From hindsight to foresight.
“… is hindsight bias a conclusion in itself…´Lyman #1084:
Bias is a tendency to hold a view which may affect our thinking; thus hindsight bias is not a conclusion unless we consciously choose to hold that view.
If we consider foresight as a process of judgement acting on information, then we have to establish if the information necessary for the judgement was available and known by the judge. Only then can the quality of the judgement be debated. Such judgement involves risk assessment and the determination of an acceptable level of risk – as low as reasonably practical (ALARP); but then who sets the standard.
In many ways this process is similar to that in determining the level of acceptable behaviour in a ‘Just Culture’, and using Dekker’s view – “it’s not the value the line which is important, but who sets it".
Previous A330 ice crystal / ADC / ASI events may have concluded that flight into such conditions was an acceptable risk because of the non-fatal outcomes (with hindsight).
AF447 was an unacceptable outcome which suggests that foresight failed; but the process of foresight was identical with previous events, thus if this is unacceptable, what risk (information) should have been judged.
Differences between previous events and AF447 might indicate a reason for the severe outcome (what), but this only represents the additional risk in that one event. It’s the difference amongst the events before AF447 which might identify the relevant contributor to the risk (why).
One difficulty with this line of thought is that irrespective of what factors are identified and mitigated, there is still some residual risk; it is probable that AF447 fell into this category.
To progress safety the industry requires to take a more abstract view for continued airworthiness (systems thinking), vice the probabilistic based certification view; and will need to apply generic safety nets to catch residual events. Aspects of these were covered by PJ2 @ #1037.