A New General Accident Theory
We need and seek a general accident theory. Accidents, injuries and deaths occurring at rates which are a major result of apparently random confluence of human actions and errors, despite our repeated and costly attempts to understand and “fix” or eliminate the causes. This is all too complicated to describe easily when we do not know or cannot write equations for the workings and failings of the human mind.
An apparent learning curve exists that causes the rate of the outcomes to fall exponentially from an initially high level as experience is accumulated with the technological system, thus following a Universal Learning Curve (ULC)  which reaches an irreducible minimum whilst not attaining “zero defects”.
This paper is an exercise in the adoption and adaptation of the principles and methods of statistical physics to the measurement and prediction of accidents and the contribution of human error. The idea here is to find the simplest physical theory and systematically obtain the most useful results. The use of physical concepts in an arena dominated by studies of human psychology and behavioural analysis may at first seem an unlikely avenue of theoretical development. But by applying physics methods to the enigmas of human behaviour, we can use analogies that are very powerful. This model agrees with and confirms the macroscopic ULC behaviour observed for technological systems, and the large contribution from human error.
KeywordsPartition Function Human Error Total Experience Error Occurrence Learning Hypothesis
Unable to display preview. Download preview PDF.
- Duffey, R.B. and Saull, J.W. Know the Risk. Butterworth-Heinemann, MA, USA, 2002.Google Scholar
- Jaynes, E.T. Probability Theory: The Logic of Science. Cambridge University Press, 1st Edition, pp. 343-371, Cambridge, UK., 2003.Google Scholar
- Sommerfeld, A. Thermodynamics and Statistical Mechanics. Academic Press Inc., pp. 214-219, New York, USA, 1956.Google Scholar
- Sherwin, C.W. Basic Concepts of Physics. Holt, Rinehart and Winston, 1st Edition, pp. 308-338, New York, USA, 1961.Google Scholar
- Duffey, R.B. and Saull, J.W. Error States — Proactive Learning, Forgetting and Elimination Using Safety Management. Proa, International Conference on (Probabilistic Safety Assessment and Management (PSAM 7), Berlin Germany, 2004.Google Scholar