ACE: Assess Cost of Error

published in The Risk of Freedom Briefing, Issue no.5 October 2000 1

Many years ago, when I was fighting the Department of the Environment on its anti-environment road schemes, I was faced with COBA, Cost-Benefit Analysis. Cost-Benefit Analysis was used to show that the proposed road was good value for money, and therefore ought to be built.

The figures on which the Cost-Benefit Analyses were based were phoney. For the Cumnor Bypass the traffic figures were accidentally doubled. For the third London Airport a Norman church was valued at 2,000. Small amounts of drivers' time were added together to produce a saving to which a high price could be assigned. The cost of land expropriated was whittled down to well below the market price, and the value allowed for loss of amenity was zero. It was hardly surprising that Cost-Benefit Analysis acquired a bad name.

I criticized it on other grounds too. Apart from its crude utilitarian assumptions, it assumed an infallibility on the part of the government which the governed found hard to believe. I used to urge the Inspectors to consider the possibility that officialdom was wrong, and then to assess the cost of error. They should consider a range of forecasts, and then calculate the cost of being wrong. At one extreme, supposing the motorway were built and the oil runs out, then all the money spent would have been wasted: at the other extreme, suppose it were not built and traffic in fact rose as much as or more than the Department had predicted, then there would be severe congestion, waste of time and waste of temper. These and other possibilities needed to be looked at, and consideration given to how easy, and how costly, it would be to redeem the mistaken decision.

The idea of assessing the cost of error was ignored by the Road Construction Units of the Department of the Environment, but now the Precautionary Principle has come in its stead. But the Precautionary Principle has, in the hands of some of its advocates, been made too rigid. Instead of urging us to ask certain questions, it has insisted on a particular answer---that we should imagine the very worst possible outcome of any measure, and then decide not to undertake it. But it is one thing to think officialdom may be wrong, another to suppose our imaginations are good guide to the future. We need to assess risks before we can identify possible errors, and consider not only what may go wrong but how it might be rectified. It is worth working through the questions we need to ask of three modern issues: nuclear power, genetically modified crops, and global warming.

What are the possible errors? Nuclear power stations might not be as safe as the authorities make out; nuclear waste might prove a greater hazard than seems likely at present; fossil fuels might turn out to be far more plentiful than we suppose. Each of these possibilities needs exploring. They should not be swept under the carpet. But neither should they be taken as serious without scrutiny. Nuclear power stations can explode: Five Mile Island and Chernobyl show that the possibility is a real one. Can it be guarded against? Yes. Can it be guarded against so as to give an absolute guarantee? No. How bad would another Chernobyl be? Could we palliate the damage? How many people would be killed, how much land poisoned. At the end of this we have guestimates of the risk and the possible damage. Different people will assess them differently, but they are formulated figures we can argue about. They can be compared with the expected costs of the alternatives. I, who was brought up in Durham, have always been sensitive to the human cost of coal-mining. I did not welcome the loss of life caused by Chernobyl, but I should set against the possible cost of a nuclear disaster the certain loss of life that continued use of coal would entail. Similarly we need to put figures on the increase of radiation that would come from spent nuclear fuel, and compare it with the background radiation we incur anyway and the increase of radiation that results from burning fossil fuels. A breakdown of power supplies needs also to be thought about. How many people would die if hospitals no longer had electricity? Again there will be disputes, but again these can be argued about. Sometimes we get agreement to within an order of magnitude---a person willing to expose himself to cosmic radiation on long journeys by air looks silly if he objects to one hundredth of that dose from spent nuclear fuel.

