What is risk? Not risk of this disease or that accident, but Pure Risk? It is a sign of the times. The Society for Risk Analysis was founded last year. There is a new magazine for the trade: Risk Analysis, an International Journal. Risk has become a profession. Social risks have long been familiar to surgeons, engineers, admirals, and entrepreneurs. Today everyone can go to a new kind of expert, the risk assessor, who offers an impartial evaluation of risks of any kind. The new expert uses lots of older lore, but history records an official beginning to “modern risk-benefit thinking.” The year was 1969.1

Risk analysis can cater to any sort of hazard, but their profession owes its existence to a relatively narrow band of possible dangers. The social risks that worry us are not a random bundle of frights. You can arrange them in an orderly way to form a triangle. At the top is the unthinkable war. Then come the two demons, cancer and nuclear meltdown. Below these are threats to the environment: foul rivers, increased radiation, acid rain, toxic waste, assaults on threatened species. At the base of the triangle is a motley collection many of which are old standbys: mining coal, smoking, drinking, getting pregnant, climbing mountains, lead, asbestos, saccharine. It is not these at the bottom nor nuclear war at the top that created the new profession of risk analysis. It was the public demons and popular causes. (Only in the past year, after the books under review were written, has total war again become both an American demon and a popular cause.)

What selects a demon or a cause out of all the risks we know? That question prompts the remarkable Risk and Culture, the joint work of Mary Douglas, the social anthropologist, and Aaron Wildavsky, an imaginative political scientist. They tell of a people in Zaire afflicted by the tropical horrors of fever, leprosy, ulcers, and parasites. Only three of their dangers are singled out by this people for worry and intervention, namely, barrenness, bronchitis, and bolts of lightning. Is that trio more curious than our own variable list of demons and popular causes? Our causes are not in general the greatest risks to which we ourselves are subject. Is our choice of a “risk portfolio” also irrational?

A social anthropologist like Mary Douglas is not one to say that stupid people rank fears wrongly. She will look at our society in the way she studies an entirely foreign culture. This book says that popular causes are created by the groups that campaign for them. But to understand which causes will be chosen, we must ask what a group needs in order to stay together long enough to be effective. The chosen cause must be of a sort that helps to maintain the group. Groups of different internal structures will fall apart in different ways, and so will perceive different kinds of risk. So Douglas and Wildavsky conclude that we need a taxonomy of possible organizations for small-to-medium-sized protest groups, and that this will help tell us what kinds of risk will be the focus of attention.

One beauty of Risk and Culture is that it elegantly leads us into a handful of simplistic models of social organization. Then it locates protest or preservation groups thus modeled along the “border” of American society, the places where protest groups spring up. The border pits itself against the “center,” the public and private institutions that hold power. The center is complacent, the border is alarmed. We get quick glimpses of groups like the Sierra Club (which is for wildlife and wilderness) and the Clamshell or Abalone Alliances (who don’t want nuclear power on their turf). The different organizations of these groups, and the risks that they protest, help create each other: if there is no group, there will be no perceived risk; but equally, without a risk of the right sort to hold the group together, there will be no group.

Having defined themselves and their risks, the groups on the border say: let these risks be diminished or removed. The center is challenged. Air must be cleaned at great expense, nuclear power replaced by solar energy. The center counterattacks. Its spokesmen declare that the world is full of risks. Let us study them carefully. Take due precautions but do not exaggerate the one risk that you highlight above others that you have forgotten. When we get costs and risks in full perspective, you will see that nuclear plants are a definite plus, and remember that we have already improved air quality by 30 percent in a decade. At this point enters the risk assessor, who, we shall see, belongs to the new profession created to assist the center in its debate.

Extensive popular agitation about environmental risks began on the “border” about 1965. The profession of risk analysis followed. Acceptable Risk is the best critical exposition of new ways to think about risks. But the title, we are told, is to be used as an adjective. The book speaks only of “acceptable-risk problems.”


Fearing that the term “acceptable risk” will tend to connote absolute rather than contingent acceptability, we have chosen to use it only as an adjective, describing a kind of decision problem, and not as a noun describing one feature of the option chosen in such a problem.

You can hear the skeptical environmentalist muttering: “We’ve been telling you there are problems; now it is your turn to show us that the risks are acceptable.”

