The Construction of Safety-II

The purpose of the deconstruction of Safety-I was to find out whether the assumptions are still valid - which in turn means whether the perspective offered by Safety-1 remains valid. The deconstruction of Safety-1 showed that the phenomenology refers to adverse outcomes, to accidents, incidents and the like that the aetiology assumes that adverse outcomes can be explained in terms of cause-effect relationships, either simple or composite and that the ontology assumes that fundamentally something can either function or malfunction. 

As is argued above, an increasing number of work situations are becoming intractable, despite our best intentions to the contrary. One of the reasons for this is, ironically, our limited ability to understand completely what we do, in the sense of being able to anticipate the consequences of design changes and other 

We can only specify the work in detail for situations that we understand completely, but there are very few of those. Because we cannot specify work in detail for the rest, human performance variability is indispensable. There are, in fact, only very few situations where performance adjustments are not required, even if every effort is made to establish an isolated and fixed work environment, for instance sterile rooms in neonatal intensive care units, or ultra-clean rooms in a unit where electronic chips are manufacturered. Even in a strictly regulated military operation such as the changing of the guard in front of a royal palace, the soldiers will have to adjust to wind, weather and intrepid tourists. The more complicated or less tractable a work situation, the greater the uncertainty about details and the greater the need for performance adjustments. (That is, more will be left to human experience and competence, and less to the capabilities of technological artefacts.) 

The ontology of Safety-II is consistent with the fact that many socio-technical systems have become so complicated that work situations are always underspecified, hence partially unpredictable. Because most socio-technical systems are intractable, work conditions will nearly always differ from what has been specified or prescribed. This means that little, if anything, can be done unless work - tasks and tools - are adjusted so that they correspond to the situation. Performance variability is not only normal and necessary but also indispensable. The adjustments are made by people individually and collectively, as well as by the organisation itself. Everyone, from bottom to top, must adjust what they do to meet existing conditions (resources and requirements). Because the resources of work (time, information, materials, equipment, the presence and availability of other people) are finite, such adjustments will always be approximate rather than perfect. The approximation means that there is inevitably a small discrepancy between what ideally should have been done, or the perfect adjustment, and what is actually done. Yet the discrepancy is usually so small that it has no negative consequences or can be compensated for by downstream adjustments. This is so regardless of whether the discrepancy is found in one s own work or in the work of others. 

It is somewhat ironic that performance adjustments must be approximate because of the very conditions that make them necessary. In other words, if there were sufficient time, information, etc., in a work situation, then performance adjustments would not be required. (This does not rule out that they might happen for other reasons, such as boredom or the human creative spirit.) But as it is, they are required because the resources are not completely adequate, which in turn means that the adjustments cannot be perfect or complete but must remain approximate. 

---