I agree with Doug. Most of what was mentioned in Lentini's LMU paper were attacks based on the topics required in 1033 (fire chemistry, fire science, thermodynamics, electricity, etc.).

One of the other things I forgot in my initial post was the recommendation that fire investigators conduct interviews AFTER they process the scene. This is completely unrealistic. While biasing information can be produced during an interview, it doesn't have to be, based on the questions that are asked. In my opinion, one needs to conduct an interview in order to determine what items may be present within the scene, or what changes/actions occurred leading up to the event. Ed Roberts presented this topic at the ITC in 2019. In one of his examples, he processed the garage, which was the area of greatest damage, for two days. When he later talked with the FD, he found out that the fire was actually in the house, but a FF was injured, so all suppression operations stopped to help the down FF, allowing the fire to get into the garage. Two days spent in a completely wrong area. With the fires that I usually get involved with, if we used this methodology, we could dig for weeks before we talked with potential witnesses. This is obviously not a reasonable tactic.

I agree that there is no such thing as an error rate of zero. In fire investigation I don't think people are concerned about a instrumental/equipment error rate. People are calling for the error rates associated with an examiner and the methodology they use (which are likely two separate things). So how do you develop an error rate for those? Unless you send a significant number of fire investigators through a known test case, there really can't be an error rate, at least that I can come up with. Most studies in my opinion have too small of a sample size to apply a reasonable extra to the entire industry. I listened to a graduate student's dissertation a few years back where he based his opinion that the identification of artifacts on wiring (arc melting vs fire melting) was invalid, based on a sample size of about 30-60 subjects. And this was over the course of a three-year period. That can hardly be enough to be statistically valid.

I'm curious to see where things lead from here with the new edition of 921 hitting the street. There are some significant changes that I believe will negatively impact the industry, particularly the changes to arc mapping. I am curious how things will play out in court when experts don't follow the current edition because arc mapping has been lumped in with fire patterns - many experts disagree with this concept, so the courtroom will be an interesting venue to see these challenges play out.