Jeff Schank: Compiling the ABM world

Jeff Schank at  University of California, Davis, is compiling a website of as many researchers and scholars working with agent-based modeling as he can find. If you’re interested in being listed on his site, drop him a line. A brief glance at the list shows he’s got many of the leading lights, so if you’re interested in seeing the range and depth of applied ABM, it’s well worth a stop. Jeff also has a number of thoughtful essays on the nature & intellectual role of ABM for understanding human behavior:

[…] In this sense, performing simulations of agents with memory and using different strategies based on an agent’s memory, is the erection of new scaffolding for building insight and understanding into the evolution of cooperation.  Of course, the scaffolding may be faulty and our apparent understanding and insight may collapse or the scaffolding may allow us to only build insight and understanding in only a limited respect.  Nevertheless, it is a starting point from which other work can be compared.

It would’ve been nice

It would’ve been nice if the IT folks at U Manitoba had given me some warning that they were about to close my account. I’m no longer going to be teaching for them in the fall, it is true; but a lot of my stuff – not to mention my agent models – are on their servers.

My own fault, I guess – I should’ve cleaned everything off of there when I decided to decline the fall courses, but still, it would’ve been nice to have had some warning.

So, if you’re looking for me @umanitoba.ca, that doesn’t work any more. I’ll be getting some new contact info before too much longer, and hopefully, some space for my simulations, too.

Some agent based modelling readings

New ABM papers to read

Just seen, and on my list to read:

Tubaro, P., & Casilli, A. A. (2010). ”An Ethnographic Seduction”: How
Qualitative Research and Agent-based Models can Benefit Each Other
Bulletin de Méthodologie Sociologique, 106 (1), 59-74 DOI: 10.1177/0759106309360111

A new article has just come out, co-authored with Antonio Casilli on ‘‘An Ethnographic Seduction’’: How Qualitative Research and Agent-based Models can Benefit Each Other. We propose a new form of mixed method for the social sciences, combining ethnographic research and agent-based computer simulation.

from Tubaro’s blog.

Also, from Paul Torrens, Agent-based Models and the Spatial Sciences:

[…] Geographers’ work with ABMs has helped to strengthen existing ties with related disciplines such as computer science and informatics, ecology, sustainability science, economics, anthropology, political science and the earth sciences. Primarily because of the value placed on spatial science and behavioral geography in agent-based modeling, work of this kind is helping to infuse geographical perspectives and ‘spatial thinking’ into these fields. This article reviews the development of agent-based modeling in the spatial sciences, its current uses and applications in physical and human geography and potential future trends in its research and development.

And reviews from the Journal of Artificial Societies and Social Simulation:

The Bounds of Reason: Game Theory and the Unification of the Behavioral Sciences
Gintis, Herbert
Reviewed by Juliette Rouchier

The Oxford Handbook of Analytical Sociology
Hedström, Peter and Bearman, Peter (eds.)
Reviewed by Flaminio Squazzoni

Complexity: A Guided Tour
Mitchell, Melanie
Reviewed by John Bragin

Mind & Society: Special Issue on Social Simulation, Volume 8, Number 2, 2009
Squazzoni, F. (Ed)
Reviewed by Brian Castellani

Building an Agent Based Model

A number of years ago, I had the pleasure of participating in a workshop led by Mike Gizzi and James Steiner; at that time, Mike had some excellent handouts and notes about the process of model building. Mike’s since moved on to other things, but via James’ site, I came across those notes – well worth considering if you are about to embark on the possibilities of agent modeling in your own research!


(click on the image for the tutorial).

Patronage as Social Computation

I’ve been reading John Miller and Scott Pages’ Complex Adaptive Systems – an introduction to computational models of social life, and Melanie Mitchell’s Complexity – a guided tour, both of which have set my wheels to turning.

Of the two, Mitchell’s book is the more accessible. Miller & Page dive heavily into cellular automata, and it takes some real leaps of the imagination to see how any of that applies to social life. One thing that they do discuss (and as does Mitchell) are Stephen Wolfram‘s theories that certain kinds of automata are universal computers. In principle this then suggests that simplified models such as cellular automata or more complicated ABM are widely applicable: that with a conceptual shift, a model may do double duty.  A model about predator-prey interactions becomes a model about shopping malls.

Mitchell also explores by-ways where the distributed, no-central-control system has as an emergent property the ability to compute solutions to problems facing the ant colony, the bee hive, the group, as a whole. Ants explore their world, and leave pheromone trails behind them. Another ant encounters the pheromone trail, leave its own traces, which reinforces the route. Trails that lead nowhere evaporate away; trails that lead to food sources are reinforced, and soon you have a map of all local food sources for the colony.  (or the Tokyo rail network, as the case may be.) No central direction.