That reaction has a germ of truth. The pure risk analyst is the peacemaker trying to invent reasonable procedures to make everyone simmer down, quickly. Fischhoff et al. outline three generic approaches in Acceptable Risk. One is traditional—use the best professional technical specialists you can hire. Secondly, there is the method of revealed preferences. Observe the risks society has tolerated in the past. Assume that we have evolved a satisfactory attitude to risk. Then introduce new risks only if they are no greater than already accepted risks.2 This idea can be spruced up to allow for changing knowledge and attitudes, but our authors prefer a third approach adapted from economics. Essentially you consider the probabilities and utilities of all the consequences that may result from taking each of the choices available to you. Ideally, you could then compute the right decision. The authors candidly describe problems of assessing probabilities and utilities, of sorting out facts and values, and of detecting hidden purposes on the part of the analysts themselves.

The book is a report. It ends with many sensible recommendations. We wryly note how a new profession concludes its list of proposals. It would like to see “an experimenting academe”—perhaps a new university department that draws experts from various fields and also makes some new jobs. It wants to “create a profession of risk management.” In the final paragraph it urges more research on acceptable-risk decisions. Here we catch phrases like “Given the enormous stakes riding on acceptable-risk decisions, our investment in research seems very small”; “bargain”; “enormous expected return on investment’; “Such research could be a good place to invest some of society’s venture capital.”

What is the expected benefit of this research? One benefit would be to “inform workers about occupational risks (so as to enable them to make better decisions on their own behalf),” but that generous gift to the working classes is not the “bargain.” We are chiefly to “consider the cost of a day’s delay in returning a nuclear facility to service.” If we could take just one day off the time needed to decide how to restore the facility: what a gain! The expected benefit is, then, possibly “shortening the decision-making period.” Here is a lesson in semantics. Changing a noun into an adjective can make us stop looking for the right decision, and start looking for quicker ways to make decisions that leave everybody happy—for the while, at least.

Who are the risk analysts? Some risk analysis is straightforward engineering. Having designed your “facility,” you had better get an outsider to draw what are called “fault trees.” This engineer figures out all the possible things that might go wrong, and how several failures might conspire to produce something worse than was bargained for. Or there might be a “common-mode failure,” where three safety devices have one shared component, so that if this component quits, all three devices are useless. Such analysis is often supposed to end with some numbers stating probability of risks and costs of disasters. These are weighed against expected benefits. Even when the numbers mean little the sheer presence of the outsider doing fault trees can help detect design blunders. No one doubts that these professionals earn their keep.

The pure risk assessor is more of a generalist than the engineer. Three authors of Acceptable Risk, for example, are research associates at a firm called Decision Research. The fourth heads the decision-analysis branch of a consultancy, while the fifth is a “a consultant to government and industry on problems of decision and risk analysis.” When we turn to other figures in the field, we find that Chauncey Starr, author of the original risk analysis paper of 1969, is vice chairman of the Electric Power Research Institute. Chris Whipple, now president of the Society for Risk Analysis, is technical manager of the Energy Study Center of that institute.

Starr and Whipple defend the center. Their institute is financed by utility companies. People on the border say it is only a front for big industry. In return Starr and Whipple regard environmentalists as part of the “conflict industry” that increases the cost of making decisions. When the border groups learn that the research for Acceptable Risk “was supported by the Nuclear Regulatory Commission under subcontract from Union Carbide Corporation, Nuclear Division, Inc. to Perceptronics,Inc.,” they must lump the book with other industry presentations. In fact that is just how the grant money is “routed.” The book is an independent assessment, not committed to any approach, and it provides systematic summaries of all the ideas and examples that come from the center. There is none of the scorn of Starr and Whipple for “environmentalists who find conflict an effective means of achieving their social objectives.” Starr and Whipple properly add:


the identification of vested interests opposing many proposed energy projects is not intended to dispute the legitimacy of their role in the conflict. Rather it is intended to point out a popular misconception; that these groups are somehow purer than the advocates of supply expansion.3

The move from “social objectives”—like not being irradiated?—to “vested interests” may strike one as rhetorical, but let’s concentrate on the idea that those who question “supply expansion” are “somehow purer.” Is that an accidental “misconception” that can be avoided? No, say the authors of Risk and Culture, for the idea of purity is in the nature of perceived risk itself.

