I recently presented a system safety course over here in Sydney. In one of the modules I discuss the roles and relationships of: System Safety (SS), Risk Management (RM) and OH&S, and how collectively they contribute to the understanding and claims of safety with respect to an equipment or activity, but that they are not mutually interchangeable roles.

Following that module, I had quite a lengthy discussion with one of the attendees who came from a strong RM background; specifically he was operating in the AVRM (aviation risk management) world, the standard for which is based on AS/NZS4360 (now gone international with ISO31000). I had something of a ‘light bulb’ moment when listening to him describing ‘risk’, in particular arguments as to how he believed he could claim to have reduced that ‘risk’.  Following this, I thought would be useful to bounce my thoughts off the forum, to garner alternative perspectives 

I feel I should caveat what I am about to report by saying that this was a discussion with one RM person and it does not transpire that all people from the RM realm think the same, but what he did say certainly resonated with some of the recent discussions here about claimed reductions in risk, particularly as it pertains to road safety and the contribution of speed to that safety.

I recall some time ago that somebody wrote into the forum claiming that ISO31000 was the new ‘safety’ standard on the market. I responded at the time, but was unfortunately unable to follow it up due to work and time constraints, that I did not consider 31000 to be a SS standard, but very much RM. RM, although sharing some common traits with SS, does not fulfil the same role; following my AVRM discussions, that position has been reinforced in my mind 

The term ‘risk’ in 31000 is described as the ‘effect of uncertainty on objectives’ where one of the ‘effects’ can be ‘a deviation from the expected’ (4360 describes it more succinctly as: ‘a chance of something happening’). These ‘risk’ definitions differ markedly from those of SS which exclusively refer to risk as being a measure of BOTH the likelihood and severity, or frequency and outcome (other terms like probability and consequence are also used). What struck me about my conversation was that RM talked about being able to claim that ‘treatment’, effecting some statistical variation in the expected outcome, was considered to have reduced the ‘risk’. He was able to do this because it is a ‘deviation from the expected’, but in SS terms the most ‘credible’  (a consistent SS aspect) outcome may not be considered to have changed (depending whether the ‘deviation’ was sufficient to bring about a change in the outcome i.e. critical to major, or catastrophic to critical – but I would argue that it is a little bit more than a ‘deviation’ at that point); in the absence of a risk matrix or discrete levels of risk or severities, any degree change can be claimed as a ‘reduction’. But the only way you can claim any ‘measurable’ effect for some fractional benefits (deviations), such as reducing the speed limit from 110kph to 100kphn say, is to aggregate it over a sufficiently large population, so that your statistical ‘deviations’ can produce ‘whole’ savings (a lucky few if you will). 

The other concerning thing about the RM approach to risk is that it is possible to ignore one element of risk (the event itself - what a previously referred to as a 1-dimensional approach) and focus solely on the outcome or severity.  This perception is reinforced when reading 31000 as it lists ‘likelihood’ and ‘consequence’ as risk treatment ‘options’ separately (which I interpret as meaning that I don’t have to consider them both together – which we do in SS). The ‘option’ of a 1-dimensional approach is disconcerting (and un-SS) from 2 perspectives: you may be trying to claim a reduction in risk for something that is already acceptably low due to the low frequency of occurrence (but you don’t know, because you haven’t look at that component of risk), or you may be claiming a reduction in risk by effecting the ‘deviation’ where the dominant component of the risk is the frequency of the event itself; in both cases, little or no ‘credible’ change in SS-risk terms, but claimed reduction in RM terms. Another concern (clearly I have many!) is that if you do not consider the whole risk space (SS-risk that is), there is a danger that an attempt to reduce the severity may result in an increase in the frequency; unless you consider both dimensions (the whole risk space) the risk may actually increase.

Andrew stated that he doubted that: ‘that the effect of an 85mph accident will be much more awful than of a 70mph one’ which is something I, and others, agreed with; Nancy put it slightly different when she said: ‘I don't think that arguments involving kinetic energy are going to answer the question’. The reason that I think they, and I, have a problem claiming speed as a credible risk reduction solution is because they are thinking in SS and not RM terms. For SS, tweaking the speed by 5 or 10mph does not alter the ‘credible’ outcome and the risk is not considered to have been measurably reduced, which his why most SS risk reductions efforts end up effecting the incident frequency in the first instance.

Let me finish by saying that I understand the kinetic energy argument and I do not dispute the maths put forward by Michael and supported by Peter, but the issue I have had all along is that, in SS-risk terms, you cannot claim to have effectively addressed (reduced) the safety risk unless you have considered both dimensions (something that RM allows you to do).

[The content of this part has been removed by the mailing list software]