Subjective Probabilities and Sensible Precaution Dominic Roser, University of Zurich, roser@ethik.uzh.ch This paper is work in progress. 1 Comments are most welcome. Please do not cite or circulate. At the conference in Bern I will talk about parts of sections 3 and 4. The claim of this paper is that, at least in the area of climate policy, we should not justify the call for precautionary policy choice on the basis of a supposed lack of probabilities. Probabilities are (almost) always available and we ought to make use of whatever probabilities are available – even if the only available probabilities have extremely low credentials. Precautionary policy choice can be justified sufficiently well on the basis of other considerations than a lack of probabilities. 1. Preliminaries Before I turn to these claims in sections 2 – 6, I would like to make some comments on Precautionary Principles in general. Many people intuitively think that the strength and the shape of the duty to protect future generations from potentially harmful climate change is heavily dependent on the uncertainty 2 associated with climate science. They think that if the state of the evidence and the state of the scientific models make exact predictions difficult it is more urgent to prevent climate change than if state of the art science made it possible to exclude many scenarios with a high degree of certainty. This general and instictive idea is captured by innumerable Precautionary Principles. Precautionary Principles are all driven by a vague "precautionary intuition" which roughly says that uncertainty ought to make a difference for policy choice because in conditions of uncertainty it is usually better to be safe than sorry. This vague intuition is shared by many, including myself. However, it is not easy to spell it out precisely. Precautionary Principles which try to capture the vague intuition often end up with extreme 3 , paralyzing or else vacuous, trivial formulations. Many of the existing 1 I hope that this also partly excuses the inordinate amount of footnotes such as this one. 2 The term "uncertainty" is often used in a broad sense as an opposite to complete certainty. Sometimes, however, it is used in a technical, narrow sense where there are three opposites to complete certainty which are characterized in the following manner (or in a closely related manner): • risk (probabilities are assignable) • uncertainty (probabilities are not assignable but the range of possible outcomes is given) • ignorance (not even the possible outcomes are given). In this paper, I will use the unqualified term "uncertainty" in the broad sense. 3 A wonderful statement of this view is given by Luca Turin and Tania Sanchez (2008, 46) in their passionate book about perfumes: "Some have argued that since perfumes do no measurable good, the level of risk that can be tolerated in their use is zero. This appears to be the puritanical subtext to some recent legislation regulating fragrance. I take the view that, much in the way that Michelangelo's David can topple over and kill you or your chaud-froid de grives au Gevrey-Chambertin can result in a kitchen fire, all art forms provide beauty at the expense of some risk (…). Nobody ever died from wearing Mitsouko, but lots of babies were born as a result of it."