Thursday, June 10, 2010

Beating the Odds

If you do felony criminal law (from either side of the aisle) and I tell you the number is 6.17 quadrillion, you probably assume that I'm talking about DNA.  
The number will reflect just how unlikely it is that the DNA in the whatever left at the scene could have come from anyone other than the defendant.*  There are four things you need to know about that number.
  1. It's from a profile.   That is, they didn't sequence the billions or whatever bits of stuff make up the DNA.  They sample and test particular spots on the DNA sequence.  That results in a profile.  And a profile is not a complete person.  
  2. It's meaningful only if the sample wasn't contaminated, the testing was done right, the demographic base was properly established and the test results properly measured against that demographic base.
  3. None of those things is as simple and free from mistake as the people in the lab coats would like you to believe.  Test results, for instance, may include bits of noise (think static on your car radio), but exactly what's noise and what's real is often subject to varied interpretation, which can lead to dramatically different results.
  4. While it's true that there aren't anywhere near 6.17 quadrillion people in the world, that doesn't mean that there can't be another person whose DNA profile is the same.  Really unlikely?  Sure.  Impossible? No.  Goes back to the idea that the number comes from a profile.
I'm not saying that when the guy in the lab coat is probably wrong when he says the tests were run and there's just a 1 in 6.17 quadrillion chance that it's not from your guy. He's probably right.  And when the prosecutor tells the jury that they should think about those odds and the fact that there are just about 7 (or is it 8 now) billion people in the world and conclude that the defendant did it, she's likely to be not just persuasive but frankly correct.
Of course, DNA doesn't just convict.  It also exonerates.  There have now been over 250 people in this country freed after DNA testing revealed that they didn't commit the offenses of which they were convicted.  The Supreme Court, though, says that there's no constitutional right to post-conviction DNA testing.  And prosecutors routinely resist the effort.  (See here, for instance.)
The highest profile case out there right now is Hank Skinner's.  He's on death row, been trying for some years to get sophisticated testing done.  The state (Texas, of course) did some, and when it didn't implicate him, decided not to do any more.  Even though it could be done at no cost to the state.  Skinner's in the Supreme Court now (decision next year) on the question of what procedural avenue he has to pursue in order to try and get testing. 
But if prosecutors often not always, maybe not most of the time (it's hard to tell), but often fight efforts at post-conviction DNA testing (even though it sometimes proves they had the right guy), defense counsel aren't always enthusiastic about it pre-trial.  Radley Balko of The Agitator, this time writing at and using Skinner's case as paradigm, says just cut to the chase and test everything.
There are times when neither the prosecution nor the defense is particularly interested in discovering the truth. That's where policy makers need to step in. In cases like Skinner's, they should establish a common-sense rule: When there is biological evidence at the crime scene, all of that evidence should be sent for DNA testing. No exceptions.
. . .
In District Attorney's Office for the 3rd Judicial District v. Osborne, last year's case, Justice Alito argued in a concurring opinion that guilty people could refuse to request DNA testing at trial, then prolong the appeal process (and stave off execution) by requesting DNA testing afterward. To find a right to post-conviction testing in the Constitution's protection of due process, Justice Samuel Alito wrote in his concurrence, "would allow prisoners to play games with the criminal justice system."
That's precisely why the testing should be done before trial. Arguing over which evidence gets tested shouldn't be part of either side's strategy. The prosecution and the defense should begin knowing that all of the evidence has been tested or will be. For old cases like Skinner's, if there's significant doubt about the defendant's guilt that testing could resolve, legislators shouldn't wait for the courts—they should make sure themselves that testing is done. A typical DNA sample costs about $1,000 to analyze, with a usual turn-around time of about 30 days. Innocence Project spokesman Eric Ferrero told me that his organization on average pays about $8,500 per case for DNA testing, since most cases have multiple samples of evidence.
As a criminal defense lawyer, I'd rather have some control.  As a citizen, I think he's right.
What got me started on this, though, wasn't Balko's piece at Slate.  It was Walter Reaves of Texas and his discussion of what happened in Ohio last week.
It's about Stephen Myers.  That's the factually innocent Stephen Myers.
Ten years ago, there was a home invasion in Licking County.  Some DNA testing was done, but there was nobody to match it against.  Until last year when someone ran the cold case DNA and up popped Myers.  Oh, he didn't match the physical description of the guy, and he said he was innocent, but the numbers were good.  Only 1 in 1.6 million people would have the same DNA profile.  That's not the quadrillions we're used to seeing (though I once had a case with a client - convicted largely on other evidence - where the number came back 1 in 4), but it's still pretty good odds.  Enough, certainly, for an indictment.
Trial was scheduled to begin in May, and Myers' lawyer was intending to attack the sampling and testing and numbers.  But then something interesting happened.
The prosecutors got the trial delayed because they decided to do more testing.  You know, DNA testing is a whole lot more sophisticated now than it was 10 years ago.  A whole lot more.  And it turns out that Myers is excluded.  He cannot - as in not possible - have been the person whose DNA was profiled.  1 in 1.6 million be damned.
It's a fluke, an oddity, that the prosecutors decided to retest.  Reaves sees in this story not just a prosecutor who did the right thing, but a lesson for defense lawyers.  When your client insists on his innocence, and after you've had a heart-to-heart about what testing might prove still wants it done, you should have it done.
Don't trust the prosecutor to do your homework for you - or, sadly, to give you the good news if she does.  Maybe.  But maybe not.  Regardless, defense counsel have a duty to investigate, not merely to trust that the prosecutor got it right.
But there's another lesson in this, too.
Next time you're on a bus or a plane or a train or in a restaurant or movie theater or anywhere where there are a bunch of people, look around.  
You never know when your not-twin, the one whose DNA profile is the same as yours, might be in the crowd.  Despite the odds of 1 in 6.17 quadrillion.   Hell, it might be one of the jurors.

