https://halshs.archives-ouvertes.fr/halshs-00003925v2Ha-Duong, MinhMinhHa-DuongCIRED - Centre international de recherche sur l'environnement et le développement - ENGREF - Ecole Nationale du Génie Rural, des Eaux et des Forêts - Cirad - Centre de Coopération Internationale en Recherche Agronomique pour le Développement - EHESS - École des hautes études en sciences sociales - ENPC - École des Ponts ParisTech - CNRS - Centre National de la Recherche ScientifiqueScenarios, probability and possible futuresHAL CCSD2006Futuresfuturiblescenariospossibilityimprecise probabilitiesuncertaintyfuzzy logic[SHS.ECO] Humanities and Social Sciences/Economics and Finance[SHS.STAT] Humanities and Social Sciences/Methods and statisticsHa-Duong, Minh2006-04-27 10:39:362022-04-19 10:12:252006-04-27 11:16:50enOther publicationshttps://halshs.archives-ouvertes.fr/halshs-00003925v1application/pdf2This paper provides an introduction to the mathematical theory of possibility, and examines how this tool can contribute to the analysis of far distant futures. The degree of mathematical possibility of a future is a number between O and 1. It quantifies the extend to which a future event is implausible or surprising, without implying that it has to happen somehow. Intuitively, a degree of possibility can be seen as the upper bound of a range of admissible probability levels which goes all the way down to zero. Thus, the proposition `The possibility of X is Pi(X) can be read as `The probability of X is not greater than Pi(X).<br /><br />Possibility levels offers a measure to quantify the degree of unlikelihood of far distant futures. It offers an alternative between forecasts and scenarios, which are both problematic. Long range planning using forecasts with precise probabilities is problematic because it tends to suggests a false degree of precision. Using scenarios without any quantified uncertainty levels is problematic because it may lead to unjustified attention to the extreme scenarios.<br /><br />This paper further deals with the question of extreme cases. It examines how experts should build a set of two to four well contrasted and precisely described futures that summarizes in a simple way their knowledge. Like scenario makers, these experts face multiple objectives: they have to anchor their analysis in credible expertise; depict though-provoking possible futures; but not so provocative as to be dismissed out-of-hand. The first objective can be achieved by describing a future of possibility level 1. The second and third objective, however, balance each other. We find that a satisfying balance can be achieved by selecting extreme cases that do not rule out equiprobability. For example, if there are three cases, the possibility level of extremes should be about 1/3.