A forthcoming study from the National Academy of Sciences is expected to send shockwaves through the criminal justice system.
People who have seen it say it is a sweeping critique of many forensic methods that the police and prosecutors rely on, including fingerprinting, firearms identification and analysis of bite marks, blood spatter, hair and handwriting. The report says such analyses are often handled by poorly trained technicians who then exaggerate the accuracy of their methods in court…
Legal experts expect that the report will give ammunition to defense lawyers seeking to discredit forensic procedures and expert witnesses in court. Lawyers could also use the findings in their attempts to overturn convictions based on spurious evidence. Judges are likely to use the findings to raise the bar for admissibility of certain types of forensic evidence and to rein in exaggerated expert testimony.
Law enforcement organizations have tried to derail the report nearly every step of the way. The report's critique of forensic evidence is much needed, but the proposed solution doesn't sound promising:
It concludes that Congress should create a federal agency to guarantee the independence of the field, which has been dominated by law enforcement agencies, say forensic professionals, scholars and scientists who have seen review copies of the study.
I wouldn't mind seeing an agency within the Department of Justice devoted to investigating and prosecuting cases of forensic fraud. If prosecutors are conspiring with or pressuring experts to deny criminal defendants a fair trial, that would be a due process violation and under the Fourteenth Amendment, the federal government would be permitted, or even obligated, to step in. Certainly Mississippi, for example, has neglected its duty to ensure that its citizens accused of violent crimes are given a fair trial.
But setting aside wilful and criminal misconduct by forensic experts, the problems with the forensics system aren't going to be resolved by creating a new federal bureaucracy. Lack of federal oversight isn't the problem. According to the New York Times article linked above, for example, the NAS report is particularly critical of the FBI crime lab, long considered the gold standard in forensics, and whose technicians often advised state crime labs on best practices.
The problem with criminal forensics is the government monopoly on courtroom science in criminal trials. In too many states, forensic evidence is sent only to state-owned or state-operated crime labs. There's no competition, no peer review, and in some cases, crime lab workers either report to or can be pressured by prosecutors when test results don't confirm preexisting theories about how a crime may have occurred. This sort of bias can creep in unintentionally, or it can be more overt. But studies show it's always there. The only way to diminish is to bring competitors into the game, other labs who gain by revealing another lab's mistakes.
Every other area of science is steered by the peer review process. It's really unconscionable that criminal forensics—where there's so much at stake—has existed and evolved so long without it. The fact that so many people have been convicted solely based on pseudo-sciences like bite mark and hair and fiber analysis, for example, ought to scare the hell out of us.
Roger Koppl wrote an excellent paper for the Reason Foundation (pdf) that outlines the problems with criminal forensics, and offers a series of reforms for ensuring that the science jurors hear in the courtroom is actually science. Koppl and I also co-wrote a shorter piece in Slate making essentially the same points. Koppl's proposals sound radical, but only because we're so used to the current system. What he's proposing is really little more than applying basic scientific principles like peer review, blind testing, and repetition to the evidence and opinions currently presented in criminal cases as actual science.
And we need to start treating criminal forensic science like actual science, with all the skepticism and repetitive testing that comes with it. What we don't need is another layer of government bureaucracy that imposes a series of negotiated, compromised-for standards and practices, then fails to properly enforce them.