Rescher, 83 (Nicholas, Department of Philosophy at the University of Pittsburgh, Risk: A Philosophical Introduction to the theory of risk evaluation, p. 68)
With “ordinary” risk situations, one is in a position to use expected values as a basis for assessment. But there are various sorts of “incommensurable” risks for which this procedure is not reasonable. When the discrepancy of the negatives at issue in a risk situation is sufficiently large, some hazards are simply unacceptablerelative to the others, and it makes perfectly good sense to ignore the “balance of probabilities” and proceedsimply and solely with reference to this disparity. In such cases one will prefer—perfectly reasonably—the certainty of a small loss to the remote prospect of a large loss, even when the hazard associated with this choice has the lesser expected value. (The only qualification here is that the disasterous outcome in view must represent a real possibility and not one whise probability is effectively zero.)
