Friday, March 7, 2008

Science pannel shoots down wider ballistics database

Panel Shoots Down Wider Ballistics Database

In recent years, some lawmakers and gun control groups have pushed for a national database that would record the ballistics signature of every gun sold in the United States. But a new report from a prestigious scientific panel says it's probably not a good idea.

The idea isn't necessarily a bad one; at least it appears well intended.

By comparing bullets taken from the victims, investigators knew that the shootings were linked. But to what gun? Whose gun? Police couldn't tell. The federal government does have a database of markings on bullets that police can search, but it includes only guns that have been already used in a crime.

That led some lawmakers to wonder whether the unique markings left by all guns should be recorded whether the guns should be fired and their ballistic signatures noted before they were sold. That way, if the guns were ever used in crimes, investigators could trace them.

Kinda like fingerprinting everyone in the US; one could then compare fingerprints at a crime scene with a national database to see who was there. Barring privacy concerns of course, which are not insignificant, but it appears as a technical matter it just doesn't work.

"We sell between 1 and 2 million handguns in the United States each year. So if you were to enter every single gun, you'd be entering 1 to 2 million every year," he explains.

Rolph says today's technology for comparing images would have a hard time sorting through so many. It would spit out lots of possible matches.

"It never tells you ... like on CSI ... 'this is a match.' It tells you, 'here's some potential candidates,'" he says.

Getting too many candidates would be a big problem ... because to know if a match is real, a forensics expert would have to pull the evidence for closer examination.

"It's OK to have them look at maybe 10 or 20. Do you want them to look at several hundred? ... It would be really very difficult to use in a practical way," he says.

So, the characteristics of rifling on fired bullets are unique but not able to be automatically identified and sorted via computers. The more database entries there are, the more possible matches and therefore the less useful the database is.

Apparently, the identification of which bullet was actually fired by which firearm is still only done by humans. Since humans make mistakes, I have to wonder how many miss-identifications have been made over the years.

From the Associated Press story on this same study, there's this interesting bit:

The report also questioned the underlying theory behind ballistic imaging. It said the idea that each firearm produces unique marks has not been scientifically proven.

Other variables make the program even more unreliable, the report found. For instance, guns leave different markings on different types of ammunition and the type of ammo used in a crime might not be the same type used during test-firing.

It seems it's quite possible that ballistic fingerprinting isn't even unique.

No comments: