To increase communication among modellers, this workshop series was launched some years ago, in 2003: M2M (“model-to-model”), the specific aim of which was to sustain the development of techniques for comparing models used in social simulations or computational social science. Several approaches were proposed and followed by modellers, during the two meetings that have taken place so far, summarised by Hales et al. (2003) thus:

“There are quite a number of ways that models can be compared, including:

• Rewriting models that others have described in papers so as to understand them more deeply and reproduce the stated results;
• Composing models where different scales are inter-related in a larger model - the results of one model being used in the other;
• Comparing different models that announce the same type of results and trying to see if they actually produce similar (or the same) results - sometimes termed ‘aligning’ of models;
• Comparing different models based on their fitness with respect to a set of data;
• Using one model as a post-hoc summary or abstraction of another model's results; thus constraining the scope of an existing model to enable more powerful techniques to be applied in a different computational framework; or
• Using models with different structures and assumptions to confirm each other's results.”

All these comparative strategies are aimed at advancing our knowledge of social complexity through MABS and related models.

Following the first M2M, a special issue of the Journal of Artificial Societies and Social Simulation gathered six papers on the topic (JASSS, 2003, vol. 6, no. 4). Replication, docking, aligning and comparison to perfect rationality models were developed. As expressed in the special issue introduction, a number of "open issues" were identified by the participants. These included:

• Could a "typology" of MABS be produced?
• Would a “taxonomy” be possible and useful?
• Can further "abstract classes" of models be identified and related to existing models?
• Should such classes be formally specified? And, if so, how?
• Will different researchers agree on abstracted classes and who's going to do all the work?

It is certainly clear that there is currently nothing close to a common framework in MABS, so any progress is to be welcomed.

MABS can be used to better understand earlier (more traditional and more empirically validated) models by implementing agents following such models but relaxing previous constraints (such as agent homogeneity). There was a strong belief (Moss, 1999) that it would be useful to integrate or align coarse and fine-grain models when scaling up to more complex cases. More generally, there is no such thing as a heuristic or methodology for model alignment—currently it is a bit of a ‘black art’ but perhaps a more structured approach can be outlined. Even if under some conditions models are found to converge it is also of interest (perhaps even greater interest) to identify the conditions under which models don't agree. It is currently unclear as to how one would go about achieving this. It is generally agreed that replication or docking of MABS models is a worthwhile exercise for increasing confidence and understanding (Ramanath & Gilbert, 2004). We expect that in the not-too-distant future the understanding of the role of replication in increasing knowledge will have improved greatly.

The aim of the workshop is to increase discussions on the topic of comparing social simulation models, including replication and transferability, and propose new techniques and results concerning the comparison of models. This of course can include new approaches that were not developed in previous M2M workshops, including more formal analyses of agent based models.

Presentations will be 35 to 45 minutes in length, plus time for discussions.