Douglas and Wildavsky fix chiefly on one kind of risk, pollution. Why should this so polarize activists against industry? It won’t do to say that pollution was worse in 1965 than ever before. Often it was less bad. The last legendary London pea-soup fog occurred in 1962. There will never be another because at that time all but smokeless fuel was banned from the capital. That was not in response to a mass movement, although smoke had long been a reformist concern. Why has pollution been a cause for the past decade in America?

Pollution—the anthropologist has something to say about that. Throughout this century it has been a central concept in describing alien cultures, most of which have some sort of ritual uncleanness. This is not a matter of being primitive, for India is as pollution-conscious a civilization as can be imagined. Mary Douglas herself has done much to give us our modern theory of the subject. She has shown how in different cultures purity rules help to determine who can marry whom, who can buy from whom. Uncleanness excludes some people and preserves others, thus establishing stability within a social group and resisting pressures from outside. The pollution rules of a cattleherding people from Uganda are here used to illustrate this idea. The Hima live on dairy products. If a Hima eats vegetables, he must fast twelve hours and be purged before drinking milk. Otherwise the cows will be contaminated. This and other pollution rules prevent integration with an agricultural people living in the midst of the nomadic Hima, competing not only for land but also to maintain different styles of living. Because of the danger of contamination, it is clear that none of the farmers can keep a cow, thus leaving the Hima supreme in their own kind of culture.

But isn’t Douglas just punning when air pollution and toxic waste are compared to the ritual impurity of other societies? Not exactly. For one thing, what people believe may be what determines features of their culture, regardless of its truth or falsehood. Even if we had worked out completely “value-free” definitions of polluted air, and had true beliefs about what air is polluted, the concept of pollution could still play a central role in the culture. Douglas and Wildavsky began with the question, how do we choose our risks? They suggest that the risks chosen by a protest group are essential to the preservation of the group itself. To see how that works for pollution, we have to understand the authors’ taxonomy of institutional types.

For Douglas and Wildavsky, border groups, such as Friends of the Earth and the Clamshell Alliance, are sects. That is a key idea. A sect is defined as a voluntary organization. As such it tends to fall apart. One way to keep it together is to think of its members as free from evil, and as confronting an impure world outside. Where the center expects the future to be like the past, the sectarian expects that in the short run things are going to get worse. A current Greenpeace ad runs, “During our life-times we may witness the loss of one million species or more.” The center must be warned. “Each sectarian,” Douglas and Wildavsky write, “needs the others’ concurrence in an image of threatening evil on a cosmic scale: for this, the idea of global, irreversible damage serves well.” It matters little to the dynamics of the sect whether it is religious or not. “Either God will punish or nature will punish; the jeremiad is the same and the sins are the same: worldly ambition, just after material things, large organization.” Indeed, secular and religious iconography run parallel. A recent mailing from the Nature Conservancy begins, “Do you want to make a deal with the devil?” A few months ago two young mountaineers in Arizona sat atop the 550-foot smelter stack of the Magma Copper Company, thereby drawing attention to the poison belching across the state. They recall St. Simeon Stylites and all other fifth-century saints who fought for purity by living alone atop columns in the desert.

A sectarian view allows no inequality among members. Such sects have a hard time making decisions. They resort to the lot, or employ leaders who have a duty to remain almost invisible. By a series of all too quick leaps our authors draw the conclusion that such sects must expand. When they cannot do so, their members become angry. Sectarianism breeds schism. Moreover, since members of the sect are pure and the rest of the world is bad, activist sects get into their most intense debate: when is it permissible to use violence against the oppressive outside, the center? Not every sect is sectarian (if we follow the authors’ confusing usage) and egalitarian. Some institutions are hierarchical and not sectarian at all. They offer stability, dignity, inequality, and firm leadership. What they fear is a change in the structure of society. A hierarchical state finds much of its risk is in foreign politics. It also consciously fears disruption from within—dissidents, crime in the streets, and the like. The sectarian view in contrast never expects internal disruption, or schism, because it knows all members are equal and good, and is amazed when a split occurs.

Two old Anabaptist sects are used to illustrate these two kinds of organization. The Amish of Pennsylvania are egalitarian. No member may use modern tools, put rubber on the iron wheels of the buggy, or have a radio. Nature cures are strongly favored. Leaders are chosen by lot. As land ran out, the Amish split angrily into orthodox and more relaxed versions. In contrast the Hutterites of South Dakota and Alberta are rigidly hierarchical. Every person has a status. The best modern equipment is used, and when a member falls seriously ill, the patient is sent to the best available clinics. When a community grows to 150 members, it divides in two. Two communities are formed, and both pack up. The one to go is chosen by lot, thus relieving the sect of making the only divisive decision left.

The classification into sectarian or hierarchical thus provides a way of interconnecting what might otherwise seem a set of bizarre and unrelated aspects of a sect. What, for the sectarian, is to be feared? The greatest risk is the defiling, greedy, and perverted world outside. This part of the border is always subject to schism. That can be avoided by forming new sects or new subsects, each one fighting a hitherto unnoticed form of pollution. And in fact that has happened often enough in the environmental movements.

The hierarchical sects have a quite different sort of worry. Here the Sierra Club provides an example. This old-time conservation group became increasingly hierarchical. When it rapidly expanded under vigorous leadership, the traditional members worried that the enormous expansion would destroy the character of the group. Then “the Sierra Club consistently expressed the typical concerns of hierarchy, according to which the worst eventuality is to change the structure of the society.” In a vote the old-timers won; the leader quit and founded the activist, officially egalitarian Friends of the Earth. This fights the evils of pollution, while the Sierra Club, its institution intact, gets on with its hikes and with saving the wilderness. Much modern technology goes into the more hierarchical organizations, with their professional lobbyists, paid staff, and the computerized mail list which gets money and members who are seldom actively involved. James Watt vastly increased membership of the Sierra Club, but the leadership remained the same and the number of Sierra hikers was almost unaffected.

Risk and Culture has much more to say about such groups, but the pattern should by now be plain. Of course it is only a sketch. The sectarian, hierarchical (and other) institutional forms would have to be validated from a wider range of examples. What we have here is a rich and inviting collection of anecdotes combined with a scanty and ill-tested theoretical map. A lot of the story hangs nicely together with our own untested impressions. The authors of Acceptable Risk speak simply of “society’s” needs as if we were one society. They write, for example, that “when lay people do force their way into hazard decisions, their vehemence and technical naïveté may leave the paid professionals aghast, reinforcing suspicions that the public is stupid.” So they urge that the public become involved in the early stages of potentially risky projects. If Douglas and Wildavsky are right, that won’t work. The public that would be involved will come from the center, while it is the border that enters with the vehemence that is the sign of the pure assaulting the impure.

Both books have an alarmingly unrealistic tone. I noted how Acceptable Risk in the end wants to speed up decision making, rather than ensure that our decisions don’t destroy us. It seems to be saying, if only we could get everyone to agree, the world would be benign whatever we do. Similarly Risk and Culture sometimes hovers near the anthropological fallacy of thinking that everything we perceive is a cultural artifact. Every once in a while the reader has to cry out that some pollution is real. Vegetables in a Hima stomach do not generally make the cows sick, while most of our pollutants do produce either discomfort or disease. Among the many distinctions elaborated in Risk and Culture, this one receives no emphasis.

Moreover, the central feature of our dangers is that they are rapidly changing. Of course the Hima were also threatened by new technology (the farmers were moving in) but not by new pollutants. New technologies eliminate many risks (smallpox) but create entirely new ones. Professor Wildavsky in Berkeley lives near Livermore, the national nuclear weapons laboratory administered by his university. Douglas taught for many years in London, not far from an island in the Thames that processes liquified natural gas. A 1978 British government report estimated the chance of more than 1,500 people being killed there in a single accident as one in a thousand per year. Do we need sociology to tell us why we are scared?

Douglas and Wildavsky rightly emphasize that our “risk portfolios” are not chosen simply by the character and degree of the risk. That does not entitle them to play down objective facts. We can ask their question about the surge of antiweapons sentiment in the United States. Why is it happening right now, when we have had doomsday with us most of our lives? Surely the new arms race did more to arouse our sleeping fears than anything we shall learn from social anthropology.

Only at the end of their book do Douglas and Wildavsky seriously turn to our real-life problem of unknown or changing risks. They “betray” their “bias towards the center” by urging a need for resilient institutions that can cope. Ecology teaches that an organism armed against a specific list of dangers responds poorly to new kinds of threat. The survivors are the resilient. What we need are central and adaptable institutions. That conclusion in no way follows from their analysis. It sounds trite enough. They could equally well conclude—but do not—that in a world of accelerating new technologies that interact in unforeseen ways, we require endlessly proliferating border sects which can force new dangers into our consciousness before we are destroyed.

This Issue

September 23, 1982