Comments

Ivan GluscicOctober 15, 2015 7:40 AM

Penn & Teller did a great episode of "Bulls**t" on forensics that I really recommend. I appreciate what forensics can accomplish, but it's not a science in the strictest sense. There's no peer review, for one thing... though maybe these articles indicate that's starting.

Scott HensonOctober 15, 2015 7:49 AM

Not all DNA analysis. It's DNA mixtures, especially for touch DNA or situations where 3, 4 or more samples are mixed. One to one DNA matches are still good, as are rape kits with two samples, one of which is known (the victim's). But for mixtures with unknown sources, there's a great deal of subjectivity in interpretation that belies the "DNA is the gold standard" meme, and math errors in the way many labs calculated probabilities. See:

http://gritsforbreakfast.blogspot.com/2015/09/a-reluctant-scoop-changing.html
and
http://gritsforbreakfast.blogspot.com/2015/09/labs-must-correct-wrong-dna-mixture.html

Tod ChristiansonOctober 15, 2015 7:53 AM

Your "Turns out it's fallible." summary of this issue is far too simplistic and inaccurate. The article it is based on is sensationalized in order to attract readers. The whole idea behind utilizing familial DNA is to generate suspects who can then be identified or eliminated based on their own complete, unique DNA profile. In general this approach would only be applied in serious crimes where other avenues of investigation have turned up no leads.

The article mentions that one person spent 30 nerve wracking days waiting for DNA results. First of all, that is a relatively rapid turn-around time. More importantly, after that 30 days if there is no match, he is no longer a suspect, it's over. What about those individuals who are suspects for years or falsely convicted because we do not use this particular objective and impartial analytical tool? What about the many unsolved cases in which the investigation has stalled due to lack of leads?

The article also cites that the technique is not really that effective, quoting a statistic of 17% efficacy in the UK. However, the cases that these techniques are applied to are usually the difficult to solve cold cases. A 17% conversion on those cases would be a very significant success.

DNA analysis by Short Tandem Repeats (STRs) remains the most powerful and objective tool we have for associating crime scene evidence with individuals (witnesses, suspects or victims) to simply pass it off as fallible based on this skewed article does not serve justice.

Full disclosure: I am a retired Forensic Scientist who performed DNA analysis and was a CODIS administrator.

GreenSquirrelOctober 15, 2015 8:05 AM

Mitch Morrissey, Denver’s district attorney and one of the nation’s leading advocates for familial DNA searching, stresses that the technology is “an innovative approach to investigating challenging cases, particularly cold cases where the victims are women or children and traditional investigative tactics fail to yield a solid suspect.”

Is this the DNA profiling version of two of the four horsemen of the infopocalypse?

So it doesnt matter if it generates more false leads than positive ones or puts innocent people in the cross hairs of law enforcement because this is how we solve cases involving women or children.....

GreenSquirrelOctober 15, 2015 8:12 AM

@Tod

More importantly, after that 30 days if there is no match, he is no longer a suspect, it's over.

If only that were true. The person will remain on all searchable records as a possible "offender" and in cases where the investigation leaks to the media even this suspicion can be enough to complete ruin a life.

Imagine a teacher suspected of a child abuse case - during the 30 days in which they are being checked they cant work, people will eventually find out why and the no smoke without fire response has the potential to ruin their entire lives.

I fully appreciate that 17% is an excellent rate for cold cases but does it justify the potential to ruin an innocent persons life?

mbdOctober 15, 2015 8:16 AM

The DNA tests used in forensics is fallible, current technology could do much better. If you really look into what DNA tests could be used for (think race, ethnicity, some phenotype (e.g. hair and color as good guesses), it is really pathetic that courts still rely on a few genetic markers for id.

HJohnOctober 15, 2015 8:19 AM

On a related note, witnesses can be problematic too, as can confessions under duress. I caught an episode of Stossel* show where the topic was the criminal justice system. It featured a lot of people falsely imprisoned. Even eyewitness testimony is highly fallible, in addition to DNA (this has never been part of one of my examinations). I'm a certified fraud examiner, so I'm pretty familiar with flaws in seemingly convincing evidence. One reason testimony is fallible is what I call the "Must Be Here" fallacy. You put a lineup in front of someone, and they trust the authorities have done their jobs and the perpetrator is there, and when he isn't there, there is a tendency to pick the best match.

