The L.A. Times editorializes for reforming the forensics system:

In 2006, Congress charged the National Academy of Sciences with studying the application of forensic science in the U.S. judicial system. Its findings, released last year, are grim. Almost every branch of forensics but DNA testing -- hair and fiber analysis, arson investigations, comparisons of bite marks -- lacks the extensive scientific research and established standards to be used in court conclusively...

In February, the science academy issued a report calling for Congress to create a national institute of forensic science, and there is more than enough evidence that one is desperately needed. As an independent agency, not part of the Justice Department, it would be charged with conducting research, setting national standards for forensic disciplines and enforcing those standards. Right now, standards vary wildly. An expert in San Diego, for example, might testify that a fiber is similar to one found at a crime scene, while an expert in San Bernardino might testify that a match is impossible to determine.

Advances in forensics have revolutionized the judicial system, aiding both prosecutors and defense attorneys, exonerating the innocent and confirming the guilty in ways that were impossible just a generation ago. The patchwork state of forensic science should not become an excuse to shy away from its use; rather, the nation should invest in the rigorous research required to standardize techniques and application.

I'm generally skeptical of the "blue ribbon panel" approach to public policy, but there are really two issues that need addressing here, and one of them could actually be addressed by the sort of federal agency the Times endorses.

That problem, as the Times explains, is setting a baseline for what sort of forensic evidence ought to be admitted at trial, and establishing what level of certainty a specialist should be permitted to convey to a jury about his conclusions. The faux science of matching bite marks left on skin to human teeth, for example, should never be admitted into evidence. There's simply no science to support it. A fiber expert can convey important information to a jury, but should be required to accurately describe the limited evidentiary value of a fiber match. The other problem you see occurs when a forensics specialist testifies truthfully and accurately, but in closing arguments, a prosecutor (and less often, a defense attorney) will exaggerate the degree to which the expert's testimony implicates or vindicates the defendant.

If a federal standards-setting agency can survey the latest scientific research to issue guidelines trial judges then use to determine what evidence should and should not be allowed, and that appeals courts can then consult when determining when improper or scientifically unsupported testimony was wrongly allowed into evidence or improperly exaggerated by a prosecutor in his closing—that all seems like a good thing. It seems unreasonable to expect a trial judge to keep up on the latest forensic and medical research. I don't see much problem in having a government agency ensure that we're using good science in criminal cases.

But the other problem with forensics is the bias—intentional and otherwise—and human error that creeps into crime lab work. A standards-setting federal agency isn't going to be able to do much about the forensics specialist who gives testimony that falls within the parameters of the agency's general guidelines, but was influenced, perhaps subtly, by the fact that he reports directly to the DA or state attorney general, or he's a private specialist whose opinions might be influenced by who's paying for his services.

That's a problem that calls for the more comprehensive sorts of reforms that economist Roger Koppl recommended in a 2007 report for the Reason Foundation (this site's publisher). Koppl and I also wrote a condensed version of his recommendations for Slate.