Mitchell’s book is also quite good when it comes to explaining how networks figure into all of this, and some basic statistical properties of networks and how they conform/are formed by various behaviors.

Which is sparking a thought in my head, as a route to a new project. In my thesis work, I was able to compute the statistics for a variety of networks in Rome. I was then able to model these networks, to see how decentralized control solved the problem of resource exploitation (in this particular case, building materials to Rome). It would be interesting to compare the statistics of various patronage systems from various cultures at various times around the world. How did these systems solve the problems of resource management? I would then reanimate these various networks in a ‘sugarscape‘ type world. Which ones are more effective? How do I measure effectiveness? Which social configurations are more fit? Then, translating that back into history & culture, it would be interesting to see how that plays out when you’ve got two cultures in roughly the same environment – Medieval Florentine patronage vs Ancient Roman patronage…

Another thought sparked by these readings: in a patronage system where clients have choice over who their patron is – ie, they can insert themselves into the train of a new patron- naturally evolve to a state where there is a single patron.  In this, I’m recasting Miller and Page’s City Formation model (pg 151, section 9.4) in patronage terms.

In this model, imagine a world where all of the available patrons are standing in a ‘police-lineup’ style line in the town square.  The population of the town is randomly sorted in lines in front of each patron.  There are a host of reasons why a client might wish to swap patrons; let’s collapse all these into two variables: displayed support (# of clients the patron already has, hence the desirability of becoming a client) and ‘difficulty’ in switching allegiances – it is hard to gather knowledge about patrons further down the chain to either side, so the further away a patron is in the line, the more difficult it is to get into that chain.  A patron’s success depends on his ability to marshal the resources made available via his client base.

So – behavior. Following Miller and Page, the agents want to have access to the most successful patron.  A client can stay where it is, or move one step to the right, or the left. If the model is run, allowing clients to assess the entire line of patrons, this ‘society’ sorts itself out so that all clients are clients of a single patron. (situation a)

That’s quite neat.

If the ability of clients is restricted so that they can only see locally (a step to the right, or a step to the left), you end up with multiple ‘poles of power’, a handful of patrons with all the clients. (situation b)

So… if we have a society in history characterized by patronage… is it more like situation a, or is it more like situation b? In this model, it’s all about the clients. They choose where to go: so the key dynamic is in how they ‘learn’ about the world. The daily display, in ancient Rome, of a patron walking to the Forum with his clients then is a way through which clients learn about the relative fitness or desirability of potential patrons: thus leading to a single patron….?

…I’m just riffing off the top of my head here, but I think there’s something to this…

ABM: Simulating Roman Social Life and Civil Violence

My paper in a special edition of Digital Studies has just come out. The special issue was edited by Kevin Kee and John Bonnett from Brock University. John and Kevin write in their introduction,

It is a truism in the digital humanities, a constant one, and a good one, that it is always in a state of transition. Such an observation is not surprising since the instrument upon which it relies – the computer – is itself in a state of flux. For the moment, its computational power remains firmly in the grip of Moore’s Law, exponentially increasing its computational power as the decades pass. Scholars, whether they want it or not, are constantly being presented with new paradigms of computing — be it cloud computing, ubiquitous computing, or high performance computing — and new tools and markup schemes to express, treat and analyze content. In any publication devoted to the digital humanities, then, it would seem superfluous to mention that change is our constant condition and our constant preoccupation, a trite observation best left unsaid. We sympathize with this view. But when it comes to describing digital humanities scholarship generally, and computationally supported scholarship in Canada particularly, we think it is wrong. In Canada and abroad, a number of important developments have recently emerged that will impinge on the practice and future trajectory of our inter-discipline. They are new, important, and are reflected in the contributions to this issue. They are of sufficient moment and frequency that we feel justified in rendering this issue of Digital Studies with the thematic stamp it now bears: that of transition.

