Traces of Crime: How New York’s DNA Techniques Became Tainted

Yes, the "science" is always good. So may i suggest we go back to the simplest and one of the older scientific methods of determining if a person is guilty or not. Take the charged party to a lake and throw them in. If the lake accepts them (well, if they drown) then the lake proves they are innocent. If they float (rejected by the lake due to guilt) then we hang them. Of course, either way they are dead, but justice and science wins and the DA's score another point for truth, justice and their career, which is really what this is all about.

I think plea bargaining should be outlawed. It encourages innocent people to plead guilty to avoid the risk of worse outcomes and also encourages bullying by prosecutors and police. Yes, it's cheaper. Justice shouldn't be cheap.

They need to include real world examples with this sort of micro DNA evidence. If I was told that the DNA sample came from fewer cells than in a flake of dandruff, well, I know I don't keep track of everywhere my dandruff floats, it probably gets places I've never been.

Thank you for having this story and investigation. On one hand, my cynic goes, there goes another set of people who have Our lives in their hands, acting in their own self interest, despite the serious, life altering harm, they cause others. On the other hand, the energy that is sweeping our planet is helping to bring transparency, this is what it looks like, today. It would behoove us to clean up our own lives, and work our way out.

There is another factor never put into the stats: what is the lab sampling error? You can be sure that if "guilty odds" of 1,000,000 to 1 are given, the lab sampling error is far greater than that

And this in Democratic NYC. The one-party state gives rise to hubris, rule bending and projects such as this that become runaway trains that are stopped only by chance, and far too late.

DNA, is already tainted to begin with.... so let's not just blame "one place, one person" and then the rest are fine. Not so. Tracking can be specific, in terms of an individual's position during the crime. Tracking individuals does not need DNA. ---- Kidnapping will be a thing of the past, for sure. Intelligent Science, not just science! Intelligent Law Enforcement, doesn't create more problems.

How about this scenario:

EXPERT WITNESS: Well, the police didn't find ANY DNA material, but they really wanted a conviction, so we ran the suspects "invisible" DNA through our software and it proved he was the donor with a 99.9% accuracy.

See the problem here?

Convictions at any cost.

If there isn't enough material to run a standard DNA test and a "result" relies on any type of speculative software logarithm, introducing the result as evidence violates the "beyond a reasonable doubt" doctrine.

Why are Americans so scared they want to lock everyone else up?

It is time for the City of New York to require ALL of the software developed by the city to be open source. This is basic level of transparency that all democratic governments should be meet. (Of course getting open source software analyzed by competent experts is another entire issue that needs to be addressed as well.)

And while we're at it, the city should make all of the intellectual property public domain. If the federal government has to do it, so should our city.

I just hope Prosecutors ans law enforcement are not more interested in a conviction than they are about the Truth and justice.

Let's take a step back here. I know that everyone mistrusts the government today, but this article is obviously one-sided. Don’t get me wrong, I have no quarrels with either side, but a better picture must be painted. Fine, one stance is that the science is bad. But I’m sure the scientists will say that the science is good. Why, then, use the media to attack one side only? While the science may have put someone in jail erroneously, what I want to know is has it exonerated people erroneously as well? Maybe those that were let go because of this science should be re-trialed and put back to jail. Fingerprinting came under the same scrutiny in the past, but it was up to the courts and to the lawyers needed to do their job. To shout loudly using the media at the top of your lungs does not make your side right. If anyone thinks OJ was acquitted because he did not do it, you’re a fool. He got away because of technicalities. Because of that a murderer was on the street and my family was less safe – that we have to thank Scheck.

The Innocence Project is only a platform for Scheck to hide his income. Don’t ever think he doesn’t privately benefit monetarily from every acquittal that comes his way. How do I know? Because I'm also a lawyer and I argue for the side that's paying me even if I disagree with it. I will say that he's much more successful in it than I.

What the heck were those geneticists doing applying "the frequency with which each piece of DNA appears in different racial or ethnic groups" to individuals!
Did they miss the lectures about how you can't deduce someone's racial or ethnic background using appearance and name? You can predict the racial and ethnic background of a DNA sample - but you should not guess the background of a suspect to estimate the likelihood of a match.
Reasonable Doubt was built into their algorithm.

