Originally Posted by kr2y5
Based on all the posts so far, it sounds like generally, all failures except for the first one tend to be brought about by a human error. I am led to the conclusion that failures in a cascade are potentially highly correlated, and what makes them correlated is the effect they have on the diver, by causing stress, and increasing the risk of an inadequate reaction. The more the diver has experienced complex scenarios, however, the better the diver handles stress, the less their performance is impaired, and the lower the effect one failure has on the other. The purpose of drilling cascading failure scenarios in training is to expose the diver to these sorts of complex situations. I suppose one could say that the degree, to which the failures are correlated decreases with experience and the number of complex scenarios practiced, and that's what makes the cascade increasingly unlikely. Makes sense.
While it may make sense - its not actually how social/psychological research believes human's interface with the world (at least currently in 2015). Most of what you said is about 30 years out of date in the accident analysis universe. You are thinking as if there is one single trigger or root cause which is not the case. There is always a set of circumstances which all have to align for any failure (big or small) to snowball into what we call an accident. I am not the best person to speak on this topic though and hopefully GLOC will see this and chime in with a better way to describe this.
=============
Okay, I’ll see what I can do! (Thanks for the PM to bring me to the debate as I don’t follow SB too much).
There are two key points I see here: what should be taught and its relevance to the real world, and what happens in the real world in terms of incident causality and outcome. (and a third at the end!)
1.
What should be taught and its relevance to the real world? Training should allow divers to deal with incidents which stress them in a controlled environment but in an unexpected manner. By this I mean that once the basic core skills have been taught so that they are able to be undertaken in a unconscious competent manner i.e. a diver can do a skill like a shutdown, or donate gas without having to recite the steps in their mind each time they are undertaken and maintain control of their buoyancy or trim and wider situational awareness (not task fixated/attention tunnelling). As a consequence of this capability they will have spare capacity to deal with real world failures which can easily become complex if they are not trapped early enough (and they won’t be otherwise the incident won’t happen!). Once this competency happens, the complication factor can be increased until the diver can deal with complicated incidents (2 or more failures) in a conscious competent manner e.g. logic is required to solve the problem rather than intuition. It is not possible to train for all the possible failures, so building blocks have to be introduced and the problem solving created. This problem solving needs to be done in a no-notice manner so that the decision-making is not primed by the diver - they part-solve the problem before it arrives, they also need to be able to monitor the cues from the world to spot the development rather than ‘Bang! 3 failures'. So, introducing multiple failures is a good thing but the chance of them happening in the same manner as incidents happen in the real world is extremely unlikely but real world incidents can certainly be used as a premise for the scenario. There is research out there in the aviation domain that has shown that crews are very competent in dealing with simulator scenarios (spotting the threats and managing the errors) but not so good in the real world doing the same. (Thomas, M. J. W. (2003). Improving organisational safety through the integrated evaluation of operational and training performance : an adaptation of the Line Operations Safety Audit ( LOSA ) methodology.)
2.
Incident causality theory. Modern causality theory has done away with linearity in terms of ‘if this, then that’ because of the complexity of the world we live in, the training we have had, the perceptions and acceptance of risk, along with hundreds of other variables out there. Richard is correct that a snowballing is required where multiple contributory factors are required to almost produce a critical mass of factors and then the incident occurs. Fundamentally, incidents start with the same behaviours as normal operations, and it is the additional of multiple small changes that create that critical mass. The problem is that humans are not effective at at spotting small changes because we pattern match to make quick decisions (Google System 1, System 2). Finally, errors are now being considered as contextual assessments of past effects judged by the severity of the outcome! Complicated? By that I mean you only know an error has happened after the event and what the outcome was. Consider the following example which I heard this week from an anaesthetist colleague and convert that to your own world.
- I like to have coffee before I go to work. I didn't sleep well the night before and I'm running late so to save time I hurriedly fill the kettle, sit it in its power stand and flick the switch. I go off to shave and dress. I come back and the kettle is still cold. I forgot to check that the power cord was plugged into the wall. Result: I don't get my coffee. I can cope.
- I'm pitching for some potentially lucrative contract one morning and I have a critical powerpoint presentation to make. Again I want to make sure my laptop is charged so I plug my power adapter into the laptop before I go to bed. I toss and turn because my financial situation is ropey and I really need the close this deal. Next morning I am again in a hurry and I gather up my laptop to go to the meeting. I meet the client and turn on the laptop. The battery is flat. I forgot to check that the power cord was plugged into the wall. Result: I bumble through and I don't get the contract. Financially painful, but hey it's only money.
- I'm on take for a medical retrieval service and we've just returned from a job. It's been a busy shift and I'm tired. I want to make sure all my equipment is charged ready for the next job, including the portable defibrillator. So I sit it in the charging dock as usual. Three hours later we are woken up for an emergency call. I grab the kit and we go out. When we arrive he patient is in VF and needs defibrillating. I switch on the portable defibrillator and it's flat. I forgot to check that power cord was plugged into the wall. Result: the patient dies.
Which one was the error?
3.
Evidence to show shortfalls in training (or behaviours) This is the crux of the problem as I see it. There is a distinct lack of data that shows what happens in the real world in terms of incidents and their causes. Anecdotally, and intuitively, we would think that there are problems at the training level for many reasons (perception of risk, speed through training, trust in automation, trust in the training system to teach correctly, quality control of instructors/CDs/ITs) but there is no hard data to prove things either way. Even incident reports that are submitted to databases like DAN or
DISMS, do not necessarily capture many of the causal or contributory factors which are down to human behaviour and decision making. This is mainly because the person involved doesn’t even realise they made the choices they did or why they made them. My research is hopefully providing a framework that will allow incidents to be better classified at the individual level but it won’t examine the supervisory or organisational failures as I already have enough of a problem covering individual level issues in a PhD and the agencies aren’t exactly forthcoming with incident data involving their instructors or staff!
So, life is complicated and error management even more so. What I can say is that the data is significantly lacking in terms of incidents and detailed narratives which are honest and complete. Without these, we will not move forward, either at the individual level with divers developing their own skills to be credible, or organisations modifying their training activities because of shortfalls. Unfortunately I think you can blame the litigious nature of the US for that.
If you want to read more about my work, or sign up to my blog newsletter, visit here
Cognitas (diving-related) or my
full-time job
Regards
Gareth
---------- Post added June 16th, 2015 at 06:49 PM ----------
As an aside, this presentation might interest a number of people on here as it concerns the trade off between efficiency (doing it in a timely manner) and thoroughness (doing it as it should be done)
http://www.abdn.ac.uk/iprc/uploads/files/Aberdeen_ETTO.pdf
========
Finally, I should say that you need to be careful with correlation and causation. Lots of incidents are down to multiple failures because the individual can't deal with them, but that doesn't mean lots of failures cause major incidents. You could have lots of failures and deal with them...
Regards
