I worked with a client who believed that, because they had run their fleet in much the same way for many years without a significant safety incident or accident, they were de facto operating in an inherently safe way. However, shortly after I started and somewhat to their surprise there was a runway overrun - nothing to do with me I assure you! No-one was hurt but it was categorised in the official investigation as a serious incident.
It turned out that they had been flying all those years with a marked exposure to runway overrun risk due to ineffective procedures, inadequate training and a patchy understanding of the implications and dynamics of high-speed rejected take-offs. They had just been lucky...
And that is one of the frustrations for anyone trying to determine the relative risk exposure between one operator and another by using statistics - history really doesn't tell you anything useful. As we know, probability is intrinsic to most systems of risk assessment but a statistical probability of say, once in 50 years, is meaningless in the short term because chance (or luck) is also in the equation. That event could occur twice in the first year and then not again for a hundred years, or conversely it might not manifest itself for entire generations before suddenly appearing as a rash of accidents. If you bet repetitively on 32 in roulette you know statistically that it should come up once in 32 bets; you just don't know when or in what distribution. Accidents and serious incidents in commercial aviation have become so rare that they are almost all 'black swans' to some extent - most include an element of random and unanticipated human activity. Statistical probability no longer applies. It is only a knowledge of the underlying behaviours, both individual and organisational, that could really help us understand how close to catastrophe an operation resides - and most operators don't share that kind of information.
It turned out that they had been flying all those years with a marked exposure to runway overrun risk due to ineffective procedures, inadequate training and a patchy understanding of the implications and dynamics of high-speed rejected take-offs. They had just been lucky...
And that is one of the frustrations for anyone trying to determine the relative risk exposure between one operator and another by using statistics - history really doesn't tell you anything useful. As we know, probability is intrinsic to most systems of risk assessment but a statistical probability of say, once in 50 years, is meaningless in the short term because chance (or luck) is also in the equation. That event could occur twice in the first year and then not again for a hundred years, or conversely it might not manifest itself for entire generations before suddenly appearing as a rash of accidents. If you bet repetitively on 32 in roulette you know statistically that it should come up once in 32 bets; you just don't know when or in what distribution. Accidents and serious incidents in commercial aviation have become so rare that they are almost all 'black swans' to some extent - most include an element of random and unanticipated human activity. Statistical probability no longer applies. It is only a knowledge of the underlying behaviours, both individual and organisational, that could really help us understand how close to catastrophe an operation resides - and most operators don't share that kind of information.
No comments:
Post a Comment