Originally Posted by lomapaseo
Primary cause, probable cause, etc. etc. confuses many readers. However it does not confuse safety professionals (who read right by those words).
If that is so in the environment in which you work then I envy you.
Most safety professionals, not necessarily those in aviation, analyse mishaps with techniques such as «Root Cause Analysis», which is,
as Wikipedia suggests here, a collection of techniques (I would put «techniques» in quotes; I don't think much of most of them). Take a look at what Wikipedia says are the «General Principles of Root Cause Analysis»:
Originally Posted by Wikipedia on Root Cause Analysis
1. The primary aim of RCA is to identify the root cause of a problem in order to create effective corrective actions that will prevent that problem from ever re-occurring, otherwise known as the '100 year fix'.
2. To be effective, RCA must be performed systematically as an investigation with conclusions and the root cause backed up by documented evidence.
3.There is always one true root cause for any given problem, the difficult part is having the stamina to reach it.
........
Notice in particular Item 3: there is "one true root cause".
(I will spare the forum another few hundred references, since most can use Google.)
And, most obviously, opherben, who self-identified as an aviation safety professional, didn't «read right by those words» either; indeed, he suggests they are very important! I think you will also find investigators at the NTSB who think there is such a thing as a «primary cause» sitting in the facts of the matter, rather than in their own selection procedures.
And if you look back at the discussion in this very forum, you will find Contributor A talking about the pilots not minding the store and saying «that is the real issue here», and Contributor B talking about the anomalous behavior of the autothrust as a consequence of a single point of failure, and proposing that as the «real issue». And A and B are thinking they are disagreeing with each other. The most obvious way of interpreting that disagreement is that each thinks heshe is proposing his selection as «primary».
So I don't buy your contention that people «read right by those words».
Just to be clear, I don't think there is anything inherently wrong with focusing on one, or a few, causes. Indeed, I sometimes think it is necessary. When talking to the airline, for example, I would likely have more to say about crew behavior than I would about automation behavior, for obvious reasons (connected with your second selection criterion below). But, as I said, when one does so it is important to make one's selection criterion explicit.
Originally Posted by lomapaseo
The meat in a report is the linkage of the contributing facts with recommendations for prevention.
That is just one way of selecting the «meat». For many of the people who deal with accident reports, including insurance companies and their lawyers, the «meat» in a report is anything which allows responsibility to be proportioned amongst those involved in the accident. Although some investigators claim this is anathema to them, a greater proportion of investigators work on this very function. Indeed, this is a very important social function, for international agreement specifies that the airline is strictly liable for any accident. When, say, features of the airport, or the air traffic control environment, or the aviation system, or the airplane, are causally involved then many people would argue that the cost should be apportioned amongst them, and indeed this is what is done. Who thinks that Bashkirian should pay the entire cost of the Überlingen accident, when the investigators have identified failures in the sociotechnical system of air traffic control, as well as inconsistencies in the various applicable guidelines of how TCAS is to be used? No one I know.
Originally Posted by lomapaseo
In my initial read of the posts above I sensed that the words "primary cause" was simply assigned to the last link in the chain of swiss cheese.
In other examples of the use of the words, it might refer to the most effective link in the chain to assign preventive action that addresses other combinations as well.
Yes, those are two different ways in which one might assign a meaning to «primary cause». Best to be explicit about which one one might be using, no?
Your first criterion, though, runs into a problem. If one uses the «Swiss Cheese» model, then it might be tempting to think, as apparently you do, that the layers are ordered, so that there is a «last» one. But I don't think Jim intended them to be ordered. It is just an artifact of the picture. Indeed, the ATSB used the «Reason Model» for well over a decade, and they neither pick a «primary cause» nor linearly-order the causal factors.
If one uses the Rasmussen Accimap model, as the ATSB now does, the causes are divided into classes, but they are not ordered. They are also not ordered in the Rasmussen-Svedung approach which Leveson uses in STAMP (used by NASA). Every level in STAMP is modelled as a feedback control system, and causal factors are identified (I put it crudely) as missing feedback loops. Neither are factors ordered in the SHEL model. They are not linearly ordered in Why-Because Graphs (used by Siemens RA and MT) either, although they are partially ordered.
Where they are ordered in is Layers of Protection Analysis, used in the process industries. That is because one works there with inherently dangerous behavior (stuff at high temperatures and pressures, and reactions, which is why in these domains one defines an accident as «an uncontrolled release of energy») which is mitigated using specific added protection functions, so it fits with the system architecture. That is not the way things are in transportation accidents. No one uses LOPA there.
Concerning
Originally Posted by lomapaseo
...the most effective link in the chain to assign preventive action that addresses other combinations as well.
Would that this could be done. It is a research problem to find a way of doing this systematically. I know a very large engineering company who has been trying to do it for almost a decade. They haven't solved it; I don't know whether they have given up.
I won't respond in this forum to any reply. I am very happy to continue discussion of these issues, with you or anyone else interested in them, but please rather by email than here. I am no PJ2 and don't want to be the target of responses such as those of jcjeant to my comment.
PBL