DNA science on trial

My least-favorite crime show is CSI. And my least-favorite line in that show is "The science never lies." Talk about a whopper!

With all of the revelations lately about faulty science -- from handwriting identification to fingerprint evidence, ballistics, arson investigation, and even bite marks -- it is reassuring that at least one type of scientific evidence rests on a solid foundation.

That's DNA, of course. After all, as I have heard DNA experts testify many times in court, the likelihood of a wrong DNA match is something along the lines of one in 1.1 million, or less.

But what if that's just one way to run the math? What if you can run it another way, and the odds of a wrong match rise to a whopping one in three? And what if the FBI knows this, and is working feverishly to keep such information hidden from the public and out of the courts?

That is the alarming cover story by Pulitzer Prize-winning author Edward Humes in this month's California Lawyer magazine.

As Humes cogently explains, the problem is not so much with the matching of crime scene DNA to a known suspect. The problem is with the cold cases, when a database spits out a match to an offender who is not already identified as a suspect. The larger the databases get, the greater the odds of a wrong match by pure chance. Especially if the crime scene DNA is degraded, and fewer than the ideal 13 loci (DNA locations) are available for comparison.

This problem was confirmed by a technician at a DNA lab in Arizona back in 2001. Running an experiment by comparing all 60,000 offenders in her state's database, she was shocked to find about 90 coincidental matches. The FBI, unhappy over her disclosure, "began threatening sanctions against crime labs that shared such information with anyone outside of law enforcement," Humes reports.

Experts maintain that wrong matches are becoming more and more inevitable, given the increasing sizes of DNA databases. In California, for example, with the third-largest DNA databank in the world, a new policy of profiling everyone arrested for -- not just convicted of -- a felony is enlarging the database by 35,000 profiles per month.

Donald Kennedy, a former Food and Drug Commissioner who contributed to the recent National Research Council report critiquing the management of government forensic labs, confirmed that the science "is being shut out of court."

That may be forced to change, depending on the outcome of several appeals around the United States. For example, a 71-year-old wheelchair-bound man convicted of a 1972 rape-murder in San Francisco is appealing his conviction on the grounds that the jury was not allowed to hear the alternate statistics. Although John Puckett had other sex offense convictions, only a cold hit to some badly degraded DNA at the crime scene tied him to this one. There were no eyewitnesses, footprints, fingerprints, or confessions. (The case is People v. Puckett, No. A121368, Cal.Ct.App., 1st District, May 1, 2008.)

The National Research Council is calling for change. In their February report, they recommend improving reliability and transparency by taking the databases out of the hands of law enforcement and prosecutors entirely.

In the meantime, jurors -- hearing only the cheerleading version of the science -- will keep believing the CSI mantra that the science never lies.

Edward Humes's report in the California Lawyer is online HERE.