How close does a DNA "match" have to be to get somebody tossed in jail? Do they actually require 100%, or is a "match" something less? What's the false-positive rate? Would a non-twin sibling's DNA "match"?
Yes, I'm aware that, in theory, DNA is unique. As is the theory for finger-prints. But, I'm also aware that theory is rarely reality.
I don’t remember the exact details, because I have watched it long time ago, but they don’t need an exact match. Actually it’s more scary - if your distant cousin has DNA “in the system” - you are in it too.
This 1M DNAs may “help” match an order of magnitude more people.
I think you're thinking of genetic genealogy, where some of those DNA testing providers (like 23andme) submit their records to law enforcement, and they can see something like "this DNA sample is likely to be a brother of this existing sample", and can narrow down investigations based on that
According to wikipedia, "the theoretical risk of a coincidental match is 1 in 100 billion (100,000,000,000) although the practical risk is actually 1 in 1,000 because monozygotic twins are 0.2% of the human population", so it looks like some twins will cause an issue for DNA-based convictions, but I think a non-twin sibling is probably diverse enough? I have different eye colour, ears, food intolerances to my brothers, so I feel like our DNA is pretty different ("pretty" on a scale of human to human comparisons, since 99.9% of DNA is identical between humans or something like that).
I'd argue that while a DNA match could be a very good starting point for a conviction, other evidence should also be presented, for the reasons you suggest. For example, there is the case of the Phantom of Heillbron (https://en.wikipedia.org/wiki/Phantom_of_Heilbronn), where European police found DNA evidence of a woman being at many diverse crime scenes, but it turned out the cotton swabs used for testing were contaminated by at worker at the factory (The swabs weren't certified for police work)
That is a coincidental match, though! It's the risk of a false positive when you compare exactly two samples.
The problem is that we're proposing a DNA database, so you're now comparing 1 sample found at a crime scene with 300ish million people. That's a 300M in 100B chance, or 1 in 333, of some person having a false positive match.
Now combine that with thousands of samples being tested every day (because let's be honest, why wouldn't you use such a database for literally every single crime?) and you're looking at dozens of false-positives per day. It is only a matter of time before it isn't someone with an airtight alibi at the other side of the country, but someone reasonably-close who was just sitting alone on their couch at home all weekend. How are they going to prove they weren't involved in the crime?
This is why large-scale DNA matching is a horrible starting point for a conviction. It gives you a false sense of security, and it makes the investigators lazy. In most cases it is far better if DNA is used as the final nail in the coffin when you've already got a solid case against a suspect.
This is a really interesting article, even though it seems a little biased towards the mother (which, post hoc, if absolutely fair). Would be very cool to see / read this from the position of the judge as the thing unraveled.
The same process we've used for centuries? You got something better? I'm sure you don't. You are also ignoring that the 12 random schmoes are selected by both sides attorneys and given instructions and facts and can still be overruled by a judge (under circumstances) plus the possibility of appeal.
Exactly. I’ve served on two juries and the experiences left me shaking my head. In one case the judge’s instructions were dismissed by a majority of the jury members with a “yeah, but…”.
In reality, juries often hold evidence like this (and also-very-unreliable eye witness testimony) in high regard, based on how it's positioned into the narrative.
Comments
How close does a DNA "match" have to be to get somebody tossed in jail? Do they actually require 100%, or is a "match" something less? What's the false-positive rate? Would a non-twin sibling's DNA "match"?
Yes, I'm aware that, in theory, DNA is unique. As is the theory for finger-prints. But, I'm also aware that theory is rarely reality.
I think this video can answer your question: https://www.veritasium.com/videos/2021/9/30/your-dna-is-alre...
I don’t remember the exact details, because I have watched it long time ago, but they don’t need an exact match. Actually it’s more scary - if your distant cousin has DNA “in the system” - you are in it too.
This 1M DNAs may “help” match an order of magnitude more people.
I think you're thinking of genetic genealogy, where some of those DNA testing providers (like 23andme) submit their records to law enforcement, and they can see something like "this DNA sample is likely to be a brother of this existing sample", and can narrow down investigations based on that
According to wikipedia, "the theoretical risk of a coincidental match is 1 in 100 billion (100,000,000,000) although the practical risk is actually 1 in 1,000 because monozygotic twins are 0.2% of the human population", so it looks like some twins will cause an issue for DNA-based convictions, but I think a non-twin sibling is probably diverse enough? I have different eye colour, ears, food intolerances to my brothers, so I feel like our DNA is pretty different ("pretty" on a scale of human to human comparisons, since 99.9% of DNA is identical between humans or something like that).
I'd argue that while a DNA match could be a very good starting point for a conviction, other evidence should also be presented, for the reasons you suggest. For example, there is the case of the Phantom of Heillbron (https://en.wikipedia.org/wiki/Phantom_of_Heilbronn), where European police found DNA evidence of a woman being at many diverse crime scenes, but it turned out the cotton swabs used for testing were contaminated by at worker at the factory (The swabs weren't certified for police work)
That is a coincidental match, though! It's the risk of a false positive when you compare exactly two samples.
The problem is that we're proposing a DNA database, so you're now comparing 1 sample found at a crime scene with 300ish million people. That's a 300M in 100B chance, or 1 in 333, of some person having a false positive match.
Now combine that with thousands of samples being tested every day (because let's be honest, why wouldn't you use such a database for literally every single crime?) and you're looking at dozens of false-positives per day. It is only a matter of time before it isn't someone with an airtight alibi at the other side of the country, but someone reasonably-close who was just sitting alone on their couch at home all weekend. How are they going to prove they weren't involved in the crime?
This is why large-scale DNA matching is a horrible starting point for a conviction. It gives you a false sense of security, and it makes the investigators lazy. In most cases it is far better if DNA is used as the final nail in the coffin when you've already got a solid case against a suspect.
Sounds like perfect for rich people, they'll have the resources to argue in court.
Don't forget the reverse. If your DNA's an insufficient match to your children, they take your kids away.
https://womenherald.com/health/the-chimerism-case-how-lydia-...
This is a really interesting article, even though it seems a little biased towards the mother (which, post hoc, if absolutely fair). Would be very cool to see / read this from the position of the judge as the thing unraveled.
A woman's DNA match to a dead baby from before she was born led to the arrest of her grandmother: https://www.cnn.com/2024/12/12/us/baby-garnet-cold-case-mich...
Remember that one still needs proof. My DNA on a victim's clothes does not mean I committed the crime.
Are you willing to trust 12 random schmoes to make that call correctly? Especially in the face of an overzealous prosecutor?
The same process we've used for centuries? You got something better? I'm sure you don't. You are also ignoring that the 12 random schmoes are selected by both sides attorneys and given instructions and facts and can still be overruled by a judge (under circumstances) plus the possibility of appeal.
Your response is very uninformed at best.
The fake-it-'till-you-make-it crime lab "scientist" is worse: https://en.wikipedia.org/wiki/Joyce_Gilchrist
Exactly. I’ve served on two juries and the experiences left me shaking my head. In one case the judge’s instructions were dismissed by a majority of the jury members with a “yeah, but…”.
In principle.
In reality, juries often hold evidence like this (and also-very-unreliable eye witness testimony) in high regard, based on how it's positioned into the narrative.
In reality, juries often hold evidence like this in high regard, based on how it's positioned on CSI:Some-City television.