Canthook

Henry Lee is the criminologist and chief emeritus in the Department of Public Safety in Meriden, Connecticut. He's world-renowned, having done exceptional work in solving, among other crimes, the woodchipper murders.

That "chief emeritus" term means "retired"; Lee is a hired gun these days, and one of the legal teams that hired him was that of OJ Simpson. Not everybody that hires him ends up putting him on the stand; apparently, if he believes the evidence points to the guilt of the defendant, he won't say otherwise. They put him on the stand during the OJ trial, though, and afterwards, he wrote a book which included a chapter on OJ, and the various pieces of evidence that pointed to someone else killing the two victims of that bloody murder.

It's fairly stark evidence, hard to argue with.

Saturday, the LA Times published a story questioning the value of DNA evidence. It turns out that state crime lab analyst Kathryn Troyer was running tests on Arizona's DNA database in 2001 when she ran across two felons with remarkably similar genetic profiles. The two men matched at 9 of the 13 locations commonly used to distinguish people. However, their mug shots show that they probably aren't closely related - one is white, the other black.

Since then, Troyer has found dozens of similar matches, each seeming to defy impossible odds.

As word spread, these findings raised eyebrows. How accurate are the FBI's DNA statistics? Should the genetic databases be opened to wider scrutiny?

Local cops generally don't like the FBI. They waltz into a crime situation, schedule a press conference, make it look like they've arrived to save the day, since the locals are hopelessly incompetent, and they do more to hamper the investigation than to assist in it. Critics point out that the FBI's much-vaunted "profiling" unit has been responsible for zero, count 'em, zero serial killer convictions.

So what did the FBI laboratory do? They tried to stop distribution of Troyer's findings, and began what the LA Times called "an aggressive behind-the-scenes campaign to block similar searches elsewhere, even those ordered by courts".

The thing is, even though everybody - except identical twins - would seem to have a unique genetic makeup, genetic profiles are not unique. They're just a tiny part of the full genome. Children of the same parents theoretically should match for half of the markers. By coincidence, unrelated people may share them as well. Nobody knows how rare DNA profiles are - and the FBI doesn't want anyone to question their estimates.

In 1903, Will West was sentenced to a term at Leavenworth - but it turns out that a William West was already there. At that point, fingerprints were unknown. The police used Bertillonage to positively identify individuals - measurements of the head and body - as well as mug shots. It turns out that not only did the two unrelated men have effectively the same name, but the same Bertillon measurements - and they looked alike as well. Their fingerprints, however, were different.

It was 1918 before the story was published; undoubtedly, there was reluctance to abandon Bertillonage when there'd been such a large investment in training and in accumulating data. Furthermore, the data was still useful. Although Bertillonage could not be used to positively identify individuals, it could be used to rule out.

And that's the situation we find ourselves in with DNA. What we need for good DNA profiling are loci - points to be compared - that are completely random. Otherwise, it's like saying, "Well, we know the criminal has brown eyes and brown hair." Most people in the US have brown eyes and brown hair, so that data is relatively low value.

However, you're going to find loci that are random; you normally get your genetic makeup from your parents, instead of from a dice throw, and they got it from their parents. In practice, genetics that contribute to the individual breeding live on, while those which do not tend to die out; most loci are going to be predominantly of one type.

While current DNA profiles can be used to screen out individuals, the only way to use DNA to positively identify an individual will be to vastly increase the number of loci used for comparison. That is, if there's an 80% chance of matching at each loci, then there's 1 chance in 9.3 that you'll match 9 out of 9 loci, 1 chance in 22.7 that you'll match 13 out of 13 loci, 1 chance in 265 that you'll match 24 out of 24 loci.

And even if they use 50 loci, and insist on a match at all 50 loci, there's 1 chance in 87,581 that all 50 loci will match. That means in a medium-large city such a Cincinnati, there will be 24 adults with exactly the same DNA profile.

The jury in the OJ case didn't say whether or not OJ killed those two people; they simply came to the quite reasonable conclusion that the prosecution didn't prove that he did. The facts were overwhelming: the police repeated lied under oath, they obviously mishandled evidence including a sample of OJ's blood, and they apparently even manufactured evidence. If I'd been on the jury, I'd have decided he was likely to be guilty, but not proven to be. When I read Lee's book, I weakened some on whether he was likely guilty. Now, I'm really starting to wonder.

There are many government agencies which really don't need to be government agencies. We get our weather, more often than not, from private meteorologists. Fedex, UPS, and DHL seem to be better at delivery than the US Postal Service. Toll roads get us where we're going as well, and sometimes better, than freeways. There are only two really essential services government must perform - protection from without (the military) and protection from within (the justice system.) We cannot afford to frame innocent people, based on DNA evidence that isn't as good as it purports to be.

Other Bloggers On Related Topics:
bertillon - DNA profiles - FBI - Henry Lee - Kathryn Troyer - LA Times - OJ Simpson - statistics - Will West - woodchipper