*I debated whether or not to include the show name, because it isn't about the host. I decided to include it in case anyone wanted to look up the episode, it was very interesting and depressing.

thevoidOctober 15, 2015 8:42 AM

@HJohn

One reason testimony is fallible is what I call the "Must Be Here" fallacy. You put a lineup in front of someone, and they trust the authorities have done their jobs and the perpetrator is there, and when he isn't there, there is a tendency to pick the best match.

i saw a documentary (probably on pbs) not long ago where they addressed this (and did some experiments testing it). indeed, most people will almost always pick the 'best match' UNLESS they are told that the suspect *may not be there*.

John MooreOctober 15, 2015 8:44 AM

It's not that DNA samples from the crime scene are wrong or that those tests are fallible. The problem is that the cops have started looking for close matches instead of exact matches. When you start doing close matches like familial matches, you are looking for 50% of the markers or less to align with the DNA markers of the crime scene sample. That will encompass a lot more people, none of whom were involved in the crime. The authorities are hoping that one of those innocent people will lead them to the person whose DNA is an exact match. It seems that the cops have gotten a bit overzealous in the use of a technology that has helped them and everyone else and they've started to abuse it.

ianfOctober 15, 2015 8:50 AM


@ HJohn,
        when you “decided to include the [TV show's name(?)] here in case anyone wanted to look up the episode” you essentially informed us that we can find it on the Internet… like, look up at the sky at night, there's that twinkling star. And when we find "it," we can start reviewing the episodes until we come across the very one that you, and only you, had in mind. Well, THANK YOU for these incredibly precise leads to nowhere, Mr. Certified Fraud Examiner.

BONUS, regrettably with direct-access URL for a change: to find out how correct deployment of DNA testing combined with old-fashioned sleuthing can lead to results, read Tobias Jones' account of the 2010 murder of Yara Gambirasio that has obsessed Italy in The Guardian.

parabarbarianOctober 15, 2015 9:16 AM

This is not so much that DNA testing is fallible but that cops and prosecutors don't care. If a person's career is boosted by his conviction rate then the bias is toward conviction over truth.

scotOctober 15, 2015 10:00 AM

The problem is, as with many science issues, seems to be that scientists are poor statisticians. "One in a million" to "one in 30 or 40" is an easy mistake if you screw up the setup, and as the birthday problem shows (as well as the profitability of casinos), people have an extremely poor intuitive grasp of probability.

DanielOctober 15, 2015 10:07 AM

It seems that the cops have gotten a bit overzealous in the use of a technology that has helped them and everyone else and they've started to abuse it.

Yes, this. When one is bully with a hammer one will find anything one can to hammer.

What about those individuals who are suspects for years or falsely convicted because we do not use this particular objective and impartial analytical tool? What about the many unsolved cases in which the investigation has stalled due to lack of leads?

The former head of the FBI already answered your question. It is not and has never been the goal of any civilized society to make legal processes or technology maximally efficient for law enforcement.

You might have made a great CODIS administrator but you make a terrible citizen when you put the demands of your own professional expertise above the good of society.

Finally, I'll add one further point regarding the first article. Whatever the problems with the lab, if they go back to 1999 they not are caused by the problem of "bad math" because probabilistic genotyping did not exist in any meaningful sense for LEA in 1999.

jaysonOctober 15, 2015 10:24 AM

Fortunately, we have infallible lie detector tests to double check the faulty DNA data.

/sarc

TatütataOctober 15, 2015 12:13 PM

German police forces were looking in the late nineties/early noughties for a cunning serial killer who had perpetrated dozens of murders all over the country, and a couple more abroad. The ghost defied all attempts at profiling, as it had about as many modi operandi and apparent motives as there were victims.

They finally got their man, or rather their woman. Her job was packing cotton swabs of the sort that was used to collect samples at crime scenes...

This material wasn't certified for that purpose, but more something like garden-grade Q-Tips.

Using a genealogy registry for police purposes is sickening.

PeteOctober 15, 2015 12:27 PM

PBS/Frontline has an episode which shows that fingerprints are not unique either.

These are all techniques to reduce the number of suspects, but claims of 98% accuracy and above are only claims. It seems.

ianfOctober 15, 2015 12:36 PM


@ Tatütata […] “[the serial killer's] job was packing cotton swabs of the sort that was used to collect samples at crime scenes...

Afraid I didn't get it… she was finally caught because of inadvertent contamination of the [off-the-shelf] cotton swabs that happened to be used at some actual crime scene, and thus led the hitherto unwitting police to that DNA-ed individual?

Does that serial killer case have a "duckable" name? Googletranslate is me friend.

.“Using a genealogy registry for police purposes is sickening.

Nevertheless, if it's a legal registry with no preconditions as to how it is to be used, then I suppose the police looking for matches of crime-scene DNA should not be automatically prevented from using it. The key here is existence of an overseer instance, which, although it has the same state provenience, lives acc. to its own regulations, and is answerable in turn to the highest democratic/ parliamentary authority.

HJohnOctober 15, 2015 3:45 PM

@ianf: when you “decided to include the [TV show's name(?)] here in case anyone wanted to look up the episode” you essentially informed us that we can find it on the Internet… like, look up at the sky at night, there's that twinkling star. And when we find "it," we can start reviewing the episodes until we come across the very one that you, and only you, had in mind. Well, THANK YOU for these incredibly precise leads to nowhere, Mr. Certified Fraud Examiner
_________

You could have went to the show's website or IMDB and found it with less typing than your ridiculous rant, no PhD is required.

It's amazing the things intellectual midgets will bicker about.

ianfOctober 15, 2015 4:21 PM


@ jayson

It took the German police 16 years, and presumably the same in DM & Euros, to start suspecting wholesale contamination of their own forensic-testing materials? What a bunch of woollies.

    At a local hospital the electronic, computer-controlled lock to some wards' medical supplies (not medicine!) store was discovered to be faulty. Nothing was missing, everything was accounted for, and no signs of tampering with sterile bandages etc were detected. Nevertheless, the entire ~€200k truck-worth contents of it were first sent to be irradiated, then donated to a charity as non-sterile, outdated material (boil before using). And yet I'm talking of a regional hospital merely calculating the cost of the alternative: potential of a contagious disease due to momentarily unsecured somehow contaminated items being far greater than the not inconsiderable outlay for restocking the supply store.

@Tatütata […] “They finally got their man, or rather their woman. Her job was packing cotton swabs of the sort that was used to collect samples at crime scenes...

Please be precise: not A, but several WOMEN of a similar Eastern European mitochondrial-DNA profile. It was the gender of the alleged serial killer that piqued my interest, as it is well known that there are few such around (and then rather conforming to Charlize Theron's "Monster" roadside-hooker role model).

SteveOctober 15, 2015 6:54 PM

I gave dna in a family study to JHU in 1993, now I worry. I had read a little before on dna forensics and mathematics http://dna-view.com/profile.htm. I have great fear now given the stupidity and ignorance of the average juror who regards it as infallibleand easily mislead by the prosecutor's fallacy, "The chance is 1/7000 that someone (anyone) other than the suspect left the stain." versus the correct statement, "The chance is 1/7000 that some (particular) person other than the suspect would leave a stain like the actual stain." And then hair and fiber analysis has been b.s. for 40 years, https://www.washingtonpost.com/local/crime/fbi-overstated-forensic-hair-matches-in-nearly-all-criminal-trials-for-decades/2015/04/18/39c8d8c6-e515-11e4-b510-962fcfabc310_story.html

we are lost

SteveOctober 15, 2015 7:22 PM

Justice is a myth perpetuated by our total immersion in television crime dramas and CSI, the real propaganda. Vicarious justice to quell and comfort us. To instill unshakable belief in forensic infallibility. Forensic "scientists" are never objective, but always witnesses for the procesecution, ready to spin their testimony to aid a conviction. And of course Justice has never been blind to money and power.

SteveOctober 15, 2015 8:03 PM

The article here: http://dna-view.com/profile.htm points out that it makes a big difference in the probabilities whether the arrest is made before or after the dna is sampled and mathed. They only test for 4 or 5 alleles, so their is let's say a 1 in 7000 chance that a randomly chosen member of the population (say male) will be and exact match a crime scene sample or 4 or 5 alleles.(unmixed). Remember they are not sequencing and entire individual's genome - be careful with the words exact match.

