David Anzola

The literature in agent-based social simulation suggests that a model is validated when it is shown to ‘successfully’, ‘adequately’ or ‘satisfactorily’ represent the target phenomenon. The notion of ‘successful’, ‘adequate’ or ‘satisfactory’ representation, however, is both underspecified and difficult to generalise, in part, because practitioners use a multiplicity of criteria to judge representation, some of which are not entirely dependent on the testing of a computational model during validation processes. This article argues that practitioners should address social epistemology to achieve a deeper understanding of how warrants for belief in the adequacy of representation are produced. Two fundamental social processes for validation: interpretation and commensuration, are discussed to justify this claim. The analysis is advanced with a twofold aim. First, it shows that the conceptualisation of validation could greatly benefit from incorporating elements of social epistemology, for the criteria used to judge adequacy of representation are influenced by the social, cognitive and physical organisation of social simulation. Second, it evidences that standardisation tools such as protocols and frameworks fall short in accounting for key elements of social epistemology that affect different instances of validation processes.

Successful instances of interdisciplinary collaboration can eventually enter a process of disciplinarisation. This article analyses one of those instances: agent-based computational social science, an emerging disciplinary field articulated around the use of computational models to study social phenomena. The discussion centres on how, in knowledge transfer dynamics from traditional disciplinary areas, practitioners parsed several epistemic resources to produce new foundational disciplinary shared commitments, and how disagreements operated as a mechanism of differentiation in their production. Two parsing processes are examined to illustrate this claim. The first one is the parsing of the qualitative–quantitative dualism, arguably the most important methodological disagreement in social science. The second one is the parsing of prediction, a key value in contemporary science. The analysis evidences that disagreements have fostered both external and internal dynamics of differentiation in agent-based computational social science. The former have permitted a more efficient use of epistemic resources, whereas the latter have forced practitioners to modify the foundational narrative and the agenda of the field.

This article addresses knowledge transfer dynamics in agent-based computational social science. The goal of the text is twofold. First, it describes the tensions arising from the convergence of different disciplinary traditions in the emergence of this new area of study and, second, it shows how these tensions are dealt with through the articulation of distinctive practices of knowledge production and transmission. To achieve this goal, three major instances of knowledge transfer dynamics in agent-based computational social science are analysed. The first instance is the emergence of the research field. Relations of knowledge transfer and cross-fertilisation between agent-based computational social science and wider and more established disciplinary areas: complexity science, computational science and social science, are discussed. The second instance is the approach to scientific modelling in the field. It is shown how the practice of agent-based modelling is affected by the conflicting coexistence of shared methodological commitments transferred from both empirical and formal disciplines. Lastly, the third instance pertains internal practices of knowledge production and transmission. Through the discussion of these practices, the tensions arising from converging dissimilar disciplinary traditions in agent-based computational social science are highlighted.

Even though the philosophy of simulation is intended as a comprehensive reflection about the practice of computer simulation in contemporary science, its output has been disproportionately shaped by research on equation-based simulation in the physical and climate sciences. Hence, the particularities of alternative practices of computer simulation in other scientific domains are not sufficiently accounted for in the current philosophy of simulation literature. This article centres on agent-based social simulation, a relatively established type of simulation in the social sciences, to exemplify this claim. The analysis advanced has a twofold goal. First, it shows that the philosophy of agent-based social simulation, mostly developed by practitioners themselves, is, on one hand, heavily influenced by the methodological features of agent-based modelling and by the loose and fragmented character of social theory and, on the other hand, distinctively shaped by contrasting views of what it implies to do social research with virtual or artificial societies. Second, it suggests ways in which cross-fertilisation could enrich the philosophical understanding of computer simulation both in agent-based social simulation and in the philosophy of simulation.

ArgumentAgent-based social simulations have historically been evaluated using two criteria: verification and validation. This article questions the adequacy of this dual evaluation scheme. It claims that the scheme does not conform to everyday practices of evaluation, and has, over time, fostered a theory-practice gap in the assessment of social simulations. This gap originates because the dual evaluation scheme, inherited from computer science and software engineering, on one hand, overemphasizes the technical and formal aspects of the implementation process and, on the other hand, misrepresents the connection between the conceptual and the computational model. The mismatch between evaluation theory and practice, it is suggested, might be overcome if practitioners of agent-based social simulation adopt a single criterion evaluation scheme in which: i) the technical/formal issues of the implementation process are tackled as a matter of debugging or instrument calibration, and ii) the epistemological issues surrounding the connection between conceptual and computational models are addressed as a matter of validation.