This most recent ABM get-together had a more varied feel, with presentations, a fascinating discussion with Rob Axtell, and short updates from Ken and Howard.
- Rob Axtell started the meeting by handing out copies of his seminal book Growing Artificial Societies, and with an introduction of what he aims to focus on while in Oxford (just started with the iNET@Oxford group). Rob is currently working on full-scale models of systems where there is enough quality data to make them e.g. of the 120 million people that work in private sector organisations. This is in part a serious computer science problem – how to run such large scale simulations in a reasonable amount of time (ref. Oxford’s ARC service). The modern computer science Zeitgeist seems to be this ambitious e.g. modelling whole heart (Prof. Denis Noble) and human brain project. We also discussed how far to go when modelling agents e.g. their cognition. The conversation turned philosophical once people started asking questions, so I’ll finish with a link to Rob Axtell’s citations on Google Scholar, noting the paper Rob recently published with Omar Guerrero. (The discussion finished off with the general note that classical models of the economy leave much to be desired largely due to their gross over-simplification of agents…)
- Anders Sandberg gave us an update on his project with insurance companies who want to understand the systemic risks in how they are using models. His meta model of how models of risks are used by insurance companies demonstrates some of the challenges of bounded rationality in competition situations. A model of models is appropriate for a philosophy project. Ander’s slides: modelling the systemic risk of risk modelling.
- Andreas Duering gave an update on his model on medieval graveyard records, specifically showing some data he’d generated with his model to ask how he can decide how significant the data is considering the small and skewed sample size (of graveyards). Unlike most graveyards, the graves of Talheim are believed to all be from the same point in time providing a snapshot of an ancient community to compare with the populations generated by his models.
- Ken gave us some updates about the exoplanet simulator he helped Maria Marinari, a school student, make over the summer, his recent trials running NetLogo models using ARC computers, and how he’d come up against the issue of running models with impossibly large parameter spaces with his Spanish flu model. He also demonstrated the new direct connection between NetLogo and the Behaviour Composer that provides a strong alternative to Java applets.
- Francesco Caravelli gave us an introduction to a model he is building of an agricultural market in Uganda where farmers, traders and a range of institutions interact according to a social network that is largely enacted using mobile phones to agree prices and times when food is moved between fields and the airport. Anders raised the topic of social stickiness e.g. loyalty between farmers and trusted traders (a major topic in Halldór Laxness classic book: Independent People where the old guard preferred loyalty and cooperatives, the new guard created the Icelandic banks of recent fame), to which Rob pointed us to Kirman, Alan P., and Nicolaas J. Vriend (2001), “Evolving Market Structure: An ACE Model of Price Dispersion and Loyalty” paper, see Networks section in Rob’s introduction to ABM page.
- Howard Noble then gave us a quick update on a new feature that will hopefully be in the next release of NetLogo called the Review Tab. The feature makes it possible to record, replay and annotate the way a person interacts with a model as they run it without impacting the performance of the model on a laptop (i.e. slowing it down or making the display flicker). This is exactly the tool Howard and David need for their OxGAME project. Rob suggested reading papers by François Bousquet at CIRAD.
- Chris Farmer finished the session off with some thoughts on the role ABMs could play in his new role at the James Martin school in Resource Stewardship. One area Chris thinks could be promising is participatory modelling. He referred to Carl Water’s book on adaptive management and one of the core idea which is that decisions are best made through discussion between all stakeholders. Of course Elinor Ostrom was mentioned at this point, along with Julia Schindler, Steve Lansing (and the Goddess and the computer short film). Rob Axtell asked the question, how should we behave even if we know what is optimal?