Author Topic: Public DNA database catches Golden State Killer  (Read 1432 times)

Mackin USA

  • Inner Core
  • Hero Member
  • *
  • Posts: 2448
  • Abstract Artist
    • View Profile
Re: Public DNA database catches Golden State Killer
« Reply #15 on: February 04, 2019, 08:55:40 PM »
and James Clapper lied to Congress about it and he is not in jail YET
Mr. Mackin

rcjordan

  • I'm consulting the authorities on the subject
  • Global Moderator
  • Hero Member
  • *****
  • Posts: 7995
  • Debbie says...
    • View Profile
Re: Public DNA database catches Golden State Killer
« Reply #16 on: February 04, 2019, 09:22:24 PM »
This has been in my feeds for a couple of days.

Quote
But unbeknown to its users, the Houston-based firm quietly and voluntarily agreed in 2018 to open its database of more than a million records to the F.B.I. and examine DNA samples in its laboratory to identify suspects and victims of unsolved rapes and murders.

FamilyTreeDNA Admits to Sharing Genetic Data With F.B.I. - The New York Times
https://www.nytimes.com/2019/02/04/business/family-tree-dna-fbi.html

rcjordan

  • I'm consulting the authorities on the subject
  • Global Moderator
  • Hero Member
  • *****
  • Posts: 7995
  • Debbie says...
    • View Profile
Re: Public DNA database catches Golden State Killer
« Reply #17 on: February 05, 2019, 05:29:59 PM »

rcjordan

  • I'm consulting the authorities on the subject
  • Global Moderator
  • Hero Member
  • *****
  • Posts: 7995
  • Debbie says...
    • View Profile

ergophobe

  • Inner Core
  • Hero Member
  • *
  • Posts: 4653
    • View Profile
Re: Public DNA database catches Golden State Killer
« Reply #19 on: February 15, 2019, 01:37:58 AM »
Quote
“There is no known relationship between the individuals, and that’s what makes these sort of cases so hard to solve,” Minneapolis police spokesman John Elder told WCCO.

There is a fundamental problem here. I recommend

Calculated Risks: How to Know When Numbers Deceive You, by Gerd Gigerenzer

https://www.amazon.com/Calculated-Risks-Know-Numbers-Deceive/dp/0743205561/ref=as_li_ss_tl?crid=2JI0IB5C0B28V&keywords=calculated+risks+how+to+know+when+numbers+deceive+you&qid=1550194344&s=books&sprefix=calculated+ris,stripbooks,152&sr=1-1-fkmrnull&linkCode=ll1&tag=ultraskiercom-20&linkId=baf225ec2df4e5ab84ff369532db07cf&language=en_US

He talks about breast cancer, AIDS diagnoses and conviction by DNA evidence.

The short version is this.

They will say things like "Based on the DNA, there is a 99.99% chance that it is from the same person." So in other words, if you test 10,000 people, you will get one false positive.

So in a city with 2,000,000 people, there are 200 false positives walking around the city. So if I *randomly* find someone ("no known relationship"), my chance that I have the right guy is not 99.99% as juries are led to believe. In the absence of any other evidence, the chance is actually 0.5% (one in 200). People's lives have been completely ruined because most people do not understand the logic there.

The book has similar analyses for other things. Very interesting and very troubling because most lawyers, judges, juries, testing labs, doctors and cops do not understand this simple statistical problem.

Travoli

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 726
    • View Profile
Re: Public DNA database catches Golden State Killer
« Reply #20 on: February 15, 2019, 04:52:33 PM »
>there are 200 false positives walking around the city
>the chance is actually 0.5%

That's a really good point.

ergophobe

  • Inner Core
  • Hero Member
  • *
  • Posts: 4653
    • View Profile
Re: Public DNA database catches Golden State Killer
« Reply #21 on: February 15, 2019, 09:24:51 PM »
Yeah, that book gave me a few simple tools like that to evaluate a lot of things.

Another example... This test reduces the risk of dying by this type of cancer by 50% and only costs $1,000 ever five years. Clearly it's worth it right?

Well, the risk of dying from that type of cancers is 3 per 100,000. By reducing the risk 50%, you've brought that down to 2 and saved one life per 100,000 at a cost of $100,000,000 every five years.

If you spent that money in another way, could you save, OMG, TWO lives for $100,000,000? I bet you could.

What does this have in common with the DNA case? In both cases, people are lead astray by looking at percentages. They should be looking at absolute numbers.

I know this is OT, but one last example.

People are told that the test for HIV is 100% accurate because it's actually two tests - a gram negative and a gram positive. If it says you have HIV, it means you have it. Doctors tell people this all the time.

That's not what it means. It means that if you have HIV, the test is basically guaranteed to detect it. If it says you're clean, you can stop worrying.

But it actually has a 1 in 1,000 error rate of false positives. The rate of HIV in the monogamous heterosexual male population is 1 in 10,000. So if you're a monogamous heterosexual male and you test positive for HIV, there isn't a 99.999999% chance you have it. There is a 10% chance.

The author of the book tells the story of a man who lost his job, house and wife only to find out three years later that he didn't have HIV. He has stories of people convicted on DNA evidence just because they were a random match only to have someone else, who also matched the DNA, get caught for the crime later.

BoL

  • Inner Core
  • Hero Member
  • *
  • Posts: 825
    • View Profile
Re: Public DNA database catches Golden State Killer
« Reply #22 on: February 15, 2019, 10:20:00 PM »
At the risk of sounding flippant, is the 99.99% stat a rounded number? After all, our genetic make up is what makes us unique. I don't know the specifics of how they use genetic evidence in courts in any particular country, I vaguely remember something about there being the potential for ~420 billion unique human beings.

I feel we could run into problems if a subset of those possibilities invariably end up in jail...