DNA Evidence is Fallible—And May be Getting More So

A review in The Atlantic of the history of DNA’s use in criminal investigation shows that DNA testing is not yet reliable. In fact, it may be getting less reliable even as its use increases.

The problem, according to journalist Matthew Shaer, is not in the science itself but in the humans who practice it. Scientific advances now permit DNA matching with very small amounts of biological material (say, a speck of spit invisible to the eye). They also permit matching when biological materials from several persons appear in a mixture; for example, a vaginal swab in a gang rape case may contain the DNA of the victim and the DNA of several perpetrators.

But making DNA identifications when samples are tiny or mixed requires some judgment. If the forensic scientist is eager to find a match, a match, genuine or not, is likely to be reported.

The article reports a disturbing research result by scientists Itiel Dror and Greg Hampikian. They sent DNA samples from an actual rape case to 17 scientists. Sixteen of the 17 said, after analyzing the samples, that they could not make a match. The police crime lab scientist who had the samples first, though, did report a match; because of that, the rape defendant was convicted and sits in prison.

(Hampikian is one of my witnesses in an attack I’m making on a rape conviction in Ellensburg, Washington, east of my office in Seattle.)

After a scandal at its DNA lab, the Houston Police Department created a new DNA lab that is said to be completely separated from the rest of the city’s police personnel, to prevent scientists from feeling any pressure to “help” detectives solve cases. In North Carolina, though, according to Shaer, law enforcement agencies that operate crime labs are still paid $600 for a DNA analysis that leads to a conviction.

To overcome the frailty of humans, some scientists have created computer programs that automate DNA matching. It is not a scientist who declares a match; it is an algorithm.

But those computer programs have their critics, too. Last year, in a case in New York, a trial judge rejected the evidence from one of them as unreliable.