Prof. Franck Varenne

Franck Varenne graduated as an engineer in electrical engineering and computer science from the Grande Ecole Supélec (Paris, 1993). After a MA in philosophy, a MA in history of science and a Ph.D. in epistemology, he became Associate Professor of Epistemology at the University of Rouen (Normandy) and member of the CNRS-Lab GEMASS (Sorbonne) in 2005. His research focuses on the history and epistemology of formal models and computer simulations in contemporary sciences. His method in epistemology is threefold: 1. historical, 2. comparative, 3. immersive, i.e. observational (empirical) and interactional (coactive). Recent papers: « Chains of Reference in Computer Simulations, in: FMSH Selected Working Paper, October 2013; F. Varenne, P. Chaigneau, J. Petitot, R. Doursat: « Programming the Emergence in Morphogenetically Architected Systems, in: Acta Biotheoretica, Sept. 2015, 63 (3), 295-308.



Searching for laws on agent-based simulations: what for?

The spreading of individual-based and agent-based models and simulations in sciences has many epistemological consequences. One of them is this acceptation by some (e.g. Epstein) that to let the computational model of a target system «grow» on the computer through step by step computations of its mechanisms suffices to explain the phenomena arising in the target system. The correlated debates on explanation and causality in agent-based models have contributed to the recent shift of the epistemological focus from laws to mechanisms in biology and social sciences. But, at the same time, some research programs still try to go a step further or - which could seem controversial - to go back to what could be called again «laws». What for?

Do practitioners search for laws in simulation results only for pragmatic reasons, i.e. for computation efficiency reason or to ameliorate the model’s robustness? Or, isn’t it for theoretical reasons too, that is in order to directly address the difficult question of emergence in complex target systems? Or else: can’t such a move back to laws be seen as correlated to a new search for a kind of - say - computer aided comprehension beyond pure computer grounded explanation? To what extent can such a seeking for law-like abbreviations of computations can be called a «new way of searching for laws» or, otherwise, a way of searching for «new kinds of laws»? When seen as new kinds of laws, do these laws have to belong to a «new kind of science» (Wolfram) or to a more modest and pedestrian reorganization of the strategy of modeling?

This research will try to clarify and substantiate these questions through case-studies. Its aim is to evaluate and discriminate similar but different simulation strategies along with their epistemological background presuppositions. The method will be the following: analysis, comparison and discussion of three different epistemological profiles and types of simulation-based law seeking.