What Is an ‘Error’?
The media reaction to this tragic event was one of disbelief, and yet, aircraft pressurisation problems, while not common, do regularly occur. The NASA ASRS database contains 171 reports relating to the Boeing 737 alone, over the period 1994-2004. Of these reports, 58 were deemed critical. Boeing reported that the rate of pressurisation events was 8.5 per million departures for the B-727, 2.7 for the B-737, 4.9 for the B-757/767 and 6.3 for the DC-9/MD-80 (AAIASB. 2006). For the aircraft pressurisation system to work, three elements need to be correctly configured. The engine bleed valves must be open, the cabin conditioning pack valves must be open and the pressurisation control panel must be correctly set. Issues with meeting these system constraints provide us with the opportunity to clarify what we mean by an error. The Threat and Error Management framework underpinning LOSA defines errors as ‘actions or inactions’ that lead to ‘deviations from organisational or flight crew intentions’ (authors italics). The definition is normative in that it suggests that the error is a ‘deviation’ and that a third party - the organisation - had an expectation of the way the work will be conducted. In short, it is possible to distinguish between the correct version of events (organisational intentions) and an incorrect enactment of plans (crew actions). Reason (1980) defines an error as an ‘action not as planned’. Here, we see that error is manifested in behaviour which departs from intention. Both definitions share the idea that the ‘error’ is an observable state, condition or action that is not what was expected. Importantly, ‘error’ is seen as an outcome. Reason’s work triggered several attempts to develop error classification systems or taxonomies. However, the behavioural precursors to the erroneous outcome are actually of more interest than the final observed output. Hollnagel (1983) suggests that, if work has an inherent variability as a result of the constant need to adapt to circumstances, then, error is an integral part of that work. We, therefore, have no need to categorise elements of work as being somehow ‘other’. Furthermore, if apparently erroneous performance is integral to normal work then we also have no need for pathologies and taxonomies of error. The argument developed in the previous chapter leads to a similar conclusion. Performance flows from underlying knowledge structures, their application to solving tasks and their subsequent control. Errors are simply symptoms of a performance: neither a cause of a problem, nor an outcome.
We saw in Chapter 1 that skills, rules and knowledge structures can be categorised as either declarative or process knowledge. While I might see a skill being performed, though, rules and knowledge are only observed indirectly through the behaviour of individuals (although they may be accessed through questioning). These memory structures represent the inputs to activity. The outputs can be considered, according to Reason (1980), as slips, lapses and mistakes. A slip is the inadequate control of motor performance. When you knock over a glass, drop an object thrown to you or trip over your feet, these all count as slips. An incorrect read back of an АТС instruction would also be a slip. Lapses are, simply, forgetting to do something. We have already seen that an interruption can lead to the subsequent forgetting to complete a task and prospective memory failure is the forgetting to do some planned future action. A mistake is a decision error: what you chose to do was not sufficient for the task in hand. Slips and lapses map onto encoded skills in that they flow from flawed enactment of behaviour. Mistakes flow from a flawed application of rules and knowledge. Violations represent a subset of mistakes (or ‘intentional non-compliance’ in LOSA terms). Whereas mistakes reflect action choices that fail to fully map onto task demands, a violation occurs when an individual elects to act outside of prescribed bounds. The language around violating is pejorative but, as we will see, violations are often well intentioned. I now want to return to the Helios accident to explore these categories in more detail. I will use other aircraft pressurisation events from around the time of the Helios crash to elaborate on performance types. My goal is to show that the Helios accident was not really an exceptional catastrophe but a tragic example of the probable outcome from a normal performance.