close x

Forums

All forums solely reflect the views of contributors and commentators. The Royal Aeronautical Society and the Human Factors Committee do not endorse or support these views in any way, express or implied, except as they correspond to the official position of the Society.

Guest  

Welcome Guest, posting in this forum requires registration.

Pages: [1]
Topic: Predictability
Flyinginn
Administrator
Posts: 3

Reputation: 0
Predictability
on: December 1, 2015, 13:52

Following Jean Paries' discussion of the lack of predictability in complex (chaotic) systems I was pointed to Dragon King theory by Dan on the CRM-Devel list. Very interesting: https://en.wikipedia.org/wiki/Dragon_King_Theory


This seems to imply that what cannot be predicted cannot be prevented. I'm not sure that's the case. Going back to Jean's example of weather forecasting, we know that the more data we collect the better the modelling. It may not help with specific predictions, but it can help with the range of required skill sets. This is perhaps closer to game theory: https://en.wikipedia.org/wiki/Game_theory


jonberman
Member
Posts: 1

Reputation: 0
Re: Predictability
on: December 8, 2015, 11:25

I felt that Jean was trying to say that collection of ever more data wasn't going to answer the exam question, and that we therefore need to design systems that are better able to respond to the emergent properties and events, rather than try to defend against them in advance. For me, the flooding in the UK Lake District is perhaps a good example of this (leaving aside the weather forecast metaphor) – the whizzy new defences have been overwhelemed, whilst the presence of those defences has arguably reduced the resilience of the people living there. But the media debate about what we mean by "a 1 in 100 year event" is interesting.


Flyinginn
Administrator
Posts: 3

Reputation: 0
Re: Predictability
on: December 8, 2015, 12:24

I agree with Jon – can't answer the exam question. But I was suggesting that better modelling means more representative conceptualization of the range of skills you could need, rather than how high the flood defences should be. So instead of saying 'well, our analysis says that this height of flood defence is adequate for all normal variation events so that's OK' we might say 'when we run the model with minor variations some outcomes show breached flood defences so let's look at how those play out and what fallbacks we need'.


I keep thinking of Sioux City, where there was no Ops Manual procedure for loss of three hydraulic systems. Perhaps nobody thought it a serious possibility. But there was a way to maintain some directional control which required an unusual skill set that Al Haynes exercised during free simulation time.


My soundbyte would be, you're only thinking outside the box if your box is too small.


Pages: [1]
WP Forum Server by ForumPress | Lucid Crew
Version: 1.8.2 ; Page loaded in: 0.047 seconds.