After a brief recap (and revisiting) of the old/new view mindmap, the second day was dedicated to a bit of ‘Archaeology of Safety’, exploring the origins of safety ‘science’. As part of the preparation the participants had to read some excerpts from Heinrich’s 1931 book. I think this is a good thing because most people in the safety world discuss, dismiss or promote his work without ever having actually read something by him.
Heinrich deserves respect because he was one of the first who attempted to treat safety in a scientific way instead of just relying on rules. We can see this in the way he tried to deconstruct accidents and find patterns/order. A typical ‘scientific’ approach is to create categories. He also formulated a number of axioms of which the three central were discussed:
- Injuries are the consequence of a completed sequence of events
- Unsafe acts are responsible for the majority of accidents
- There are many more unsafe acts than minor injuries and there are many more minor injuries than severe injuries
We were then encouraged to discuss Heinrich’s legacy with a background in our own experiences and the organisations where we work. One may conclude that - although we have progressed - the Heinrichian approach in essence is still the predominant view in the safety world.
In the next exercise we had to find out how Heinrich got his axioms, what data he used, what he did with them and what biases he had in the collection and analysis of data. Part of this is described by Heinrich in his book, part is criticized by people like Manuele. I found the thinking about the biases to most interesting element, including:
- The data is self-selected to blame (a large part was taken from closed case insurance records, so the people filing the claim had to gain something by claiming either the person or the machine),
- The other part of the data was based on records from plant owners and probably biased by the people who wrote them,
- Heinrich only allowed for either a human cause or a technical cause, if a case listed two he would chose one in a relatively arbitrary way, preferring the human cause,
- He only looked at proximate causes (the thing happening right before the accident),
- He believed in psychology as the solution for Safety problems.
A conclusion is that we should take the ‘scientific’ in Heinrich’s approach with a pinch of salt. As JB said: “Heinrich has a lot to teach us today - but be aware of the flaws!”. I think that what Heinrich has done was coming up with some good ‘common sensical’ ideas and the find (or construct) some kind of scientific looking basis for these ideas. Most likely Taylorist views influenced Heinrich’s work. Taylor (1911) sees behaviour as something that has to be controlled:
- Prescribe the method (work is decomposed in ‘atoms’; there is one best method to do the job; the expert is smart, the worker is stupid),
- Select and train,
- Monitor performance,
- Manager plans, workers work,
- For every act of the worker there are many acts of management.
Skipping 60 years we then went fast forward to James Reason and his first book “Human Error”. It was interesting to see that Reason’s work bears clear influences from Heinrich. I have long thought of the Swiss Cheese Model as being about the spaces between the dominos, so I had linked Heinrich and Reason there, but there is a much greater heritage that I didn’t really think all that much before. Finding Heinrich’s legacy and looking at differences was the theme for the next exercise.
There are quite a number of parallels/overlaps between Heinrich and Reason. This may be as simple as the use of words (“unsafe act”) and the domino-like structure of the SCM, but also the linearity of thinking, the idea that solutions were to be found in psychological factors and the notion that it was possible to decompose accidents in a number of sequences and the notion that humans might be a problem instead of a resource (something that Reason would have a revised view on in his book “The Human Contribution”).
Reason differs clearly from Heinrich in that his focus shifts from the direct causes (‘active failures’) upstream towards underlying/organisational causes (‘latent failures’), a greater multi-causality and especially the idea that things are depending upon context.
Some time was spent discussing inconsistencies in Reason’s work, including the ever shifting SCM (according to JB “a cartoonish way of illustrating defenses in depth”) and the disagreement within the Human Factors community about the notion of violations.
Personally I found this Day of the Learning Lab the least ‘giving’. This was partly because I had been through the ‘study and reassess Heinrich’ exercise a few years ago (read some of the results here). Another reason is that it was the part that had a rather negative tone, since it was more about deconstructing and criticising (which of course is important, but I felt we were a bit stuck in just one mode) than finding useful elements and how to use them.
>>> To Day 3
>>> And a reflection on something that happened (and not happened) this day