Although mailing lists can be dauntingly intense for your mailbox, there are rare gems that make you glad that you keep checking them. One such example is the MECS Conference ‘Agent Cultures and Zombielands’, hosted by Institute for Advanced Study on Media Cultures for Computational Simulation (MECS), Leuphana University in Lüneburg. As fate would have it, I had booked a train from Utrecht to Berlin on the starting day of the conference. So I decided to exit during my stop in Hannover, and head over to Lüneburg. The focus of this conference was social simulations, and we saw a variety of social scientists and simulation experts sharing their knowledge and research.
The first day kicked off with a keynote by Robert Axtell, a pioneer in computational social science. Axtell presented an overview of the past, present, and future of artificial intelligence and agent-based modelling. Agents provide humans with a way to see things at a lower level of organization, microscopic vision. The unique perspective simulations, is that many agents interacting in a simulation provide us with a macroscopic perspective, that enables us to see the higher levels of organization, which enables a visualisation of the development of social issues. Eric Winsberg presented his research concerning defining simulation vis-a-vis experiment. Though there are many nuanced similarities and differences between a simulation and an experiment, the most defining difference according to Winsberg is the type of background knowledge. Knowledge producing activities rely on arguments, and these arguments rely on background knowledge. I must admit that I had to look up later what he actually meant by this. I found this quote by a Jesper Jekert (2012, 3) that explains Winsberg’s point nicely:
Instead he argues that the important difference between simulations and experiments is what forms the basis of the hope that there are formal similarities, and how researchers justify their beliefs that there are such similarities. For simulations, the hope is based on the fact that we (think that we) know how to build good models of the target system, whereas for experiments the hope is based on the fact that the object and the target belong to the same kind of system. In both experiments and simulations, background knowledge is needed to gain new knowledge, but the background knowledge is not quite the same.
Day two was equally exciting. Alexander Galloway and Rainer Hegselmann each described their own anonymous pioneer relevant to the field of agent-based modelling and AI. Galloway shared his historical research on mathematician Nils Aall Barricelli. Hegselman compared Thomas Shelling‘s famous model of segregation to the lesser-known computational pioneer James Sakoda. Sakoda produced work comparative to Schelling, but more complex and used a computer to calculate his mathematical model on neighbourhood segregation. While Shelling was awarded the Nobel Prize in economic sciences, Sakoda was mostly known for his origami books. The tragic fate of an unknown but brilliant mind.
A particularly interesting lecture was given by Klaus G. Troitzsch from the University of Koblenz-Landau. Troitzsch questioned whether their team could run simulations to model extortion racket systems and possible interventions. Comparing the data of their simulations with criminal data obtained from the police they could test the validity of these results. I particularly liked his definition of a simulation:
“A thought experiment which is carried out with the help of a machine, but without any direct interface to the target system” (Troitzsch 1997, 46)
All in all the conference was very inspiring, and gave me much food for though for my own research. Thanks for the organizers Sebastian Vehlken and Ricky Wichum, I will definitely visit MECS again in the future.
Jerkert, Jesper. “Science in the Age of Computer Simulation–By Eric Winsberg.” Theoria 78.2 (2012): 168-175.
Troitzsch, Klaus G. “Social science simulation—origins, prospects, purposes.” Simulating social phenomena. Springer Berlin Heidelberg, 1997. 41-54.
MECS Jahrestagung 2016 © Leuphana