News for hackers. Stuff that matters.
Re:FBI overstated forensic hair matches in nearly all trials... by mikeash on Saturday April 18, 2015 @09:04PM (#9401674)
I'm constantly amazed at how terrible criminal forensic science seems to be. Everything except DNA matching appears to be no better than voodoo, and even the DNA stuff is tricky. Is there something about this particular field that attracts charlatans and incompetents? Is it purely a motivational problem, wherein the people who pay for these results are more interested in a conviction and don't care if it's right?
Re:FBI overstated forensic hair matches in nearly all trials... by MCRed on Saturday April 18, 2015 @10:15PM (#9401810)
Think about the hypothesis that all fingerprints are unique. That's voodoo. First off you would have to fingerprint everybody to check it. And secondly they don't actually match fingerprints, but match a description of the fingerprint that throws out most of the data (like taking a million books and reducing them all into one byte hashes of their text, and then claiming that Moby Dick is a plagiarized version of Henry V because they produce the same 8-bit hash!)

The problem is, these things were indicators, but they have been manipulated to be used as if they were proof.

If the suspect is a red head male, and the witness describes seeing a red head male, then the jury is inclined to think that's the same guy.

IT's human nature.

Even in New York City.

Where there are at least 84,000 red headed males!

Re:FBI overstated forensic hair matches in nearly all trials... by revelation on Saturday April 18, 2015 @09:56PM (#9401763)
DNA testing may be theorethically sound, but in practice it's restricted to testing about 13 different loci, which is a miniscule porton of the actual information at hand.

Starting from that weak position, the FBI then ignores basic statistics:

http://freakonomics.com/2008/08/19/are-the-fbis-probabilitie...

Re:FBI overstated forensic hair matches in nearly all trials... by fenomas on Saturday April 18, 2015 @09:32PM (#9401723)
Some time ago when I was reading about the accuracy of fingerprint analysis (similarly a horror show), the culprit seemed to be the professional association of analysts. IIRC their official line was that fingerprint analysis was 100% accurate(!), and they forbade members from taking part in research studies on the subject...
Re:FBI overstated forensic hair matches in nearly all trials... by scotty79 on Sunday April 19, 2015 @07:39AM (#9402794)
I think it's because it's a field that demands exact answer from fuzzy data. You can't do that unless you are willing to accept some wishful thinking as true.
Re:FBI overstated forensic hair matches in nearly all trials... by realusername on Sunday April 19, 2015 @05:04AM (#9402494)
I've worked a bit on fingerprint recognition devices (using various vendor SDKs) and people clearly overestimate the accuracy of these kind of devices. The majority can be fooled with a fingerprint made by melting gummy bear with some tape to gather the fingerprint. Creating fake fingerprints is really easy and cheap. I'm also not mentioning the crappy firmware these devices are generally equipped with, everything close to metal seems to have really low quality firmwares, most of the time, you need to compensate on your side for some unfixable problems in the firmware.

The main problem is TV I guess, for the average guy, these kind of technologies are completely infallible and people tend to forget that just like anything else, it's just an indicator that can be fooled / misinterpreted / badly gathered.

Re:FBI overstated forensic hair matches in nearly all trials... by gaius on Sunday April 19, 2015 @06:52AM (#9402720)
Yes it's a bit tinfoil-hat but when you look at shows like CSI and 24, you couldn't imagine a better propaganda campaign that "evidence is infallible and torture is OK".
Re:FBI overstated forensic hair matches in nearly all trials... by mml on Sunday April 19, 2015 @09:37AM (#9403078)
I've been of the opinion that shows of that nature are a pretty transparent attempt at gas lighting the populous into feeling as though such behaviors and practices are acceptable, desirable & necessary.

I'll go polish my tinfoil hat now.

Re:FBI overstated forensic hair matches in nearly all trials... by m_mueller on Sunday April 19, 2015 @11:07AM (#9403405)
It seems to me the 'tinfoil hat' thing has become the way of shaming people into not questioning the government, so now every time someone like you would like to say something critical about the government, you first have to excuse yourself from it. It almost seems like people have been brainwashed tinfoil hat on.
Re:FBI overstated forensic hair matches in nearly all trials... by PseudoRandomHN on Sunday April 19, 2015 @10:31AM (#9403247)
[replying with bogus account] I spent a few months myself working with a range of low to moderate cost OEM modules (<$200) from several manufacturers and their SDK's and I concur. Not all were easily tricked with a fake print but every device/SDK had some problem. We decided not to include fingerprint recognition as a primary means of authentication in our products. I've heard rumor of a competitor who did include it that actual field performance hasn't met expectations and they're retreating to use it as a secondary means.
Re:FBI overstated forensic hair matches in nearly all trials... by EliRivers on Sunday April 19, 2015 @06:49AM (#9402712)
It's well known that a large body of fire investigation knowledge is basically made up nonsense. On the plus side, the very fact that it is well known to be nonsense is effecting changes. There is a push towards actually having to use scientific methods, which is both gratifying and horrifying.
Re:FBI overstated forensic hair matches in nearly all trials... by thaumasiotes on Sunday April 19, 2015 @03:04AM (#9402289)
> Is there something about this particular field that attracts charlatans and incompetents?

Yes, a strong preexisting demand and no way to fulfill it honestly. (Really, the first is doing most of the work.)

Same reason you get charlatans and incompetents in love spells, medicine, and longevity.

Re:FBI overstated forensic hair matches in nearly all trials... by hawleyal on Monday April 20, 2015 @10:56AM (#9408490)
Abusing science for criminal justice. They're not necessarily incompetent. Just trying to apply science in a non-science related field is hard and somewhat futile.