Errors: causes, consequences and management methods in occupational health and safety

"Human error" has been studied in the field of occupational health and safety (OHS) since the late 19th century. Although widely questioned, minds are still marked by Heinrich's research, which quantified (very precisely) at 88% the proportion of accidents that were attributable to errors and other "dangerous acts" by the worker performing the task, 10% being due to dangerous mechanical or physical conditions, and 2% being simply unavoidable.

This view has been challenged since the end of the 20th century, but is still very much present in the beliefs of the general public, and even OHS professionals. This article sets out to turn this logic of the human being as the "weak link in an otherwise smoothly functioning system" on its head, by presenting the mechanisms underlying individual and collective behaviours that make humans fallible and, by extension, the systems that depend on them.

To get away from this paradigm, we'll start by avoiding the expression "human error" in this article, for two reasons:

• the first is that it implicitly implies "human error" in the singular, and therefore an individual vision of error which, as we shall see, is incomplete and offers very few solutions;

• the second is quite simply that the expression "human error" is a pleonasm of the same order as "climb to the top". Errare humanum est, error is unique to humans, and only humans can make mistakes. Of course, other living beings - living organisms in general, as well as certain artificial systems endowed with cognition - also react to changes in their environment, and are liable to adopt objectively sub-optimal behaviours that we might call errors.

But these entities perform pre-programmed behaviors, whether innate or acquired through direct learning, with a given capability. The cuckoo's victim birds make no mistake when they feed the cuckoo's chicks at the expense of their own, just as the DNA polymerase enzyme, responsible for DNA replication, makes no mistake when it incorporates nucleic bases into the new strand that do not correspond to the original gene: both perform predefined behaviours with evolutionarily optimal capacity.

Even if, as we shall see, human beings can also act in this "automatic" way under certain conditions, error can only be human, since it implies a notion of choice and the ex ante ability to judge these choices and their consequences in relation to a maximum (intolerance to error) or an optimum (capability assumed on the basis of explicit or implicit cost/benefit analysis).

So to err is human, but it remains shameful. This apparent paradox, which in the end is not so, so often do human beings...