Prasidhi's Share: The Halo Effect

Thank you for sharing this great episode Prasidhi (and WELCOME btw!!). It’s a terrifc first share, so much to think about. And I really liked your point about how the corporate environment changed the way the main protagonist processed risk…
What particularly stayed with me was this idea that we tend to judge the decisions people made based on how things turned out, rather than what they actually knew at the time. And how uncomfortable we are with the possibility that sometimes bad things happen even when good people are doing their best. We seem to need to find someone to blame, or some intention behind what happened.
I also loved the point about “epistemic humility”; being more comfortable with not knowing, and a little less confident in the conclusions we draw, especially when we're working with incomplete information.
It made me wonder how often I do exactly what the episode warns against, without even realising it!

Forgive me, but this article troubled me, at first I could not quite understand why but I think my discomfort is due to the fact that the examples given do not, for me in any case, subserve the argument that is being made around epistemic humility or Húbrin mísei (Shun hubris) which are indeed very important principles. I hope I can make a coherent case for my view.

A very useful term used in risk management is a"never event"which is something that should absolutely never happen but did. The response to this is what is called a root cause analysis. where, once the facts are ascertained. every step in the process map is subjected to the recurrent question"Why?"The usual aim is to learn and prevent recurrence unless there has clearly been negligence.

With this preamble I would suggest that the unexpected death of a human being is a never event which requires immediate detailed investigation to determine causality. I would suggest that this is a fixed responsibility and failure to do so constitutes both moral and legal malfeasance.

In the first case I would suggest that a car bursting into flames after being rear ended with the ensuing horrible death of the occupant or occupants constitutes such a never event. To require more death to occur without thorough investigation of the first event seems to me to be a gross failure of the duty of care and so, even given the corporate environment, the application of a process map that is utterly inappropriate given the severity of the occurrence, constitutes, to my mind at least, criminal negligence. To frame such occurrences in terms of profit and loss and for it to be accepted as reasonable discourse seems extremely troubling to say the very least.

In terms of the second case I would suggest that there were several criminally negligent actions which fulfilled the legal requirement of mens rea (that a reasonable person would feel guilt under the circumstances in the case of driving while inebriated) and actus reus ( that the act itself is criminal as is the case in shooting a red light)

And so I do not believe that either of these cases can be used as examples to illustrate the very important concept of epistemic humility given that both of these individuals made grave errors of judgement which made them directly culpable for avoidable deaths and to my mind cannot be exonerated or excused even by the most circuitous reasoning or with the benefit of hindsight.

I do hope this makes sense,......

Postscript: Thank you Prasidhi, I found this very thought-provoking indeed. I very much look forward to having difficulty getting to sleep in response to what I am sure will be equally engaging future posts!

Hi Ricky,

If I understand your argument correctly, you’re saying that every accidental death should be investigated and corrective actions taken. This is under the assertion that every accidental death is a “never event”, something that should never happen. Have I stated this correctly?

Investigating accidents and formulating recommendations for future avoidance is in fact the operational method for the airline industry. The industry has also expanded the incidents to be investigated from accidents to “near misses”. As a result, air travel is extremely safe.

In contrast to the airline industry, where professional trained pilots are in charge of professionally maintained multi-million dollar aircraft in a highly regulated industry, the private operation and maintenance of motor vehicles is subject to not just human error, but a much wider degree of basic competence than that of the airlines. Accidental deaths are to be expected, despite the huge improvements in safety, accident avoidance, and survivability made by the auto industry and road designers.

Assuming that the analysis done of competitive vehicles is accurate, the Pinto’s exact failure mode (fire/explosion after being rear-ended) occurred at 2.5 mph higher speed in other vehicles, yet these other vehicle models did not become the butt of jokes and the subject of lawsuits. For other vehicle models, these accidents were judged to be “par for the course”, an unavoidable outcome of millions of imperfect humans. Is it worth spending $30 million to make the vehicles slightly safer? What percentage of accidents would have been mitigated by that $30 million?

Or does this come down to the judgement that the Pinto was slightly worse than its peers?

Thank you so much Barry, let me try to formulate my view in terms of the line of reasoning that you have taken.

Most motor-vehicle accidents that cause mortality can have the reasons behind them explained easily with a accident investigation on the scene. I would suggest that individual accidents which cause mortality and which cannot be easily explained by comparison with the expected outcome of that accident type should be interrogated until a reason for such outliers causing unexpected mortality is ascertained. It will always be possible that no cause will be found but it is quite likely that if there is a cause, as there was here, it would be discovered. The use of comparative data is thus to my mind completely irrelevant and, much worse than this, might be used as an excuse not to interrogate such accidents and in this way avoid the cost incurred by finding a problem which needs to be rectified. The indication for detailed investigation should, in my view, be on the basis of the index accident itself.

In summary then, my belief is that any fatal accident, the cause of which cannot be explained and is not in keeping with the usual outcome of that accident type should be interrogated and that no further statistical rationalisation is required to justify this. I would perhaps go so far as to suggest that legal statute, if it is not already present, should be created to ensure that this is the case and the final arbiter of whether such assessment should occur would be the initial police investigator.

I do hope this makes sense. I look forward to further conversations or a continuation of of this one if you feel it warranted.