Volume 3, April 2003
Reviewed by Elizabeth A. Corley, Ph.D. **
* McGill-Queen’s University Press: Montreal & Kingston, 2002, 185 pages.
** Assistant Professor, School of International and Public Affairs, Columbia University
How do we make policy decisions to avert potential global catastrophes when predictions from scientific models are highly uncertain? In his book, Apocalypse Soon? Wagering on Warnings of Global Catastrophe, philosopher Stephen Haller provides an answer to this timely question. Throughout the book, Haller addresses the varying levels of risk regarding global catastrophes and he proposes how we might move forward with policy-making regarding these risks, even though scientific information about the future is often sparse or non-existent. Haller’s arguments are presented in a clear and concise way that makes this book both engaging and accessible.
The main objective of Haller’s book is to explore how environmental policies should be made in the face of great scientific uncertainty - the type that we find with global warming models. Haller argues that current and historical warnings from scientists about global catastrophes like overpopulation, global warming, nuclear winter, and ozone-layer depletion are mixed blessings. On the one hand, they may help us avert potential global disaster but, on the other hand, they can tie up resources and immobilize policy-making as we attempt to avoid a disaster that might never come to pass. He discusses the use of scientific models in predicting levels of risk for global disasters (such as global warming) while arguing that we should place little confidence in these models and their predictive abilities because the large-scale nature of global models makes it impossible for their developers to test against their predictions and improve the models by trial and error.
If the potential costs of global catastrophe are so high and the level of risk is unknown, how do we make policies based on this mixed information? Do we assume that the models are correct and prepare for the worst case scenario? How do we evaluate scientific models that predict global catastrophe to determine if they are credible? These are a few of the questions that Haller specifically addresses in the book.
In Part I of this book (Chapters 1 – 6), Haller outlines several examples of past and present global models and he discusses how these models are usually evaluated. In Chapter 1, Haller presents examples of global systems models that predict large-scale catastrophes; specifically, he explores the history of nuclear winter models, ozone depletion models, overpopulation models, and global warming models.
In Chapter 2, Haller discusses the nature of scientific models and explains how the two model goals of explanation and prediction can contradict each other. Some models, such as models of traffic circulation, can provide both an explanation and prediction of the future. If a model, however, is only able to yield either an explanation or a prediction (but not both), then the model will be rejected or accepted based only on the successful property.
In Chapter 3, Haller presents several criteria for assessing whether or not any prediction can serve as a good test of a model. Usually, scientists tinker with models to assess their reliability, but in the case of global warming, the confirmation of predictions will come too late to be helpful for policy makers. He argues that “the nature of the problem in question is that we are urged to take immediate action before predictive success is possible.”
In Chapter 4, Haller explores options for testing whether a model can accurately predict future conditions without waiting to see if those predictions come to pass. He argues that to accurately evaluate a predictive, scientific model three issues must be addressed: 1) how accurate are the initial data that we supplied to the model?, 2) how reliable is the model?, and 3) are the probabilities the model assigns to predictions justified? Haller believes that global models are weak in all three of these areas.
Throughout Chapters 3 and 4, Haller presents traditional methods for evaluating scientific models using trial and error or “direct tests” of the models. In Chapter 5, however, Haller presents “indirect tests” of models and says that these tests allow scientists to evaluate theories and models even when trial and error cannot be used. Indirect tests of a model’s strength for prediction utilize historical agreement, predictions in the near future, sensitivity analysis, and analogues to similar physical phenomena. Haller discusses each of these indirect tests in detail and applies them to global warming models. He concludes that neither indirect tests nor direct tests of global warming models instill confidence in the models.
Haller summarizes his concerns about the explanatory and predictive powers of global warming models in Chapter 6. He is especially concerned with epistemic difficulties that are “peculiar to the problem of assessing a model’s reliability without the benefit of direct testing through trial and error.” In the case of global scientific models, the epistemic problem is that model predictions could be simple logical implications that provide information about what is possible, but they may not give us any sense of the most likely scenario. In short, we use global models and we either expect that they will be applicable to the situation at hand (such as global warming) or that they will not be applicable. If the model is deemed applicable, it exhibits successful explanation and prediction of phenomena and we are satisfied with it. On the other hand, if the model’s predictions do not match our observations, we might still be satisfied with it and just explain why the observations deviated from the ideal case. Haller argues that we never know beforehand if a model is applicable or not; when models are used as an ideal, they cannot be falsified. If any explanation of the deviation of observations from the ideal is required, it is not supplied by the model. These types of explanations are supplied by the researcher and are value-laden.
