Framed for Murder by His Own DNA

April 20, 2018 3:48 AM

When the DNA results came back, even Lukis Anderson thought he might have committed the murder. Traditional police work would have never steered police to Anderson. But the DNA hit led them to seek other evidence confirming his guilt.
posted by mattamatic (11 comments total)

I sort of wish there'd been more focus on the detective at the end. There's this mention at the end that he's learned to have less blind faith in DNA, but I also wonder whether he's learned not to try and bully information out of people. Would it have been possible to realise the guy who was the putative link to Anderson wasn't lying about not knowing anything?
posted by hoyland at 4:33 AM on April 20, 2018 [6 favorites]

It seems appalling to me that those who have been proven to be wrongly accused receive no shred of compensation for their wrongful imprisonment. Yet when I think about the idea of offering compensation, my mind jumps to that possibly apocryphal story about people in some country going back to kill the injured people that they had hit with their car, in order to avoid paying compensation for the rest of the victim's life. In other words, if there is compensation offered for the wrongfully accused, would that incentivize our oft-corrupt law enforcement even more to hide evidence and so on?
posted by inconstant at 6:26 AM on April 20, 2018 [2 favorites]

Situations with accidental transfer should be rare, right? I wonder if it will work out that in a crime scene where you take a thousand samples, 850 will the the victim or victim’s family, 100 will be recent visitors, and 40 are those committing the crime (assuming the perpetrator isn’t one of the friends or family). Then it isn’t enough to get only one hit, you should be getting two or three or more.
posted by Monday at 6:31 AM on April 20, 2018 [1 favorite]

Fascinating story. And a near miracle that a homeless man had a rock-solid alibi.

This is one of the dangers with big DNA databases. If you only took DNA from people implicated in the crime for other reasons, you couldn't get a false match on a stranger like this. If you're scanning a database of millions it's way more likely you'll find someone who is plausibly, but not probably, connected and then they need to prove their database.

The flip side is catching the actual killers in this case probably required their DNA.
posted by mark k at 7:17 AM on April 20, 2018 [8 favorites]

Riveting read, thank you.
posted by St. Peepsburg at 8:49 AM on April 20, 2018

It's important to remember that while DNA has extraordinary value for exoneration, being a great tool for proving that someone could not have committed a crime, it's still as garbage as any other forensic science when it comes to conviction.

Even if you have a clean sample, you have to look at something called the False Positive Paradox. Basically, when looking at a test, if the incidence rate of what you're testing for is lower than the false positive rate, a positive result is completely unreliable even on an accurate test.

Let's imagine a test for a disease with a 95% accuracy, and a population of 1000. 40% of them have the disease, and you want to know which 40% it is. So you test, and because of your 95% accuracy rate, you get 400 true results, and (1000-400)*.05 = 30 false positives. A given person who tested positive is still 93% likely to actually have the disease, and not be a false positive.

But if you take that test to a different population of 1000, only 2% of whom have the disease, you get a very different result. You get 20 true results, and (1000-20)*.05 = 49 false positives. A given person who tested positive is 29% likely to actually have the disease. More than two thirds of your positive results are false positives.

Now suppose you have something like DNA analysis, and you're checking a perfectly clean sample from a crime scene that you know can only contain the DNA of one person, the criminal. Say your test is competently executed and is 99.999% accurate. But your incidence rate in a population is 1, and DNA databases are HUGE. Run that on a DNA database of 5 million people, you might get one correct hit, but you're gonna get 5m*.00001 = 50 false positives. Even if your test is 99.9999% accurate, and you do get only a single correct hit, you can statistically expect roughly 1/6 odds that a given positive result is actually a real positive.

And yes, defense attorneys are making a stink about this, but we've had enough trouble trying to get courts to throw out stuff like hair, bitemark, or handwriting analysis that are basically astrology and proven to be constantly falsified.
posted by kafziel at 10:07 AM on April 20, 2018 [26 favorites]

"gagged him with mustache-print duct tape. ... The coroner would later conclude that he had been suffocated by the mustache tape."

What a strange detail to include, and even weirder way to refer to duct tape.
posted by Pig Tail Orchestra at 10:12 AM on April 20, 2018

It's really depressing that this man would be in prison for the rest of his life if he hadn't been in the hospital that night. The science behind the story is interesting, but the reality is without a rock solid alibi it wouldn't have mattered. No jury would have acquitted him without it.
posted by something something at 11:37 AM on April 20, 2018 [2 favorites]

Imagine we scrutinised the prior month in the life of every lottery winner, searching for some coincidence that seemingly brought them together at roughly the same time and roughly the same place as any of the recent murder victims in their metro area. Central station, the bar district, the mall, the football stadium. Who connects them in the 6 degrees game? It's not difficult to envisage a crossover at some point.This is effectively the exercise that's being carried out when you trawl a huge DNA database and snag a "hit" with no immediate connection to the case.
Humans are brilliant at seeing patterns where none exist, and at creating justifications for assumptions. Given that this level of investigation is generally reserved for extremely serious cases and that the people accused of such acts are generally not able to afford crack law teams. it's imperative that there are systemic checks and balances to mitigate such biases.
posted by Jakey at 3:58 PM on April 20, 2018 [5 favorites]

It's really depressing that this man would be in prison for the rest of his life if he hadn't been in the hospital that night.

True, but he also wouldn’t have been a suspect if he hadn’t been taken to the hospital since the transfer came from the paramedics.
posted by LizBoBiz at 12:43 AM on April 21, 2018

Situations with accidental transfer should be rare, right?

From the article:

There now exists a small pile of studies exploring how DNA moves: If a man shakes someone's hand and then uses the restroom, could their DNA wind up on his penis? (Yes.) If someone drags another person by the ankles, how often does their profile clearly show up? (40 percent of the time.) And, of utmost relevance to Lukis Anderson, how many of us walk around with traces of other people's DNA on our fingernails? (1 in 5.)

[...]

Consider a case in which a man is accused of sexually assaulting his stepdaughter. He looks mighty guilty when his DNA and a fragment of sperm is found on her underwear. But jurors might give the defense more credence if a forensic scientist familiarized them with a 2016 Canadian study showing that fathers' DNA is frequently found on their daughters' clean underwear; occasionally, a fragment of sperm is there too. It migrates there in the wash.

What the article does mention as rare is research on DNA transfer.

I was also kind of surprised that the only sort of "forensically aware" crime-committers mentioned were ones who attempted to eliminate DNA evidence: what about intentionally contaminating the crime scene with, say, dust from a high-traffic public place? If the crimer can introduce 1000 or 10000 other human genomes, or large quantities of DNA fragments from specific people, at what rate can they successfully mislead a forensics investigation?
posted by XMLicious at 4:11 PM on April 21, 2018 [2 favorites]