Saturday, May 30, 2015

The Pyrite Standard

Here's the deal.

We know that most of what passes for forensic "science," the CSI stuff that gets people put in prison forever, or executed, is bullshit.

  • Bite mark comparison - bullshit
  • Tire track comparison - bullshit
  • Hair and fiber matching - bullshit
  • Ballistics - bullshit
  • Handwriting comparison - bullshit
  • Fingerprints, yep fucking fingerprints - bullshit

We know that dog alerts mean nothing.
We know that people confess to crimes they didn't commit.
We know that lab techs fake results to get the results the cops and prosecutors want them to reach.
We know that eyewitness identification is altogether unreliable.

Really, the miracle is that we ever get it right, that sometimes the guy they say did it really did.

Fortunately, we have DNA.  You know, the gold standard.  

When the numbers come back and the guy in the white lab coat explains to the jury about alleles and genetic markers and says that there's a one in 623 septillion chance that the bit of whatever found in some damning place came from someone other than the defendant - oh, and by the way, there aren't 623 septillion molecules of water in all the oceans, lakes, rivers, and aquifers combined . . . .

Whew.  Now we know we nailed the bastard.

Ah, that was so 2014.

Turns out the gold standard is fool's gold.

Spencer S. Hsu in the Washington Post.
The FBI has notified crime labs across the country that it has discovered errors in data used by forensic scientists in thousands of cases to calculate the chances that DNA found at a crime scene matches a particular person, several people familiar with the issue said.
Of course, it's minor.  Just a blip.  
In a bulletin sent to crime labs, the FBI said the problem stemmed from “clerical mistakes in transcriptions of the genotypes and to limitations of the old technology and software.”
You know, some secretary.  Probably a blonde.  And, really, it was microsoft's fault.  A software glitch.  

Oh, sure, some pointy-pointy headed professor in Dayton, Ohio's been pointing out this kind of thing for a decade.
“The public puts so much faith in DNA testing that it makes it especially important to make those the best estimates possible,” said Wright State University statistics professor Daniel R. Krane, an expert whose work has been cited by defense attorneys. “There is no excuse for a systematic error to many thousands of calculations in such a context.”
Krane, who identified errors 10 years ago in the DNA profiles the FBI analyzed to generate the population statistics data, called the consequences of the disclosure appalling, saying the data has been used in tens of thousands or hundreds of thousands of cases worldwide in the past 15 years. He said when he flagged the problems a decade ago, the FBI downplayed his findings.
But, hey.  What does he know.  If he were any good he'd be at Harvard.  Or working for the Bureau. I mean, it's not like anyone was wrongly convicted because, well, because.
The bureau has said it believes the errors, which extend to 1999, are unlikely to result in dramatic changes that would affect cases.
In Ohio (and probably your state, too, they've been doing DNA testing on thousands of old rape kits that have been sitting, ignored, in evidence lockers in police departments around the state.  They're bringing 20 year old rape charges against people based on DNA - even when the knew the person's identity at the time, even when the person admitted back then that he'd had sex with the accuser. Because DNA.

And the legislature, our legislature, is preparing to enact a new law that would restart the statute of limitations whenever DNA.  Because DNA.

Which, it turns out, is - what's the term?  Oh yeah.



  1. Doesn't your post essentially amount to the proposition that all proof offered in court to prove a person committed crime is bullshit? Therefore, it appears to follow, no person can (should) be convicted of a crime. Is that really your position? I assume this is rhetorical hyperbole, but maybe not. I enjoy reading your blog but this post makes me shake my head wondering.

    1. I don't think so. At least I hope not. What I'm really arguing is that when we put blind faith in forensic so-called sciences, we're making a serious mistake.

      There's nothing inherently wrong with DNA technology. But the uncritical use of the databases to figure out what the scientific results mean, databases generated by the FBI for purposes of locking up bad guys, that's a different matter.

      Trials, in any event, and as I frequently say, aren't about truth. They're about proof. And proof is what the jury believes. It's important that they not be wowed with false declarations of infallibility and that evidence - all evidence - be treated with skepticism.

      And we should be, always, less certain of the results, especially when they point to guilt which is after all what gets (or doesn't get) proved rather than not-guilt which is really just a failure of proof, than we tend to be.

      I'm not a nihilist, but I am a skeptic.

  2. While I love reading the blog ... 5 gallons of water has more than 623 septillion water molecules (unless you're an old-fashioned Brit).