In Part II of his book (Chapters 7-10), Haller explores the role of traditional decision-making theory as a tool for guiding actions regarding potential global catastrophic events. He points out that it is difficult (if not impossible) for policy-makers to react cautiously to every risk that is presented to them. In Chapter 7, Haller argues that “science by itself cannot drive policy because other values must figure in decision-making.” Even if existing global warming models did give us comprehensive scientific probabilities about global catastrophe, we would not make policy decisions based solely on the science; desires and values are also part of the policy-making process.
In Chapter 8, Haller explores the strengths and weaknesses of the precautionary principle as a tool for policy-making when faced with conditions of extensive scientific uncertainty. He presents the concept of “epistemic conservatism,” which is the maxim that science follows when either a Type I or Type II error is unavoidable. Epistemic conservatism is defined as the concept that new theories or models are treated with skepticism until they are proven to be valid and useful. Haller argues that if we adopt this notion of epistemic conservatism for dealing with models of global warming, the “burden of proof will lie with those who assert that there is such a risk, rather than with those who want to continue business-as-usual.”
While this approach would protect us from over-reacting to the threat of global warming, it could also lead to global disaster if the models that predict dire consequences of the warming prove to be true. Haller says that we need to know if “there are any good prudential or moral reasons for escaping epistemic conservatism when confronted with the possibility of global disaster.” He believes that in the case of zero-infinity dilemmas, cost-benefit analysis does not provide a sufficient decision-making framework because making decisions in cases of low probability is quite difference from decision-making in cases of uncertain probability. Haller turns to the concept of the precautionary principle as an alternative decision-making framework for zero-infinity dilemmas. Adopting the precautionary principle as a basis for decision-making in cases of uncertainty transfers the burden of proof to those who are skeptical about global warming models.
In Chapter 9, Haller presents arguments by scholars who have called for new conceptions of rationality. These scholars argue that this new concept of rationality should place more emphasis on ethical and democratic values because decision-making about global risks is mostly a matter of ideology anyway; therefore, we should openly embrace making decisions based on political, ethical, and cultural reasons, rather than acting as though we are making decisions based on scientific certainty. Haller disagrees with this view because he argues that it is a “result of running together epistemic values and action-guiding values” – that is, it does not distinguish between claims that something is the case and claims that a particular type of action should be taken.
In Chapter 10, Haller revisits the precautionary principle and points out that while epistemological arguments provide support for the concept of epistemic conservatism, ethical arguments provide support for the precautionary principle. Some scholars use moral arguments to override the problems with the precautionary principle – they argue that some risks are not morally justifiable. Haller says that ethical arguments in support of the precautionary principle “shift the burden of proof to those who claim that no harmful effects will occur.” Haller criticizes five ethical arguments that could support use of the precautionary principle; he concludes that these arguments cannot generate sufficient, widespread support of the precautionary principle.
Ultimately, Haller borrows arguments from Pascal and William James to argue that we must embrace the precautionary principle in cases of scientific uncertainty, such as global warming. Following Pascal, Haller suggests “that if one allows the perceived high costs of precaution to sway against precautionary action, then one just does not understand the infinite costs of catastrophe.” Haller argues that Pascal’s Wager, which lays out why one should believe that God exists, is not about belief, rather is about a gamble. Even if there is a very small probability that God exists, we should bet on it because the magnitude of the potential reward of the decision is so great. Haller argues that the principles of gambling in Pascal’s Wager and global warming are similar. Even if the costs of acting like global catastrophe will occur are tremendously high, they are “nothing compared to the infinite negative value of a catastrophe.” Thus, when we make a bet in favor of the occurrence of global warming (of the existence of God) we are not believing, rather we are acting. Haller concludes that we do not necessarily need to believe that global catastrophe will occur, as long as we act to prevent catastrophe.
