Wed Oct 23, 2002


Conservatives are up in, um, arms about the possibility of ballistic fingerprinting. The idea is that gun barrels make unique, consistent patterns on bullets they fire, and so bullets found at crime scenes could be compared to a database of factory-fired rifles. Conservative, pro-gun bloggers like Glenn Reynolds and the Volokhs hate the idea have been regularly linking to critiques of it.

I have no view on ballistic fingerprinting because I haven’t read the research. I was struck, though, by the analogy to regular fingerprinting, and reminded of a book I’ve been meaning to read: Suspect Identities by Simon Cole. It’s a social history of fingerprinting and criminal identification. Cole argues that fingerprinting established itself as a legal touchstone only after a long battle against other identification systems (like anthropometric ones that relied on measuring various body parts). Beyond that, the reliability of fingerprinting hasn’t been properly tested for a long time: the rate of false positive and false negative matches might be quite high. (The Economist reported a while back that, in one case, 8 out of a sample of 35 state law enforcement agencies failed to match latent prints at a crime scene to those of the person convicted of the crime. That’s about a 20% false negative rate, assuming the guy was rightly convicted. I haven’t looked into false positives.) The law’s reliance on fingerprints is very well institutionalized, but it’s not as scientifically solid as it may appear. The same goes, almost certainly to a much greater degree, to the reliability of eyewitness testimony. There’s a very large body of research on the problems of eyewitnesses. Courts aren’t about to downweight their testimony any time soon, because even witnesses who misrecall the facts are often compelling because they believe what they’re saying.

If ballistic fingerprinting doesn’t work at all, it shouldn’t be used in court, of course. But how good would it have to be before it would be about as good as eyewitness testimony and regular fingerprints?