Beta-Gamma science for sustainable agriculture: taking the implications of complexity seriously

Life is the interaction of non-equivalent observers guided by different but legitimate goals and non-reducible models of each other.This implies that when studying the evolution of living systems and the sustainability of human progress it is impossible to define in substantive terms "optimal solutions" or "best courses of action". In fact, these concepts do entail a set of pre-analytical ideological decisions: "best" or "optimum" for whom? On which time scale? In relation to which selection of attributes of performance? Who should decide about this?When coming to the availability of hard data to study sustainability things do not get easier. Empirical information is generated by a mix of abstractions about reality (the definition of the identity for the modeled systems, which is associated with a finite set of categories - variables - adopted in the model) and real signals from reality (those found when operating the selected measurement scheme). Therefore, empirical information should not necessarily be considered more "substantive" than the information provided by philosophers or informed actors.In spite of this epistemological impasse, hard scientists do not have alternatives to the old reductionistic agenda of "reduzieren" and "classifizieren" (they can only adopt one formal narrative at the time). They can only base their representations of reality on a finite set of models with which it is possible to simulate patterns of behaviours validated in the past. These models obviously do not evolve in time.The complexity revolution provides a way out of this impasse. To provide useful information to societal processes of decision making about sustainability scientists can: (i) use simultaneously non-equivalent relevant narratives referring to different scientific disciplines and different scales. This implies adopting integrated packages of models able to reflect useful perspectives and relevant mechanisms (Multi-Scale Integrated Analysis); and (ii) get involved in a process of evaluation which is multicriterial and participatory. This implies the requirement of adding a "quality assurance control" to the process of decision making based on scientific information (Societal-Multi-Criteria-Evaluation).Innovative concepts derived from complex systems thinking [e.g. learning holarchies, mosaic effects across scales, impredicative loop analysis, useful narratives to surf complex time] can be used to enlarge the repertoire of analytical tools available for analysis of sustainability. When using these concepts it becomes possible to look for meta-models (useful metaphors and narratives) when choosing a scientific problem structuring, rather than relying solely on statistical tests and differential equations.In the final part of the thesis, practical examples of applications are discussed to show that the adoption of these innovative concepts can actually help in developing useful analyses of the sustainability of agriculture. Principal applications refers to: (1) the definition of a mechanism of benchmarking for typologies of societal metabolism. This method of analysis can be used to define typologies of farming systems. This typologies make it possible to infer typologies of constraints that the socio-economic context entail on farming systems. (2) the establishment of a bridge between the characteristics of the farming systems and typologies of ecological impact on terrestrial ecosystems. (3) an analysis of farming systems seen as nested hierarchical systems. In this case, different agents operating at different levels can be characterized on a multi-criteria performance, This implies that the various strategy matrices adopted at different levels can be effectively studied .