So I was quite honored to be included in this collection. The abstract from my contribution:

Behaviour Space: Simulating Roman Social Life and Civil Violence

Shawn Graham

Agent based modelling, also known as individual-based modelling, holds great promise for historians as a tool for formalizing and visualizing historians’ understandings of historical processes. It also provides a means to explore exploring the emergent consequences of such assumptions. Such models specify the possible behaviours for a single individual, and then enable individuals within the simulation to interact, with each applying the behaviours in a context-specific manner. Artificial societies begin to emerge from these interactions, allowing us to study their characteristics. Moreover, when these models produce behaviours that cohere with patterns embedded in historical or archaeological data, it becomes possible to interrogate aspects of past experience otherwise lost to us. This article presents PatronWorld, an agent based model of the ancient Roman daily ritual of salutatio, a ritual in which clients gathered to make a morning greeting to their patron. It explores the ritual’s role in the development and maintenance of patronage networks, and its relationship to the emergence of civil violence in the Roman world. The model is also based on a framework that describes the social network surrounding land holding (the foundations of wealth in antiquity) in the City of Rome from the late first – early second century AD. Civil violence was a constant feature of Roman society, frequently targeting the men upon whose social connections the political economy of the state depended. Results from the model suggest the social conditions that made the state vulnerable to periodic bouts of violence, and suggest new directions for further research.

I’ve recently been reading John Miller and Scott Page’s Complex Adaptive Systems: An introduction to computational models of social life (2007, Princeton UP) and there are some models and insights there which are making me rethink the complexity of the model that I created for Behaviorspace. But that’s a post for another day…

(For those interested in novel uses of GIS for historical situations, see in the same special issue,

‘Buried Beneath the Waves’: Using GIS to Examine the Physical and Social Impact of a Historical Flood Abstract HTML
Mathew Novak, Jason Gilliland )

Agent based modelling of the Hispanic Baroque

The Hispanic Baroque project at UWO has quite a complex simulation underway, exploring

[…] the origin, evolution, transmission and effectiveness of the baroque patterns of behaviour and representation in the Hispanic world.

[…] The objectives of the project are: to describe the most common, resistant baroque patterns in different environments; to establish its relationship with processes of social identity and organization; to analyze the technologies of culture that have made this adaptability of the baroque possible; to determine its effectiveness based on the reappearance in Neo-baroque phenomena of the contemporary world; with the participation of other disciplines, to create new tools that fortify investigation methods in the humanities.

It’s a tremendously ambitious project. I had the pleasure of speaking to participants in the project a few weeks ago, on complexity and agent based modeling, detailing my own research and the mistakes/accomplishments I’ve had in trying to deal with cultural complexity in the past.  I’m pleased to say that I’ll be working with these folks a bit more in this new year, though we’re still in the progress of getting-to-know each other. The Hispanic Baroque, as a period, as a worldview, as Atlantic History, is brand-new territory for me. It’ll be quite exciting, I think!

For more on this project’s Agent Modeling work, please see this page. The model itself can be explored here. It appears tremendously interesting, and I suspect there are important insights emerging from this model. However, I would wish that there was a bit more documentation with the applet, explaining the mechanics and ‘things to try’.  The model appears to be linked to this model, with the outputs of the latter becoming the inputs of the former.

Are there other examples of linked agent-models out there?

Dynamic Modeling in a GIS Environment Autumn 2009

Courtesy of Andrew Crooks GIS and Agent Modelling blog , I learn today of a series of seminars exploring the latest in GIS & dynamic modelling, at the Global GIS Academy

Of particular interest (to me, at any rate) are the following:

October 28th

Ling Bian (Buffalo)
A dynamic social network model for disease transmission

The work in this presentation was sponsored by a health care agency and some of the results reported remain confidential until we have permission from that agency to make the presentation available.

See also: Bian, L. (2003) The representation of the environment in the context of individual-based modeling. Ecological Modelling, 159 (2-3): 279-296.

Bian L 2004 A conceptual framework for an individual-based spatially explicit epidemiological model. Environment and Planning B 31(3): 381-95. @ www.envplan.com/abstract.cgi?id=b2833

Bian, L., Liebner, D. (2004) A network model for dispersion of communicable diseases . Transactions in GIS , 11(2): 155-173. @ www3.interscience.wiley.com/journal/118490206/issue

November 18th

Raja Sengupta (McGill)
What’s so spatial about Agent-Based Models?

Download PDF (3.1 Mb) of presentation

See also: Sengupta, R., and Bennett, D.A.(2003) Agent-based modeling environment for spatial decision support. International Journal of Geographical Information Science, 17(1): 157-80

Sengupta, R., Sieber, R. (2007) Geospatial Agents, Agents Everywhere…. Transactions in GIS, 11(4): 483-506.

Derek Karessenberg (Utrecht)
Integrating spatio-temporal GIS data with spatio-temporal models.

Download PDF (1.3 Mb) of presentation HERE (ITEM 4)

See also: Karssenberg, D., Schmitz, O., de Vries, L.M., and de Jong, K (2008) A tool for construction of stochastic spatio-temporal models assimilated with observational data. 11th AGILE International Conference on Geographic Information Science 2008, University of Girona, Spain. 7 pages