DNA Doesn’t Lie, But People Do

Em Carpenter

Em was one of those argumentative children who was sarcastically encouraged to become a lawyer, so she did. She is a proud life-long West Virginian, and, paradoxically, a liberal. In addition to writing about society, politics and culture, she enjoys cooking, podcasts, reading, and pretending to be a runner. She will correct your grammar. You can find her on Twitter.

Related Post Roulette

13 Responses

  1. Chip Daniels

    One thing not commonly discussed is how convicting an innocent person allows an actual killer to walk free to kill again and again and again.Report

  2. Dark Matter

    So the guy who helped OJ get off was willing to say anything about anyone if we paid him?

    I guess this explains a bit of what happened there. With the benefit of hindsight, the DNA evidence against OJ was stupidly large and convincing.Report

  3. MarineVet

    While generally scientific, DNA is still an interpretive measure and there is no real science that backs up their claims of uniqueness. They never do a 100% fully DNA sequencing, they hit a few spots and call it a day. It might be a fact that DNA is varied enough to be 1 in 2 billion or it might be 1 in 10^200 with more combinations than there are electrons in the universe. But the fact is that they are not testing DNA far enough to make that quantification scientifically.Report

    • Dark Matter in reply to MarineVet

      They never do a 100% fully DNA sequencing, they hit a few spots and call it a day.

      True, but 99.9% (wiki) of all DNA is the same for all humans so there’s no point.

      As of April 2021 we have 20+ million people in these profiling DNA databases. Ergo we have an extremely good idea how unique these entries are.

      It would be major news if we had non-twin non-clone identical entries. It hasn’t happened yet and we think it will never happen.

      It’s like using monkeys to randomly create 10,000 page books. If you find two books that are the same, odds are good they’re different copies of the same monkeys work and not two different monkeys.Report

  4. DavidTC

    I think we need to seriously consider implementing double-blind testing of evidence.

    Put a third party between the police and the testing lab. The third party gets handed a piece of evidence by the police, it then hands four pieces of evidence to the testing lab, and the testing lab doesn’t know which one is real. They have to test them all.

    Now, that obviously would be expensive, but, even if we don’t do that, I want to point out that we almost never bother to test these labs when they don’t know it’s a test, and as such a third party in the middle would be a good idea, because we could do that.

    Right now, they get handled a piece of evidence by the police and they generally know what the results ‘should’ be. Whereas if there was someone in the middle, someone with no interest in this, we would be able to _occassionally_ hand the lab completely bogus evidence and see if they came back with positive results. We wouldn’t have to do it for everything, but if the lab knew that ‘try to match this fingerprint to a suspect’ or ‘test this for blood’ tests were sometimes completely made up and if they return a positive they’re immediately going to be fired, we might have less of this.

    Granted, this is a problem with the entire criminal justice system, not just the labs. Everyone can lie, and there’s basically nothing in the system designed to check that or test it. It’s just a lot harder to prove the police were blatantly lying about something they observed than to retest a piece of evidence. Although it would be interesting to test the police in the same way, set up situations where someone is factually not breaking the law, and monitored, and wait for the police to react.Report

    • Dark Matter in reply to DavidTC

      Interesting Idea. It wouldn’t need to be three-and-1 (25% of tests matter), you could probably do 1-4 (80% of tests matter) and you’d be just fine. That would reduce the expense a LOT. In theory you could go with 1-0 if you’re not telling the lab what the correct results are but whatever.Report

      • DavidTC in reply to Dark Matter

        I mean, honestly, literally just inserting a third party in there so that the police are not directly talking to the labs in such a way that they can make it obvious what outcomes they want would solve a good chunk of the problem. But I just suspect that the police would instantly come up with a backchannel and subvert all that.

        Which, to be clear, the police would do anyway, which means any testing would have to be pretty identical. They want a towel tested for DNA, the middle party needs to go and buy an identical towel and put identical bloodstains on it from someone else, and send that in…and I was going to say ‘send it in together’, but that _also_ puts the testing lab on alert. They really need to send it in such a way that the lab will mistake it for the original, and only after it’s tested send the actual evidence in.

        The problem there, of course, is that it can slow things down. OTOH, we don’t seem to have a problem with the court system operating extremely slowly in literally every other possible way, so it seems odd we’d suddenly care about that.

        Of course, the thing about DNA evidence specifically, is instead of duplicating the tests, we can simply duplicate the things being matched to. Don’t let labs run tests against one DNA pattern, make them run them against a 100 different ones, some amount of which would be supplied by the police and the rest would be added in by the third party. Aka, do a ‘lineup’.

        Same with fingerprints.

        Which is, I think, how a lot of people assume it works, but actually isn’t. The only place any court requires lineups is random civilian witnesses. _Expert_ witnesses get to stand there and tell you how they compared the fingerprints/DNA at the crime scene to the fingerprints/DNA of the suspect, which they knew was the suspect at the time, and and they match. And I think people do not understand the level of subjectivity in that, they think it’s some magical computer stuff where it pops up MATCH IDENTIFIED and a photo of the suspect that the computer found out of a millions of records, and it really really is not.Report

        • Dark Matter in reply to DavidTC

          RE: Towels
          Separate labs? One to collect the DNA, one to process it.

          RE: 100 tests
          If they know they’re getting random spot checks, enough to establish their level of honesty, then they’ll toe the line.

          RE: Expert Witnesses
          I got to hear one at my trial (I was in the jury). She was an expert at running blood alcohol tests. Seriously impressive lady.(*)

          Given that the cops really didn’t want to convict him (they paid his fine(**) and let him drive off) my expectation is they didn’t tell her what the results should be.

          (*) Attractive, Intelligent, Well Spoken, Experienced, and a Degree beyond a PhD is running that test. The cops beclowned themselves but she effortlessly handled all questions from the defense with an aura of competence.

          (**) Yes, really. College Campus cops dealing with a no-neck football player who I gather was important.Report

Leave a Reply

Your email address will not be published. Required fields are marked *