*In this way, we distinguish DNA from fingerprints, where the examiner will testify (and yes, I have the transcripts) that from his examination of the fingerprint and the latent print, it is absolutely certain and he cannot be wrong (not even 1 in 6.17 quadrillion times) that the accused and only the accused could have left that latent print.  (We know those examiners are sometimes wrong, but they apparently don't know it; alas, neither do jurors.)


  1. Jonathan C. HansenJune 11, 2010 at 6:21 PM

    You know, I wanted to comment on this topic last night on Gamso's blawg, but was too tired. Seeing this picked up by Windypundit reminded me. There is another aspect to these probability inferences that is often overlooked.
    There is an excellent video regarding this topic on YouTube from the TED talks, that shows how totally unable people in general are at reasoning with statistics (
    For instance, in the Stephen Myers case, the probability of 1 in 1.6 million may well have been presented as there is only a one in 1.6 million chance that he didn't do it. But that is not correct, as it is not the whole story. As pointed out in the video, one has to take into account another piece of information; in this case the "prevalence" of other people that have the same combination of DNA segments.
    A one in 1.6 million chance, multiplied by, say 300 million as the number of people in the US, is 1 in about 187 that he DID do it. That is, given the large number of people in the US, there will be 187 or so others that have the same combination of DNA segments that were tested as does Stephen Meyers.
    The video is very good, well worth watching, and certainly opened my eyes. I hope no one is offended if I post this comment on both blogs...

  2. Jeff-what would you make of this hypothetical:

    Client against whom there is a fair amount of implicating circumstantial evidence, likely enough to convict in the absence of some explanation that currently does not exist, insists on his innocence. The overwhelming likelihood is that if the client did not in fact commit the crime, one of 3 other people, all of whom have been swabbed, did.

    There is an item that is likely to contain the perpetrator's DNA. Client insists that his DNA will not be on it, after a heart to heart come-to-Jesus, you're completely fucked if your DNA comes back on this. If the DNA of one of the other 3 people is on the item, it would almost certainly result in a dismissal. Client's speedy trial right is too close to accomplish the testing without a time waiver and client refuses to waive time.

    Do you waive over his objection based on his unwavering claim of innocence and the significant chance that if his DNA is not on the item that case is much better if not gone? Or do you take his refusal to waive time as a passive acknowledgement of the fact that his DNA will probably be on the item and proceed to trial without results?

    Obviously, the question is not entirely hypothetical. I have this situation currently and great minds I have consulted disagree. Interested in what you or other CDLs here think?

  3. Jeffrey Gamso - Kudos to you for not only doing your job, but for passionately, persuasively and effectively doing your job - consistently. I've been searching the internet for your email address. Please write me at
    Thank you!