Maybe Someone Stole the Tent

How often have I said to you that when you have eliminated the impossible, whatever remains, however improbable, must be the truth?
Sherlock Holmes, The Sign of the Four by Sir Arthur Conan Doyle

Would that it were that simple. 

Oh, Holmes's logic is impeccable enough.  Right up to the point where it's rot.

There is, and really I shouldn't need to elaborate on this, one basic problem.

Eliminating the impossible isn't the same as eliminating all but one theoretical possibility.

There are also lesser problems: Mistaken determinations, lack of information, false information, and so on.  We don't know enough, never will, to be able to eliminate all possible error or even, dare I say, all impossibility.

Life, like the universe itself, is far too complex and random.  Our own capacity is too limited.  Many things are possible - remotely possible sometimes, but possible nevertheless.  Occam's razor, like the related principle of parsimony, is a useful tool for cutting through the clutter, but it proves nothing.

The specific subject today is arson, more properly, it's fire which is sometimes arson but sometimes not.  (See Willingham, Cameron Todd.)

This morning, Scott Henson, at Grits for Breakfast, who's been following fire investigation in Texas and the gymnastics of the Texas Forensic Science Commission has gone through to avoid actually investigtating and reporting just how problematic it is, wrote about Dee J. Hall's article, "Burning Questions: Old Assumptions Hard To Put Out," from the Wisconsin State Journal. 

John Lentini, a prominent fire investigator and one of the harshest critics of the current state of fire science, said some of the probes amount to little more than "witchcraft and folklore."
He cited a 2005 test designed by the U.S. Bureau of Alcohol Tobacco and Firearms in which fire investigators were asked to identify the general area where two test fires were started in separate rooms. The fires were extinguished less than three minutes after achieving "flashover" — the point when, Lentini says, "a fire in a room becomes a room on fire."
Each time, just three of the 53 investigators got the area of origin right, and it was a different three each time, Lentini said. Subsequent tests have produced similar results.

Let's focus on that for just a minute.

ATF did a test.  That's not me, not some bleeding heart, sleazoid, criminal defense lawyer or plaintiff's lawyer (much fire investigation is done by insurance companies desperate to prove arson so they don't have to pay out).  That's a federal law enforcement agency.  They're in the business of trying to put people behind bars.

So they did a test.  53 fire investigators each examined two fires.  That's 106 examinations.  6 correct results.  6 of 53 investigators were wrong half the time.  The other 47 were wrong all the time.

When a fire investigator says the fire started at location X, you can bet that it started somewhere else.  You'll be right more than 95% of the time because that's how often the professionals, the experts, are wrong according to the law enforcement folks who rely on them.

But wait, as Ron Popeil would say, there's more.

Hall's article actually focuses on the investigation of a fire at J.J.'s Pub in rural Marquette County.   That fire was investigated by Deputy State Fire Marshall Joseph Siehelr and investigators hired by the insurance company to prove arson.  And so they did.  Along with V patterns, which ATF says don't mean a thing, and origin point examinations which we've seen are wrong 95% of the time, Siehelr used the Sherlock Holmes method.

Perhaps most important, Siehelr used a form of reasoning known as "negative corpus" in determining the blaze was an arson. Siehelr testified he and the experts paid by Awe's insurance company ruled out all accidental causes in their area of origin, "which leaves no other possible conclusion than for this to be incendiary."
The National Fire Protection Association's Guide for Fire and Explosion Investigations is considered the gold standard in the field. It has been revised for 2011 to add strong language saying such reasoning never should be used.
"It is improper to opine a specific ignition source that has no evidence to support it even though all other hypothesized sources were eliminated," the guide states. In those cases, it says, the investigator must label the fire as undetermined.
Denny Smith of Kodiak Fire and Safety Consulting of Fort Wayne, Ind., a national expert in using the process of elimination in fire investigation, said it's "pretty clear" Siehelr's reasoning "meets the criteria of what shouldn't be done."

That's a nice way of saying it's all bullshit.

Oh, the fire at J.J.'s Pub might have been set.

It's not impossible.

But that's no reason to think it's the truth.

Richard Wiseman, a professor of psychology at the University of Hertfordshire, working with the British Association for the Advancement of Science conducted a study to find the funniest joke in the world.  This came in second.

Sherlock Holmes and Dr Watson were going camping. They pitched their tent under the stars and went to sleep. Sometime in the middle of the night Holmes woke Watson up and said: “Watson, look up at the stars, and tell me what you see.”

Watson replied: “I see millions and millions of stars.”

Holmes said: “and what do you deduce from that?”

Watson replied: “Well, if there are millions of stars, and if even a few of those have planets, it’s quite likely there are some planets like earth out there. And if there are a few planets like earth out there, there might also be life.”

And Holmes said: “Watson, you idiot, it means that somebody stole our tent.”