Never forget, regarding computer generated findings, GIGO [garbage in, garbage out]. States expect that folks possessing no real knowledge & understanding of science, scientific method, DNA issues, & statistics will can properly come to a decision regarding guilt or innocence. Where even Judges may not understand those same issues & concepts, how can one expect a jury of high school graduates or in some cases college graduates to know when to determine that there exists reasonable doubt? Between you me & the wall, a 30% margin of error equates with reasonable doubt, aka, the innocence of the accused.

I pity the poor ignorant criminal or even smart criminal who has to decide whether to take a plea, although innocent, or roll the dice, rely upon that previously described genius jury or over worked defense attorney or mystified judge or eyes on my future prosecutor so as to keep the accused out of prison. BTW, defense attorneys are not free & Legal aid is overworked.

You see, DNA evidence is about convicting the right individual, not just any individual. As a society we must never lose sight of that reality. The science we rely upon to deprive someone of his/her freedom, must be free from nonsense like a 30% margin of error or properly vetted using appropriate scientific method. All scientific opinions must be considered & no one fired to protect the process. The victim, of course, must await the conclusion of the process to have his or her say, if the victim is alive.

The secrecy of the algorithms that are used to make statistical determinations that can affect liberty is hard to understand in a democracy. We see the same secrecy in policing decisions about when and where to allocate police, in parole predictions about who is likely to recidivate or remain crime-free after years (and sometimes decades) in prison, and is who is likely to skip bail or commit another crime while released. Here, secrecy was invoked to limit the ability to evaluate the mathematics of a DNA match, with profound liberty interests and consequences. Protecting the "intellectual property" of a privately-developed algorithm has to be offset by the possibility of error and profound human costs of a flawed statistical model. And secrecy is the antithesis of science, where replicability and reproducibility are fundamental to the scientific enterprise. Lawrence Tribe was almost years ago when he decried the idea of "trial by mathematics." Mathematics is not the problem, secrecy in this case is antithetical to due process and fundamental liberty.

Tracing and Tracking Systems need to improve. When that happens, DNA will be a thing of the past. "Intelligent Science" should be the standard, not just science.

Clearly, police and prosecutors should be penalized for false convictions, not incentivized for convictions.

Why?

Not just because innocent people should not be incarcerated (also very expensive), but because they enable real perpetrators to prey on us!

I don't know what the evidence was against Mr. Herskovic but one assumes that alternative theories -- transference, incorrect statistical calculations or assumptions, lab error -- were presented to the jury, which weighed those arguments against the totality of the evidence and still found him guilty beyond a reasonable doubt.

Either there was more evidence than this article would suggest or Brooklyn juries are incredibly pro-prosecution. The latter seems unlikely.

This example demonstrates why there's no excuse not to let defense experts vet black-box forensic methods. Similar errors arose in DNA mixture calculations used in all Texas crime labs until 2015. Now, proprietary black-box methods are being used to replace the erroneous, old DNA-match calculations in Texas labs, but most front-line lab analysts don't understand the math and defense experts can't doublecheck them. What if they're as flawed as the NYC method? See: https://gritsforbreakfast.blogspot.com/2017/08/courts-punt-on-forensics-...

DNA evidence is only as good as the collection technique. As a chemist involved often in detecting environmental crime, I have often been appalled at the sloppiness of field techniques. While the lab is rightly held to very high standards of calibration, traceability of materials, training and so on, field data is often collected by people with little or no understanding of the procedures. I was once responsible for a huge investigation which involved 30,000 samples. 10% of the samples were replicated in the field, and 10% were replicated in the lab. The field error rate (I.e., non-agreement between replicate sample results) was approximately 1% and the lab error rate was approximately 0.06%. I have no reason to suspect that the data for forensic evidence would be very different.

It is curious that Barry Scheck did not publicize or otherwise bring legal action to stop the flawed DNA techniques that he voted against as a commissioner on the Commission on Forensic Science.

If anyone could have gotten people's attention, it would've been the founder of the Innocence Project and the lawyer who got OJ off by attacking LAPD's bad forensic techniques.

But Barry Scheck has always played a deeply conflicted and contradictory role in the development of how DNA evidence is collected, analyzed and used in American jurisprudence: Sure, the forensic evidence against OJ was deeply compromised, but was OJ innocent?

The same DNA flaws that could convict the wrong man could free the wrong man, and therein lies the conflict for something like the Innocence Project: no defense lawyer would attack forensic science as rigorously as an objective scientist would attack forensic science.

