RE: Safety Definitions of Hazard and Risk



Date view Thread view Subject view Author view

John McDermid (jam(at)it.uq.edu.au)
Fri, 19 Jun 1998 08:20:32 +1000


Chaps and Chapesses, I don't want to get embroiled in great long discussions - I have other things I need to do more urgently, but Peter Ladkin says: > John McD thinks it's impossible to give precise definitions of hazard > and risk, but he has offered no argument for that view. (Note that > to show something is possible, it's sufficient to exhibit an example, > but to show something is *im*possible, you have to give a general > argument.) > So I felt I ought to respond directly. I took the above as axiomatic, however I think it is easy to make things explicit. There are two main factors: 1. Acceptability of risk is a societal/psychological issue not a purely technical one, i.e. in some circumstances we willing (voluntarily) accept alternatives (do things) which have a higher probability x severity (risk) measure than others. In fact we do this all the time, e.g. flying to Australia when I could have stayed in York, deciding to overtake a car, etc. I'd also refer you to Nancy's erudite discussion here. 2. In practice, almost all (perhaps all) states of complex systems are hazardous to some degree. Thus we are always at risk to some degree. As hazards depend on environmental factors, not just technical/system failures, the degree to which we are at risk varies with time. For example, with civil aircraft, a disproportionate number of the accidents occur in the final phases of flight (no sick joke here, I mean final wrt intent). I don't have the figures to hand, but its about 50% of accidents in the final 5% of the intended journey time. Both these mean that its a judgement what I choose to call a hazard for a particular system. We can give guidelines, but different people will, in general, make different judgements about what is a hazard, and what is an acceptable risk (and certainly different stakeholders will). One can think of other factors, e.g. that the hazards might genuinely be different for different stakeholders - maintenance crew as opposed to operators, however I think the above two factors are the critical ones in showing that there are no "ultimately precise" definitions here. I think this is a general argument. I would expect a precise definition to mean that there was a right or wrong answer, on which suitably informed parties would agree. As acceptability of risk is judgemental our definitions and analysis must be so (I LIKE nuclear power plants; I can see several GWatts of coal-fired power plants from my house with the right climatic conditions; nasty dirty things which emit carcinogens, etc., the nuclear plans in England are a long way away, and the prevailing wind is from the South West, and I think they've been over-engineered, and ...). I don't think this means we shouldn't define terms, etc. but we should be realistic about the degree of precision we can achieve - certainly I don't think we can reduce such things to logic (and I would interpret Nancy's recent mail as supporting this view). John John McDermid Temporarily at the SVRC, University of Queensland e-mail (as usual): John.McDermid(at)cs.york.ac.uk


Date view Thread view Subject view Author view