seconds, which in government work would have somehow cost the taxpayers an extra $20,000.


That’s because they use an outdated type of DNA test. For $99 (less during sales) anyone can get an autosomal DNA test from Ancestry or FamilyTreeDNA that absolutely will be unique. The state of the technology has moved well beyond what law enforcement uses.

closed #34

This topic was automatically closed after 5 days. New replies are no longer allowed.