The genetic modification of crops alarms people. The natural stock of genes might be ousted by artificial ones; some genetically modified organism might have unexpected malign properties; and of these some organism might be able to reproduce itself uncontrollably. These mights are different. The extinction of species and loss of genetic variety have happened, but through human greed and improvidence rather than as a natural corollary of genetic alteration. Few people want to eat old, traditional English apples when they can get Cox's orange pippins easily. But it is perfectly feasible to continue to grow old-fashioned varieties just in order to preserve them. Crops created by artificial modification of the genome are no different in this respect from crops created by selective breeding of the sort that has been practised for centuries. In either case there is a natural tendency for the new to drive out the old just because the new is---for our purposes---better. We need to remember that our purposes are limited, and there may be features we have overlooked that will tell none the less. Every new brand of wheat evolved by selective breeding in the last half century has after a few years proved susceptible to rust or mildew; monoculture is inherently dangerous, but the massacre of the English elms by the Dutch elm disease was not due to artificial genetic modification. The same point can be deployed to parry some, but not all, oBFIQYom unexpected malign consequence. Traditional innovations have sometimes proved disastrous---the introduction of rabbits to Australia, the breeding of killer bees in South America. Whereas with nuclear fuels we are dealing with known potentially adverse effects, with biological organisms we are afraid of unknown ones, and it is impossible to make estimates of them. But we still can guess. If we were modifying the small-pox virus, we might well end up creating a lethal pathogen, but if we are trying to alter wheat so that it can, like legumes, fix nitrogen on its roots, the possibility of the modified wheat seed being a fatal virus is vanishingly small. Still, should we not play absolutely safe? Here the cosmic ray aircraft argument applies. Risks we are prepared to run in our ordinary way of life are not risks we must avoid when considering new courses of action. We cannot reasonably ban all forms of genetic modification while still countenancing genetic modification achieved by selective breeding. If the vanishingly small possibility of the genetically modified wheat seed being a fatal virus is too large for us to face, then the possibility of the selectively bred wheat seed being a fatal virus is also too large for us to face: the green revolution was not absolutely risk-free, and if absolute freedom from risk is what we want, then we must forgo all further biological improvement. But that argument is not as broad as some GM enthusiasts make out. There are risks and risks. And as with nuclear power some potential disasters are very great. We need to distinguish between different forms of modification and between different sorts of genome. Plants are not bacteria, and it would be silly to prohibit all forms of agricultural research for fear of its generating a new plague. It would be a bad thing if a rogue plant escaped into the wild, but an altogether worse one if a lethal pathogen did. Different safeguards are needed for different fields of research. And in each case the safeguards need to be publicly arrived at after public discussion in which the worst-case scenarios are fully thought through and critically scrutinised.

Assess the Cost of Error applies differently to global warming on account of the difference of time scale. The actual errors we might be making are comparable. The changes in the atmosphere might be unimportant; the increase in average temperature might be just a random fluctuation; a further increase in average temperature might be harmless. But by the time we know whether these fears are unjustified or not, it will be too late to remedy some errors. If by some gross oversight the seed of a mega giant hogweed were released, it would be a terrible disaster, but we could do something about it. At enormous cost we could seek out and burn every deviant plant. But if the prophecies of global warming are not mistaken, and we suppose they are, we shall not, when we realise our mistake, be able to reduce the amount of carbon dioxide in the atmosphere: whereas, if we believe those prophecies and they are in fact mistaken, the cost, though great, will be one we can carry. Restricted economic growth can be made up for later; we shall have made fewer journeys by car, and spent more time staying at home or in trains; but so did our grandfathers, and they survived. Assess Cost of Error is not a calculus but an approach. It is untidy, because it holds that alternatives should be argued about, and arguments are inherently untidy. There are different risks of different sorts of error occurring, and the costs of their outcomes have to be assessed against different backgrounds. In spite of its awkwardness, it has the great merit of taking alternatives seriously and actually arguing about them, and in this it is preferable to the bland insensitivity of those who do Cost-Benefit Analysis, and the blind terror of those who are prey to unformulated fears of the worst they can imagine actually happening.

1. tel. 01 665 510 325; fax 01 665 510 607; E-mail

Return to home page

Return to bibliography