A new treatment of low probability events with particular application to nuclear power plant incidents

Abstract

The disquiet engendered by the occasional unexpected catastrophic failures of major scientific innovations held to be assuredly safe is impeding essential technological progress. While quantitative technical assessments can improve a design and enhance reliability, they neither give dependable prior revelation of the causation of these low probability events (LPEs), nor do they allay the public concern aroused by them. The ‘New Treatment’ attempts to resolve the ‘Zero-Infinity Dilemmas’ posed by the threat of these LPEs in a philosophical approach which rejects mathematical idealism in favour of the engineering pragmatism that is characteristic of the scientific method. The LPE is seen as a rare singularity in the stream of ‘Event-noise’ of mundane, untoward incidents that afflict industrial systems. Engineering inspection which has evolved as a regulatory mechanism can be effective in reducing this ‘noise’ and the competent inspector is able to recognise the ‘loss of resistance to failure’ which so often comes before an LPE. Despite that an intractable residual risk remains, but this can be made societally tolerable if there is a justifiable conviction that safety is defended by trustworthy engineering. In an age marked by an increasing complexity and sophistication in its science and technology, the engineer is emerging as the proper central decision maker in such matters, standing athwart pure administration and theory on the one hand and expertise and practice on the other. Moreover, evidence that his intellect may be peculiarly adapted for such a role is coming from recent findings in neuropsychology. The Nuclear Installations Inspectorate is cited as an exemplar of the engineering inspection needed to apprehend those human fallibilities in design and application to which most catastrophic failures of technology are due. However, such regulatory systems and the assessment functions associated with them lack accountability and, as an interpretation of Goedel's theorem suggests, cannot assess their own efficiency. Independent, disinterested appraisal assisted by Signal Detection Theory is offered as a remedy.