The definitional problem of hazard definitions
I guess you could say that clarity in the key concepts in engineering standards is an essential element for their effective use. Ideally any standard should contain a precise, concise, complete and clear set of definitions to reduce the likelihood of errors. So why in the standards governing the safe design of systems is there so much difficulty in pinning down such a fundamental safety concept as what actually is a hazard?
Today there exist multiple, overlapping and inconsistent definitions of hazard. as well as vociferous disagreement in the safety engineering community as to what a singular definition should be, see for example the Peter Ladkin’s initiated ‘What constitutes a hazard’ thread on the York safety critical mailing list archives (2000). Hazard has been variously defined as including or being; an event, system state, condition, set of conditions, physical situation, occurrence (either singular or in combination), occurrence in combination with an environmental event or state, that is a source of harm, real or potential that may (alt. inevitably) lead to an accident or cause injury, illness or death. Good grief!
The first point that we should realise is that ‘hazard’ as a definition is not a scientific concept that can be defined relative to an absolute empirical standard but is actually a philosophical concept the same as ‘justice’, ‘truth’ or ‘art’. When we ask ‘what is a hazard?’, we are actually engaged in a philosophical pursuit not a scientific one. The pursuit of such philosophical definitions for key terms is also a time honoured tradition, some of the earliest being found in the Plato’s dialogues (428-347 BC). Western philosophy is also littered with the unsuccessful efforts of philosophers to provide a singular definition of such terms. A philosopher proposes a definition of X, a counter-example is provided to refute it, a new definition of X is proffered and so the cycle repeats.
So how do we deal with the problem? In Philosophical Investigations (1953) Wittgenstein tackles the definitional problem by proposing that what we are really doing when we attempt such definitions is looking for a singular set of attributes when in practice there is an overlapping set of features where no one feature is common to all. Attempting to firmly define a concept such as ‘hazard’, with a necessary and sufficient set of features is therefore an ultimately futile effort. Wittgenstein termed this concept of overlapping features ‘family resemblence’, giving as his most of quoted example the family of games. Games may be physical, mental, team or singular, competitive or cooperative, played for money or not, and so on. But, as a whole, we intuitively understand that games form a family where each game resembles another to a greater or lesser extent.
So does this fuzziness mean that we are lacking in our knowledge of what a hazard is? Not according to Wittgenstein. His point is that such a view reflects a failure to understand how our language really works. Humans don’t normally set up crisp bounds around concepts, instead our usage of terms such as ‘game’ , or ‘hazard’, is inherently flexible with the meaning of a term defined by how we use it or explain it, and this is all the meaning it can have. Any attempt precisely define such a term as ‘hazard’ is simply an exercise in establishing an artificial boundary rather than revealing any hidden truth.
All very interesting, but so what? The first logical consequence of this is that if we have a standard that includes any philosophical concept (and safety standards are full of such philosophical terms as safety, risk, loss and hazard) we automatically introduce the problem of family resemblance. Secondly where any standard singularly defines such a philosophical concept, this definition simply represents a local rule or subset specifically developed to suit the purpose of that standard and should not be construed as an absolute and global definition. Needless to say the idea that safety standards rest upon such a philosophical conclusion is somewhat disturbing, perhaps we need to look further than a list of atributes to get a handle on what a hazard is.
1. Wittgenstein, Ludwig (2001). Philosophical Investigations, Blackwell Publishing.