On risk analysis and absentmindedness

[jenk]

This interview transcript illustrates some of the chicken-and-egg factors in creating systems that prevent errors. But this one bit on root cause analysis caught my eye...

I think the term ‘root cause analysis’ is an unfortunate one. It implies that there’s some real root, there’s some cause back there which set the whole thing going like a chain reaction. But of course, there’s never one cause. So I don’t really like the term, but I do like what it implies, which is to go back from the immediate sharp-end people, from the immediate things that went on, and ask questions: ‘What was it that provoked this person to do this particular unsafe act? And what were the decisions upstream from that, that left those people for example, short-handed, or with inadequate tools or equipment? What decisions were going on?’ say on the Board level, in the tension between protection and production. But it’s a finite process.
- Jim Reason, Emeritus Professor of Psychology at the University of Manchester.

This is something I've run into with RCAs. Either mgmt is looking for one simple thing to fix and doesn't want to deal with the overall system, or (in the shrinkwrap world) the documentation necessary may not exist. Le sigh.

As a bonus, from the same interview, on the Columbia accident:

Because if you go back to, for example, to the Columbia accident report that came out last summer, the Columbia spacecraft that crashed on re-entry over Texas, the Columbia Accident Investigation Board went back through all the decisions that were made to send it off. They suffered from, I think, a kind of outcome bias, because there was a terrible loss of 7 astronauts, somehow there must have been a terrible blunder to have set them off. But actually, if you go back to the things that determined the decision, it was the decision which I’m sure we’d make again, namely, there had been 113 shuttle flights, all of them had had foam damage of the kind which penetrated the wings of this spacecraft and caused the burn-out. They were under schedule pressure, it was a low-level scientific mission, you know, there was no good reason not to fly it, and once they realised that the wing had been hit, there was not an awful lot you could do about it. But my point is this, that investigators have the hindsight bias, because to us in retrospect, a particular event seems like an inevitability, but to those who are engaged in it, those who only have foresight, they don’t see the convergence at this particular event.