The truth is all forensic science as practised on a daily basis by law enforcement and by defense attorneys is not peer-reviewed science. If the results are favorable to you or your client, you are disinclined to challenge them for he sake of scientific objectivity.

In a jury trial, the jurors must be able to evaluate the reliability of the data collected by police. 25 years ago, I was a juror in a trial where the evidence included DNA analysis of blood samples, and it wasn't difficult for us to determine that the evidence was good. But today I would be especially suspicious of police evidence that uses the FST software - it is illogical to trust a software program that throws away data that might benefit the defendant.

A well written, balanced, and important article.

DNA evidence is a major advance in criminology. Though imperfect it is vastly superior to the tools of evidence which proceeded it, such as eye-witness identification.

But the technology has its limits and its flaws, particularly when it is pushed to the limits. A picogram is an unimaginably tiny amount of DNA. And so problems have arisen when DNA technology is pushed beyond its capabilities, and used to analyze samples that are, by rationale standards, too small to be analyzed, or when the samples are contaminated by DNA from multiple individuals.

One of the major problems described here is that virtually no one, including those who gave testimony, had the expertise to look under the hood, and evaluate the software; everyone simply accepted everyone elses word for the reliability of the tools, a veritable house of cards. This was compounded by the inexcusable lack of transparency of the NYC ME office, who fought to keep the software’s source code secret.

There are several lessons. Protect whistleblowers, don’t fire them. Be transparent, and in cases like this the software source code and underlying technology must be public.

In Massachusetts last spring, 21,000 drug convictions were vacated because of a rogue technician at the state crime lab, whose malfeasance should have been suspected far earlier than it was. Let’s not repeat this.

DNA patterns are relatively easy to read and a trained person can readily determine if the pattern from a crime scene matches a suspect's DNA profile. However, if a computer is needed to detect and analyze truly minuscule amounts of material, the odds for error increase rapidly. Simply stated, if that is all the police and prosecutors have, they don't have a case.

Thank you for a thorough report on an area I've long been interested in hearing more about- types of DNA analysis used in forensics and their reliability. I know something of testing of complex systems and its statistical analysis. I have seen a number of cases of users of such information hastening to likely erroneous conclusions based on their faith in the absolute reliability of answers coming from black boxes whose inner functioning, when painstakingly inspected, was dubious. Producing reliable information from testing complex systems is more expensive than many institutions' management realize or are willing to pay for if they do.
I hope to see more on this subject.

I have a PhD in Molecular Genetics and spent my entire career generating and analyzing genetic material in the study of evolution. What I repeatedly found was that degraded material or even poor technique never leads to a "match". You get garbage. You don't suddenly come up with a perfect match to some random person or plant or animal. The biggest problem with these amazing new, very sensitive techniques is the possibility of picking up contamination. If a suspect's DNA is present in the same laboratory there is a danger of cross contamination. But otherwise, generating sequences that matches some random person is statistically impossible.

I've been professionally interested in forensic DNA typing since 1990. One small but striking point is that Bruce Budowle is now lining up with the opposition to the OCME testing strategies. Thirty years ago, we raised exactly his current objections - that validation tests make over-simple assumptions about the simplicity and cleanliness of test material - to the FBI procedures then in use and he testified that these objections had no scientific basis.

"...the lab calculated that it was 133 times more likely than not that the remainder belonged to Mayer Herskovic..."
Apart from the other problems discussed in this article, is it not obvious that this is a comically weak argument for guilt?

Unsubstantiated denial is always the first response of those that know they are on shaky ground. As long retired Quality Control executive I know the bit. Just reading the article set my truth meter up several notches. My best guess is that in time we will see more resignations and perhaps more severe consequences.

Hubris--the bane of good science.
It is inconceivable how essential evidence good be deemed "secret and proprietary. What justification?

With so many legitimate questions about the validity and reliability of those techniques, one wonders why they are being defended so vigorously. Someone has skin in the game. I'd guess it's either financial, or merely reputational, but there are real lives on the line. With margins of error in the 30% range, these methods should never have been admitted as evidence. With no one from the lab having comprehensive understanding of the software, it should never have been purchased, much the less applied. This is as egregious as the Law and Order episode about a rogue prosecutor in Jack McCoy's office, and should result in the same re-examination of every case it impinged on.

This is corruption plain and simple. Developing a scientific method to be used as proof with a software code that's kept secret? Incredible. Proprietary? How can evidence produced in a criminal court case be secret. The logic completely escapes me.