"With regard to process of elimination in science; consider a problem like SARS or COVID - something is killing people and a sample is sent to have genetic sequencing done to figure out what might be making people sick. Is it a virus? A bacteria? Some fungus or parasite? (There is no evidence to support any of these hypotheses, yet we can form them anyway...Could it be a virus? Is it, maybe, a bacteria?)“

You have to “rule in” before you can “rule out.”

How did the hypothesis go from “something is killing people” to a virus or bacteria? Without data or evidence, there are hypotheses that can be considered without data or evidence to explain “something is killing people.” Murder? Car accident? Mass suicide? Etc.… Why SARS, COVID, Virus, Bacteria?

The answer may be that there is data that was available to be used as evidence. You examined a body, there were documented symptoms, other relevant and reliable data was available. You can consider any such hypothesis without evidence, but there is ultimately little value in such. The value is in the collecting of more data to try and find evidence in this specific case that allows the formulation of a valid hypothesis of what reliably did happen, as opposed to what could have happened. You do not do this in a vacuum, but you use, your education, knowledge, skill, and experience to help with efficiency.

I can consider a murder hypothesis and then I go and try to collect data that might provide evidence of murder. A good way to focus the collection of data in tractable problem-solving manner. If I do not find any evidence of murder, I consider other hypotheses based on my education, training, skill, and experience. I am not eliminating a hypothesis because I never formulated it with evidence to begin with. Again, you must “rule in” before you “rule out.”

Inductive reasoning is a step in the process of applying the SM, but it is not the end point. It must be followed by deductive reasoning to reach a reliable determination. Again, using the SM does allow for both forms of reasoning, but does not allow for only inductive reasoning OR the step of deductive reasoning before the step of inductive reasoning.

The use of K-9s in fire investigation might provide an analogy. Let’s assume for this analogy that the use of a dog is representative of inductive reasoning and the use of a lab test is more representative of deductive reasoning. What is the dog used for? A fire scene presents an infinite number of samples that can be taken as data. Not a tractable problem. We use a dog to identify a more efficient and tractable means to collect samples. The use of the dog is producing data because it is information that can be documented and verified, but it does not provide evidence since the data is not relevant and reliable to the issue of the reliable presence of an ignitable liquid. I can consider the hypothesis of an incendiary fire and use the dog to try and collect more data that might yield evidence of such, but I still do not have evidence to formulate a hypothesis of an incendiary fire.

I can then proceed with more tractable and efficient data collection through the submission of the collected samples to a lab for confirmation of the presence of a non-indigenous ignitable liquid. This data may be relevant and reliable in this context to be used as evidence to formulate a hypothesis of an incendiary fire in a specific incident. Now I can try and collect more data to be possibly used as evidence to try and disprove this hypothesis using the SM. This is deductive reasoning.

Inductive reasoning before deductive reasoning is a reliable application of the SM. The use of deductive reasoning before inductive reasoning is not.

“The sample is thrown into a gene sequencer - the human gene sequences are eliminated (the human genome has been sequenced so, really smart people can see that "these" sequences are human and can be eliminated as "normal") What is left is something else. Eliminating known bacterial genome sequences. Eliminating known fungi and protozoa sequences and comparing what is left to known virus sequences may allow investigators to narrow the pathogen to a viral source. (Further tests can be done to confirm that hypothesis.) To arrive at that point, did they not go through a "Process of Elimination"? Or was what they did not science?"

One can never “confirm” or “prove” a hypothesis. One can only try and disprove a hypothesis with evidence.

Further tests are for trying to produce data and perhaps find evidence that disproves the hypothesis that was formulated with evidence. While trying to “eliminate a hypothesis” and trying to “disprove a hypothesis” is arguably the same process, the “process of elimination” is a special case of the application of the SM. It is only a reliable methodology under a specific set of circumstances. More specifically, where there is a finite and known set of outcomes like the multiple-choice exam example that is commonly used. The investigation of fires is an “open-ended” problem. EOP, for reliability requires a closed system, not an open system.

As previously discussed, there are multiple general outcomes that can be produced by the application of the SM. If you have a valid hypothesis that is uniquely consistent with the available evidence, you can reach a reliable determination. This does not require that you “eliminate” all hypotheses until only one is left. In fact, there can be multiple alternative hypotheses that can be formulated with the same evidence. If there is no evidence to disprove these valid hypotheses, then they remain valid hypotheses. In some cases, there may not be a hypothesis that is uniquely consistent, yet none have been eliminated with evidence. “Undetermined.” If one of the hypotheses is uniquely consistent with the available data, then you have reached a reliable determination.

Not “science.” How do you “rule out” a hypothesis that you have not first “ruled in?” Forget about an infinite number of hypotheses that can be generated without the need for evidence. Let’s assume for argument that you have a number of hypotheses that were generated without evidence. You then are able to “eliminate all but one hypothesis.” What evidence did you use to eliminate all but one hypothesis?”

A typical response has been “there was no evidence that the hypothesis happened in this case, so it was eliminated.” So, the remaining hypothesis, as a product of the “process of elimination,” was not eliminated because there was no evidence?” How is this a reliable methodology that your determination was not made with any evidence in a specific incident? How is this not circular logic that renders the methodology unreliable in this application?

Douglas J. Carpenter, MScFPE, CFEI, PE, FSFPE
Vice President & Principal Engineer
Combustion Science & Engineering, Inc.
8940 Old Annapolis Road, Suite L
Columbia, MD 21045
(410) 884-3266
(410) 884-3267 (fax)
www.csefire.com