A 'confluence of interests' supporting debunked arson indicators

For those who've followed the saga over the Todd Willingham and Ernest Willis arson cases at the Texas Forensic Science Commission, this Wisconsin State Journal article ("Burning questions: Old assumptions hard to put out," May 19) helpfully frames the problems with arson investigations without all the politicized death-penalty baggage surrounding the topic here in Texas.

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.

I wonder how investigators at the Texas state fire marshal would fare on such a test, given that they've continued to stand by investigative conclusions based on the same, debunked premises described in the article? Notably, as arson science improved - or more accurately, as actual science began to seep into arson investigation for the first time beginning in the late 1980s - arson convictions declined precipitously:

Since 1980, the number of intentionally set fires has been on "a long-term downward trend," currently accounting for about 8 percent of all structure fires, down from about 20 percent 30 years ago, the National Fire Protection Association reported last year. Roughly half a million buildings in the United States are damaged or destroyed by fire each year, the NFPA estimates. ...

At least some of the decline, Lentini believes, is because investigators are taking a more cautious approach.

The Wisconsin story focuses on a fire in a bar where a deputy state fire marshal named Joseph Siehelr and an insurance investigator concluded arson simply because, as in the Willingham and Willis cases here in Texas, the investigators could not conclusively identify another cause:

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."

Most of the discussion surrounding arson science in Texas has come in the context of the Willingham case, and thus in the context of the criminal justice system and ultimately debates over the death penalty, which Grits has long argued diverted focus from the bigger picture. But most fires don't kill someone or result in the death penalty, and flawed arson investigations also come into play in more workaday cases where an insurance company simply doesn't want to pay a claim. The attorney in the Wisconsin case described noted "a confluence of interests," between insurance companies and arson investigators: "The insurance company is looking to get out from under a policy, and the state fire marshal wants to hold someone responsible for the fire. There was an incredible rush to judgment — a mindset of going onto a scene looking for a crime, not looking for what happened."

Perhaps the best place to attack crappy arson science isn't in criminal court at all. Maybe what needs to happen is some deep-pocketed civil attorney needs to vet complaints over denied insurance claims at the Department of Insurance to find clients - possibly even enough for a class action suit across multiple states - against insurance companies denying claims over arson based on debunked indicators. The Wisconsin attorney is right there's a "confluence of interests" between insurance companies and law enforcement, but unlike fire marshals offices, insurance companies don't have either sovereign immunity or court-endowed qualified immunity, as do fire investigators. Federal courts in the past have accepted "negative corpus" reasoning in such cases, but given the strong language in the most recent NFPA discrediting that approach, the issue seems ripe to be revisited on a much broader scale.