As Becker points out, the potential costs of a lethal pandemic are astronomical. The Spanish flu epidemic of 1918-1919 killed tens of millions of people, and because of the extreme mutability of the flu virus it is entirely possible that an equally lethal strain may re-emerge. True, the death rate would probably be significantly lower today because of improvements in medical care, but it might not be. The new virus might be even more lethal, or more people might be infected; in either event as many or even more people might die. The mutability of the virus can make existing flu vaccines completely ineffectual. Because the virus is airborne and has an infectious incubation period of several days (that is, a period in which a carrier of the virus is asymptomatic but infectious), a large number of people can be infected before the disease is even discovered and measures for preventing its further spread implemented. International air travel can spread the disease throughout the world in almost no time. The interesting economic issue, besides the costs inflicted by the disease, stressed by Becker, is the optimal response to the danger of a lethal flu pandemic. Ideally one would like to be able to calculate the expected cost of the pandemic and compare that with the cost and efficacy of the possible preventive and remedial efforts. The expected cost would be the cost inflicted by the pandemic discounted (multiplied) by the probability that, in the absence of preventive measures, that cost would be incurred. Unfortunately, that probability cannot be estimated. But since we experienced such a pandemic less than a year ago, the probability cannot be considered trivial. Since the potential cost if such a pandemic does occur is so enormous, efforts at prevention or mitigation deserve serious consideration. If, following the Lancet estimate discussed by Becker, we guess that a repetition of the 1918-1919 pandemic would inflict a total cost worldwide of $20 trillion (this estimate excludes the narrowly economic costs, but they would probably be lower, in part because thinning out populations can raise per capita incomes, especially if the very young and the very old, and poor people in overpopulated countries, are the principal victims), and if we indulge a further guess that there is a 1 percent annual probability of such an event, the annual expected cost would be $600 billion. Of course this would not imply that the world should spend $600 billion a year on trying to prevent, or reduce the costs of, a lethal flu epidemic. Costs and benefits have to be compared at the margin. The marginal benefit of additional expenditures on preventing or alleviating the costs of a flu academic is probably zero after a few billion dollars of expenditure. Indeed, it could be negative, because large expansions in the number of research personnel working on vaccines against lethal airborne diseases increase the number of people who have the skills required for bioterrorism--a concern to which I return later in this comment. On the prevention side, the most important measures are (1) global early warning systems and (2) the development of very broad-spectrum flu vaccines, that is, vaccines that provide protection against possible mutant forms of the virus. Neither of these preventive approaches are terribly expensive. Of course, one could spend unlimited amounts of money on vaccine research, but this would be inconsistent with the marginal principle: the returns to additional resources on developing a vaccine that would protect people against all possible mutations of the flu virus probably diminish rapidly. Crash programs have limited efficacy in solving deep scientific puzzles. On the response or remediation side, expansion of hospital facilities and arrangements for large-scale quarantining should be considered, although here the costs are likely to become prohibitive quite rapidly. For example, there are fewer than one million hospital beds in the United States; imagine how much it would cost to expand this number tenfold in order to be prepared for a major epidemic--yet even a tenfold increase would accommodate only about 3 percent of the U.S. population. Analysis is complicated and the prospects darkened by the threat of bioterrorism. The flu virus, like the smallbox virus, lends itself to weaponization; both viruses are airborne and have significant infectious incubation periods. Smallpox is more lethal than any known flu virus, including the 1918¬¨¬¨-1919 virus, the death rate (as distinct from number of deaths) fron which was very low. But increasing the lethality of a virus, and also modifying it to make existing vaccines ineffectual against it, are well within the current state of scientific knowledge, and require only modest technical skills and inexpensive manufacturing facilities. Optimal responses to pandemics, whether natural pandemics or ones contrived by terrorists, are complicated not only by diminishing returns, but also by the multiplicity of catastrophic threats. Even if multiplying the number of hospital beds tenfold could be a justified measure in preparation for a possible lethal flu epidemic, it would be immensely costly and this would as a practical matter preclude financing measures directed against other catastrophic possiblities, which include abrupt global warming, asteroid strikes, biodiversity depletion, nuclear terrorism, and (as we have become acutely aware) global depressions. We need an overall "catastrophe budget" that would match expenditures to the net expected benefits of particular measures targeted at particular catastrophic threats.