So now what happens most likely is that law enforcement does a a canvas search of volunteered dna or a canvas of a dna database, essentially a random selection of the population at large, and if large enough there will be at least one match. They stop searching at that point. Now law enforcement is as ignorant about math science probaility and dna as anyone. They say, "we got him, we got the guy! his dna matches." They go to the victim and say, "we got him we got the guy his dna matches! he's the one! All we need from you are a few more details - anything you can remember, to put him away, so you can have closure." The unfortunate victim or family just as ignorant about dna, believe the cops. The cops lead and coach additional details. They arrest if they haven't already, go to court, and with forensic and procecutorial spin it's a slam dunk conviction. Scary.

On the other hand if other evidence leads to arrest in the first place, then dna is sampled from the suspect and it matches, then the dna is very strong coroboration...

It makes a difference

ianfOctober 15, 2015 8:25 PM


@ Steve “Justice is a myth perpetuated by our total immersion in television crime dramas and CSI, the real propaganda.

    One form or another of justice was there well before the advent of television, or, for that matter, even drama. If not the current (in the history of Mankind pretty mild) form of justice, what would you propose instead?

    Wait!—I know: we'll put Mr. Certified in charge, he'll sort us out. ;-))

Forensic "scientists" are never objective, but always witnesses for the prosesecution, ready to spin their testimony to aid a conviction.

And of course Justice has never been blind to money and power.

    Name one area of life where that doesn't apply.

TPOIVABOctober 15, 2015 11:52 PM

Wasn't there a 'problem' with duplicate results from DNA searches found in the Arizona database reported some time ago? When doing these familial searches wouldn't you find many close matches? Can one ask the DNA expert how many close matches or duplicates they found? Do the searches stop at the first match thus never finding the duplicates? Is every 'new' DNA entered into a database first checked to see if it already in there of close (familial) to any others?

tyrOctober 16, 2015 12:05 AM


The real flaw is philosophical. The general acceptance of
certainty in some form. It turns out if you dig there is
no certainty in anything. So we have false prophets for
the infallibility of a method based on a false premise.

Once it gets accepted by law enforcement as infallible a
chain of tragedy ensues. They really want to solve cases
because work interferes with donut time so any expert
with a magic procedure will do. thoughtmaybe has a nice
documentary titled CSI on the subject. Scientism pushes
this kind of thing on the gullible all the time, not in
this area alone.

There are a lot more snake oil salesman running around
selling some form of certainty than there are people
doing peer review trying to limit the damage of flawed
systems. If you wind up as the defendant being opposed
by infallible forensics in front of a jury who thinks no
mistake is possible, then no evidence that proves you
could not have been physically present will save you.

Let's not even contemplate the rigor of collection that
takes place to obtain the infallible evidence.

Humberto MassaOctober 18, 2015 6:28 AM

The problem is never evidence; is HOW evidence is gathered (and many times, WHY it is gathered).

ianfOctober 18, 2015 9:47 AM


@ Humberto

I wouldn't put it in quite those words, but collection of crime scene forensic samples always needs to be first subjected to the same kind of elimination scrutiny, that the samples later will be examined under. True scientists (and intel analysts) always start by doubting the veracity of obvious and/or expected and/or easily extracted results of an investigation.

Unfortunately, in many cases and too many places, the forensic sciences are viewed as a tool for the prosecution, waved about when it fits the state's narrative, suppressed when it doesn't. Observe, however, that, even in the most stringent Western jurisprudences, forensic evidence per se is not needed for a conviction; other factors weight heavier than whether there were any fingerprints or blood and fibre traces.

Indeed, there are plenty of cases of later admitted miscarriages of justice where there was no, or next to none technical evidence, yet severe sentences were meted out anyway.

Forensic evidence is thus like a club used by that party which best can explain its importance in the criminal context. For that reason alone, any DNA evidence over which hangs a whiff of a suspicion that it might have been contaminated, should automagically be excluded. Just as, to take another known example from the now-exonerated Amanda Knox' case, any prosecution during which the prosecutors' own technical experts manage to irretrievably destroy ?HOW? all the accused's hard drives ?HOW? with corroboration of her movements and mail history in the course of data-forensic examination of same, should automatically have been abolished.

Leave a comment

Allowed HTML: <a href="URL"> • <em> <cite> <i> • <strong> <b> • <sub> <sup> • <ul> <ol> <li> • <blockquote> <pre>

Photo of Bruce Schneier by Per Ervland.

Schneier on Security is a personal website. Opinions expressed are not necessarily those of IBM Resilient.