Additionally, following William James, Haller claims that “deciding a course of action on nonintellectual grounds, which the arguments for precaution specifically call for, is justified under certain conditions – conditions that define what James calls a genuine option.” Haller agrees with James that “there are certain momentous decisions that must be made under uncertainty and, in these cases, the caveat of scientific reasoning to suspend judgment is misplaced.” An option is momentous if the situation would lead to a decision that is irreversible or if the stakes are particularly high.
Haller argues that Pascal and James’s arguments provide reasons for action, but not necessarily for belief. He says that if we act on the hypothesis that God exists, “both Pascal and James point out that the benefits of this action will have an immediate pay-off and might even have an infinite pay-off in the afterlife. These are reasons for faith.” Haller believes that the same is true for having faith that global warming is occurring. Haller says that “the evidence from the models does not provide sufficient evidence for belief, and this is why they fail to convince everyone. However, they do provide reasons for action.” We merely must accept these models as working hypotheses.
In summary, Stephen Haller provides a comprehensive criticism of the reliability of global warming models and a deft exploration of the concept of policy-making in cases of potential global catastrophe. This book clearly and consistently demonstrates that the prediction abilities of many global warming models are weak and cannot serve as a sufficient basis for calculating the probability of global catastrophe. One complaint however is that Haller presents arguments against the predictive powers of global catastrophe models, but discusses the models in the abstract without focusing on the specifics of a few of the widely accepted global warming models. Therefore, a reader without extensive knowledge of the science behind global warming models might wonder if the weaknesses that Haller presents are common to all models or simply a few of them (i.e., can each of these models have all of these weaknesses?). If Haller had introduced the specifics of a few of the more widely accepted models, I think that the reader would be left with a better sense of the strengths and weaknesses of the models. However, this focus would have lead to a different (and doubtless much longer) book.
Throughout the book, Haller’s philosophical arguments are presented in a thoughtful and engaging manner that is accessible to a broad range of audiences. The organization of the book is particularly strong. Haller guides the reader through his arguments from beginning to end. First, he presents the weaknesses of global catastrophe models and then explains to the reader why policy-makers must heed the warnings of these models, despite their weaknesses. While some readers might not agree with Haller’s evaluation of current global catastrophe models and his discussion of their weaknesses, most readers will find his arguments noteworthy and thoughtful.
In conclusion, Stephen Haller has written a thoughtful book that is accessible to a broad range of audience types and would be of interest to philosophers, scholars of policy studies, natural scientists, and social scientists, as well as practitioners, policy-makers, and the general public.
Jeffrey, R. C. (1983). The Logic of Decision. Second edition. Chicago: University of Chicago Press.
Keeney, R. L., and H. Raiffa. (1976). Decisions with Multiple Objectives: Preferences and Value Tradeoffs. New York: Wiley.
Lindblom, C.E. (1965). The Intelligence of Democracy. New York: Free Press.
March, J.G. and H.A. Simon (1958). Organizations. New York: Wiley.
Raiffa, H. (1968). Decision Analysis: Introductory Lectures on Choices Under Uncertainty.Reading, MA: Addison-Wesley.
Simon, H.A. (1947). Administrative Behavior. New York: Macmillan.
Simon, H.A. (1957). Models of Man. New York: Wiley.
 Haller, p. 33.
 Haller, p. 65.
 Many policy scholars have argued that the rational decision-making model is weak because it lacks behavioral realism and its assumptions are unrealistic. For an introduction to the field of rational decision-making model, see Raiffa (1968), Keeneyand Raiffa (1976), or Jeffrey (1983). For critiques of the rational decision-making model, see Simon (1947), Simon (1957), March and Simon (1958), or Lindblom (1965).
 Haller, p. 79.
 Haller, p. 91.
 Haller, p. 92.
 Zero-infinity dilemmas are posed by decisions that involve a very small (or close to zero) probability of a catastrophic outcome that could be infinitely harmful.
 Haller, p. 114.
 Haller, p. 142.
 Haller, p. 156.
 Haller, p. 157.
 Haller, p. 158.
 Haller, p. 161.
 Haller, p. 162.
Return to Home Page