Franzl;
Jim Reason's Swiss Cheese model, extremely useful, has provided a good insight into human factors but is being challenged in the way any theory is challenged - through increased research and new knowledge, some of it made possible by a powerful computing capability which did not exist at the time the model was introduced.
Alongside (and not in place of!), Reason's notions are those of some superbly-insightful writers such as Sidney Dekker, John Stoop, Charles Perrow, Nancy Leveson, certainly PBL (Why-Because Analysis), who has contributed here in the past and others who are taking "systems theoretical" approaches (to describe it broadly...they may disagree!). Each are worth the trouble in looking up and reading, just as Reason is, for an understanding of this approach. Such an approach is (again!), entirely blame-free; - rather, it attempts to find things out.
Let us examine two notions. This isn't unique and is said in my own words. Others will have expressed these notions differently.
One way to think of an accident is a series of elements or things, which then interact (or don't interact). The focus is upon "things" as (philosophically) solid items with a fixed nature, which then interact with other "things" with their fixed natures which are then portrayed as "causes", which have "outcomes". It is perhaps a mirror of the way the western world approaches most things...in a Cartesian manner, or a mental model that looks at the world as "mechanically linked" in terms of cause and effect, (note the singular form of these words!).
Because human factors deals naturally with the way humans see their world, another way to examine an accident sequence (and view the world!), is in terms of relationships...that which occurs "in-between" things. It no longer sees "things-in-isolation" but instead sees primarily relationships...what goes on in-between things and how relationships change those things. So, an organizational system, which can be printed out as the usual "org-chart" isn't a "thing", it is a living organism which materially affects the behaviour of people within such a system. Therefore, the notions of "cause" and "effect" are significantly changed. So much so, that analyzing a series of events from a "Cartesian" view, (as one might do a physics experiment), cannot work and a better model is needed. Charles Perrow first broached this notion in 1984 in a ground-breaking book entitled, "Normal Accidents". Perrow is emminently worth reading and listening to.
Diane Vaughan (sociologist), wrote about the Challenger accident in a way that analyzes the organizational structure of NASA - the relationships between engineers and managers - there was very little analysis of "things" in Vaughan's work.
Many of us here understand this stuff intuitively but many others do not, and are perhaps a bit stuck in a Cartesian world view in which the notions of local cause/effect = blame are legitimately/automatically attached to any understanding of what happened in the accident and why. Concepts like blame and accountability are legal terms and the legal discourse is quite different than the discourse of the safety process, which is being discussed by
Safety Concerns.
This applies to AF447 in ways that have already been very well described and written about here by those who know this aspect of flight safety work. The importance of the sociologists' work in this cannot be over-emphasized, but nor can the engineers' work be set aside. The two need to work closely and this is where the field is tending - a systems theoretical approach.
In the July - September 2011 issue of
ISASI Forum has an excellent article on this approach - it is the paper presented by Sidney Dekker and John Stoop at the
2010 ISASI conference entitled, "
Limitations of 'Swiss Cheese' Models and the Need for a Systems Approach". I tried the link to that issue of the magazine and it doesn't work yet, but it does work for the entire
Seminar Proceedings Vol 14, and you can find their presentation there.
I believe we will learn far more about AF447 using this approach and indeed this has already been put into practise in many of the posts here, but there is always more learning! The benefit of such robust process is it provides a solid basis upon which pet theories and recurring themes may be judged in terms of relevance and consistency as well as their contributory value towards understanding, and where indicated, safely set aside.