It seems NIST wanted to get the model to accurately depict what was known through the video to have taken place. With their methodology the results were factual data that could be used to analyze what took place and what needed to be done through the code development process to prevent it from happening in the future. Their methodology and process reduced the potential error rate of the final model.

As you are aware of, the error rate (uncertainty) of a particular process of one of the questions that is being asked in the courts. In speaking to a number of people, it is suggested that the error rate (uncertainty) for the CFI calculator being used in the field by an investigator could average around 30%. I disagree with using a fixed error rate. I would believe the error rate would change as a function of time. The longer the fire burns the greater the error rate of modeling but closer to the fire’s origination the less the error rate (uncertainty). I thing, and maybe there will be some guidance with the new CFUcalculator as to the error rate (uncertainty) of the conclusions reached by the program.

Jim Mazerat
Forensic Investigations Group