Real world, virtual risk

Who dares wins - and also often loses, as investors in the Channel Tunnel have been discovering. But Peter Bernstein puts the emphasis on winning: "Higher risk should in time produce more wealth, but only for investors who can stand the heat."

The story he tells, he tells very well, but it should be called a story of risk with significant omissions. The missing parts can best be described in terms of three categories: directly perceived risk (climbing a tree, driving a car to the Channel Tunnel); quantified risk, probabilistic estimates of failure (for new vaccines or the reinforced concrete in the tunnel); and virtual risk, in which the scientists and statisticians do not know or cannot agree (the likelihood of a terrorist planting a bomb in the tunnel that will kill hundreds of people).

The management of the first has probably changed little since our ancestors climbed down from the trees. The dangers we face are different but we still respond to them intuitively in ways programmed into us by evolution.

The management of virtual risks is guided by belief, conviction and superstition. These risks remain the realm of the gods. Like the computer programmer's virtual reality, they can be "real'' or completely imaginary. Virtual risks are products of the imagination that work on the imagination. Such risks are usually labelled "uncertainties''. But we do not respond blankly to uncertainty; we impose meaning on it - meaning that has no firm basis in science. And our responses continually change the world to which we are responding. Virtual risks cannot be determined actuarially. They are inherently unknowable because they are the product of an infinitely reflexive regress. Investors, polluters, beef eaters, travellers, terrorists, and those responsible for security are all guessing about the responses of others, to their responses, to the responses of others. It is not clear whether the recent fire in the Channel Tunnel was a result of accident or sabotage. But it is clear Eurotunnel's estimate that a fire that will kill 50 or more people will occur only once in 600,000 years is fatuous. When fully operational, the tunnel could have more than 10,000 people in it at any one time; and what will happen in it will be a result of people behaving unpredictably in response to the behaviour of tunnel operators who are, in turn, trying to guess what they will do.

Against the Gods does not address the first category of risk in the book and a more important weakness blurs the boundary between quantified and virtual risk. It is essentially the story of the achievements of the risk quantifiers and their success in displacing the gods. Bernstein elevates these achievements to a defining role in history: "The revolutionary idea that defines the boundary between modern times and the past is the mastery of risk: the notion that the future is more than the whim of the gods and the men and women are not passive before nature."

In his conclusion this mastery becomes the shaper of progress: "The central theme of this whole story is that the quantitative achievement of the heroes we have met shaped the trajectory of progress over the past 450 years ... these heroes have transformed the perception of risk from chance of loss into opportunity for gain, from fate and original design to sophisticated probability-based forecasts of the future, from helplessness to choice."

Bernstein devotes considerable space to an inconclusive discussion of uncertainty. For coping with it he recommends the methods of his heroes: "Under conditions of uncertainty, both rationality and measurement are essential to decision-making.'' But it is advice without practical content. If we don't know, we don't know. Sophisticated probability-based forecasts of the future are only as reliable as the assumptions upon which they are based. They are useful only to the extent that the past is a reliable guide to future. Bernstein argues that his heroes - Pascal, Gauss, Galton and Kenneth Arrow to name but a few - have enlarged the domain of risk under scientific management, and reduced the domain of the gods. But the uncertainty within which we all live our lives remains boundless. The heroes appear merely to have shrunk infinity.

Or perhaps enlarged it? Might the increased command of science and technology over nature be illusory? Might pharmacologists and genetic engineers be creating new problems faster than they are solving old ones? Such questions have no agreed answer. They lie in the realm of virtual risk.

The recent fire in the Channel Tunnel provides a topical test of the achievements of the risk quantifiers. Eurotunnel's Safety Case is a 295-page testament to the art of quantified risk assessment. It concludes: "It has therefore been possible to perform a deterministic safety analysis of the system and to use the potential risks which have been identified as the basis for Quantified Risk Assessment.'' Although no one died, the physical damage, and the consequent disruption of the service, were not anticipated in this assessment.

All said, this is a fascinating book. But, at its centre is an unresolved ambivalence; Bernstein is a risk enthusiast on the verge of losing his nerve. After applauding the triumph of probability theory over the gods of ignorance and superstition, he closes with Keynes quoting Locke: "God has afforded only the twilight, as I may say, of probability, suitable, I presume, to that state of mediocrity and probationership He has been pleased to place us in here.'' It is an intriguing note on which to end a book devoted to the defence of hubris (the book's title is a definition of hubris). Is it offered, one wonders, to propitiate Nemesis, who invariably punishes such presumption?

John Adams lectures in geography, University College London.