Accidents are surprising. That which previously was trustworthy, suddenly no longer is. And the mismatch between expectations and actual performance calls for an explanation.
To learn from failures organisations and societies institute accident investigations. The task of such probes is traditionally to find out what happened, why it happened and what can be done to prevent such events from recurring. The hope is that in the rubble, somewhere, there is an identifiable source of trouble – the seed of destruction, the root from where it all came. Only if such origins can be located there is a possibility to prevent similar events from reoccurring.
Guided by this conviction the search is on for deviations, the abnormal, anything that can be repaired: a misbehaving employee, a bad manager, a procedure that wasn’t followed, or perhaps a technical malfunction.
However, the universe does not cooperate with attempts to localise a distinct cause. Every event is preceded by other events. Every part is connected to multiple other parts. And every condition is the result of other conditions. The number of contributing connections quickly mushrooms beyond anything localisable.
The more intense the search for the origins of trouble, the more the explanatory power gets diluted. The search is bound to eventuate in everyday normal and approved structures, components, or goals. Accident causes (if they exist at all) are unstable buggers residing somewhere, but nowhere specifically, between debris and normality. Should we arrive at the beginning of something – at a root cause – it is rather an expression of a capacity to discredit or ignore complexity, than a true or objective depiction of the world as it is.
But what if we weren’t looking for faults? What if man-made normative ideas about how the world should work did not guide ‘learning’? What if investigations didn’t start with the assumption that someone screwed up, or that something failed? Perhaps we would see how people at all levels of organisations struggle to meet competing and conflicting goals. Perhaps we would see how systems every day face variation, change and unexpected disruptions. Perhaps we would realise that success happen not because of detailed work instructions and procedures, but despite, and that in fact people are ingenious contributors to safety. Perhaps we would see that people and organisations struggle to make multiple ends meet with suboptimal resources at hand.
And what if we were not so focused on parts and components? What if there was a shift toward understanding how a phenomenon is a consequence of a set of relationships between components – that the interesting things lay not inside the components but in the qualities between them?
If we want to improve safety, then it is a fragile approach to identify and control for weak or faulty parts. Instead, safety improvement efforts should be about strengthening the connections that make up the node in focus. If a part is in trouble, then open up the paths that can nourish it, improve its connections with the surrounding system, make it grow. Don’t throw it out.
Maybe then we can learn to cope with reality as it is, as opposed to how we want it to be.
Interesting article, Daniel. Perhaps it would be useful to think about the ’causes’ of accidents as having rhizomatic characteristics; having no discernible beginning (or end), with multiple antecedents and limitless (often unseen) potentialities.
“If a part is in trouble, then open up the paths that can nourish it, improve its connections with the surrounding system, make it grow”
Daniel – Keep encouraging posts that use metaphor. Eventually ones will emerge that make sense to everyone. The H&S establishment tend to be resistant to metaphor but but, in my experience, use inappropriate metaphors (is the ‘accident triangle’ shape truly representative of the actual ratios between events of different secerities!)
The ‘growing’ idea has not yet been fully explored. I like the concept of creating an environment in which peope become better at what they do by growing stronger. What are the organisational conditions that encourage employees to grow stronger in what they do?
Andrew
Andrew, I believe ‘parts’ and ‘components’ are also metaphors, ie of a machine. I’d like to argue that any word is a metaphor, signifies a part of reality, and that reality does not dictate what words it favours.
Organisations as living networks is quite intriguing to think about. I think that there is quite a bit of content on this website that could hint what makes people grow and improves the resilience of such a network. Self-organisation, engagement, participation, diversity, creativity, recognise everyone contribution (empowerment) In other words, using more of that we see that nature relies on to make complex systems thrive.
Yes, someone explained to me that ’causes’ are remains from outdated philosophies. Describing a flow events connected to an accident, creating a story that makes sense and bring about meaningful learning opportunities is the best we can hope for.
Daniel – I was coming to believe that it would not happen in my lifetime, but maybe just maybe it will. It sure as hell won’t come from the regulators. It might come from industry realising that helping the people in an organisation thrive leads to improved profits.
Has Australia become the epicentre of humanitarian approaches to H&S? Were the origins of this in Lund with you, Sidney Dekker and Eric Wahren?
Andrew
Actually I think the tree, with roots, is a pretty good metaphore…
But not in the way it is used ‘traditionally’.
The tree does not become a tree out of the nothing. It starts pretty small as a seed and grows and grows. First it is a little branch and continuous to develop. As it develops it becomes bigger and more complex. Sometimes (in gardening) people help the tree to grow ‘properly’, the cut out old branches and remove other trees in the area so this particular tree might flourish to the big old oaks from the stories of old…
But just one root does not cause the tree, it is the growing in time as the whole becomes more and more complex, which we at some point in time see as a tree…
Just try to make the exact same tree, it is not as simple as it seems. One might need the exact same genes and reproduce the exact same environmental circumstances to get the same tree.
I’m afraid I’m not really making myself really clear, but what I’m trying to say is that a tree metaphore might be used properly to show complexity, whilst everybody might think the opposite.
“However, the universe does not cooperate with attempts to localise a distinct cause. Every event is preceded by other events. Every part is connected to multiple other parts. And every condition is the result of other conditions. The number of contributing connections quickly mushrooms beyond anything localisable.”
Yet improvement actions that target specific factors can in fact lead to overall systemic improvements in safety, because systems can often be understood and points of leverage can be found… and where you find a point of leverage, a root cause is probably lurking nearby.
At this point, I expect someone will try to tell me that causes are constructed, not found, and that I’m applying an outdated, Newtonian, deterministic worldview (both being popular slogan around here.) They’re only partly true. For instance, when I do a root cause analysis, I investigate the problem/event and the system that experienced it to find as many relevant factors as I can — including their interfaces and interactions. Then I look for the combinations of factors, interfaces, and interactions that seemed to act as drivers or as risk adders.
After that, I look at other features of the system that experienced the problem. I try to understand which system features created the “found” causes. I analyze the system further, applying both analytic and synthetic thinking as required to build up an understanding of the system. How does it actually work? How was it intended to work? How do we want it to work in the future? What features are keeping the system from working the way we want it to? What features are create, promulgate, and fail to mitigate the risks? Which of these features can be affected by actions taken within the system?
What about systems that are truly chaotic, or that become chaotic when pushed beyond a certain level of criticality? In the former case, I really have to ask why such a system was allowed to exist in the first place, especially if it had the potential to harm or kill. In the latter, I need to figure out what excites the system to that critical point and the kinds of disturbances that can cause an avalanche.
To summarize: I use investigation to find causes and I use analysis/synthesis to refine them. Both deterministic and stochastic factors are sought, as needed, to explain the system causes that created, nurtured, and triggered the problem and/or the factors that didn’t prevent, mitigate, or warn. In other words, the causes are both found *and* constructed, in the same way that observations (found) can lead to scientific hypotheses (constructed).
Bill,
Nice reflections! I struggle with the idea that anyone can find something through mere observation. How can you know that you have a ‘finding’ in front of you, and that it isn’t ‘noise’? What makes the finding jump out?