SAFETY OR RISK?

It is common to talk of the issues of hazard, danger and project failure as matters of 'risk', and in a number of countries there are now requirements that 'risk assessments' be carried out before major new industrial construction projects can be approved. The term risk itself, however, is not an entirely helpful one : it is, of course, a crucial idea in the insurance business, but the rationality used to calculate insurance premiums is only rarely reflected in the way in which people actually behave.

Also, it is by no means easy to deal formally with the risks that are inherent in a given technical arrangement or installation. The documented difficulties of such technical risk assessment (see, for example, Refs 4 and 5) are added to when we examine risk management and safety control from a sociotechnical perspective. Since formal risk assessments of a system can never provide more than a partial view of the hazards: decisions about risk will almost certainly be in error. In most real-life, open-world situations, the information needed to make a good-quality assessment of risk is typically not available.

It is therefore important to ensure that risk prediction is always complemented by strategies for the ongoing control of safety.637 Within engineering design, reliable information may exist about the likelihood of failure of specific, commonly used components under certain conditions-those components that have been frequently used and also often rigorously and repeatedly tested. However, information about the likelihood of failure of combinations of such components, or of larger, more unique subsystems, not to mention the total system itself, is much less commonly available.

Also, while we now know much about the way in which individual people assess and deal with risks (see Chapter 8 in this volume), it is also important to recognize that the perception of risk and hazard is not a private or an individual matter. Individual variations notwithstanding, the dangers that we emphasize and the dangers that we ignore are in large part already selected out for us by the groups and the society to which we belong.' Rather than being an independent factor that can be assessed and precisely specified, most of the dangers that we face are to a great degree interdependent with, or are generated by, the activities of our own groups and our own society.

Those who live by hunting and fishing need to contend with the dangers of the environments into which they place themselves, as well as with the potential hazards of self-inflicted injuries from guns, spears, arrows or trawl nets. Military societies transform death and danger by making their confrontation a part of the destiny of the warrior who should prefer to die bravely and gloriously rather than to survive as a coward. Our own endorsement of a way of life that uses cars and aeroplanes, electricity, plastics, high-rise office buildings and the other trappings of contemporary industrial society brings with it an associated set of risks of traffic accidents, pollution, collision and collapse on a grander scale than has previously ever been possible. We provoke the hazards whose risks we then have to learn to cope with.

Nor is it the case that risks are equally distributed within modern societies : whether we consider general death rates, occupational death rates, the danger from accidents in the home, inability to work as a result of chronic illness, or a range of indicators of general health, it is overwhelmingly clear that it is more 'dangerous' to be poor, since figures for all such factors increase as we move down the socioeconomic scale-'the poor risk more'. It is also a complication that the risks that we can most accurately assess in advance are the ones that are well known and well structured.

Since, as we shall see below, most major system failures arise as a result of ill-structured combinations of factors, whose dimensions become apparent only with hindsight, it is clear that such events cannot be well captured by means of highly exact anticipatory risk assessments. A final caution to be expressed about the idea of risk relates to the way in which attention given to a numerical assessment of risk, which, it is claimed, can be calculated in an impersonal and a detached manner, detracts from or covers over the issues of moral judgement and blame which crop up whenever things start to go wrong. In the design stage of a new project, a risk assessment figure may have the appearance of a precisely and accurately calculated certainty, but in the unfortunate event of a failure in which people are killed or injured-or even one in which they merely lose money-this apparently impersonal risk assessment number has to be interpreted as one element in the multiple issues of guilt, blame and responsibility. On matters of danger and safety and their social implications, we can never take a neutral attitude.