Close window  |  View original article

The Science Doesn't Lie?

It does, if you don't understand it.

By Petrarch  |  November 21, 2007

Turn on the TV most any evening, and be prepared to spend time watching a police crime lab. With the proliferation of CSI shows - New York, Miami, Las Vegas, who knows where next - American audiences have been introduced to the science of criminology.  Time after time, some apparently perfect crime is unsnarled by scientists in the lab, analyzing a stray hair, piece of fabric, partial fingerprint, or other obscure but infallible identifier.

As a result, juries in the real world are increasingly demanding fancy scientific proof in order to return a conviction - the "CSI effect."  And since they want to win cases, district attorneys are happy to comply.  So we get tremendous increases in state lab budgets, outsourcing of analysis to independent private labs, and of course, the famous FBI forensic lab.

It's interesting to consider the underlying reasons behind this phenomenon.  At one time, the ultimate proof in court would be a sworn witness giving testimony, preferably a police officer.  But having been exposed to the many weaknesses of witnesses, from Officer Mark Fuhrman's perjury in the O.J. Simpson murder trial, right on up to President Clinton lying about his extracurricular activities, juries are understandably more cynical about the truth of what witnesses have to say.

They'd like to see something a little more concrete - something scientific, impartial, and reassuring.  Combine that with the Hollywood indoctrination of CSI, and this development makes perfect sense.

The only trouble is that "someting scientific" isn't always concrete.  Forensic evidence in the real world is rarely clear-cut and significantly more nuanced than forensic evidence in the TV world.  It has become increasingly clear that mistakes are rife in police labs, with impossible DNA returns and other such disturbances.

Recently, we see that even the FBI doesn't always get it right.

For many years, the FBI has done bullet-lead analysis comparisons for prosecutions.  The idea was that each batch of molten lead that's cast into bullets is slightly distinct from the last batch - it contains trace elements that the smelting process didn't quite remove, in different proportions based on any number of influences.  So a crime scene bullet can be compared to unfired bullets in the suspect's possession, and if they match, the suspect must have owned the bullet used to commit the crime, and therefore is involved somehow.

After decades of usage, researchers discovered that the statistics in the test are faulty so the conclusion is often wrong.  The bullet lead doesn't differ enough to say anything useful.  And the batches are too large - sure, one boxcar-load of bullets may differ from another boxcar-load, but you still have ten thousand different people with chemically identical bullets, who have nothing more in common than that their bullets were shipped from the manufacturer on the same day.  How is this helpful?

The test looked scientific and sounded scientific, and it allowed juries to feel better about convictions.  But how many times was the test the incriminating factor for an innocent person?  We just don't know -- because the juries saw the science (grouped with all the other measurements and figures), and were further reassured by it.

Our jury system requires a certain amount of work and thought on the part of the jury.  There is no shortcut.  A juror must listen to, and weigh, all the evidence presented; make adjustments for how truthful he believes it to be; and make a judgment call.

Cases are to be proven "beyond all reasonable doubt," which is far different than being proven beyond any doubt whatsoever.  Scientific evidence is only as reasonable as it is accurate.

Hollywood-saturated juries, lazy prosecutors, police blunders.  Innocents convicted - and, in some ways worse, guilty men who may now be wrongfully set free to commit other crimes.  What a mess.  There won't be a happy ending with justice served, like there is on CSI.