News Category: ‘Past Friday Seminars’

Friday, 20th November: 3.00pm

Why I am Not a System Dynamicist

Rob Axtell, Department of Computational Social Science

System dynamics models in the social sciences are typically interpreted as a summary or aggregate representation of an autonomous dynamical system composed of a large number of interacting entities. The high dimensional microscopic system is ‘compressed’ into a lower dimensional form in the process of creating system dynamics models. In order to be useful, the reduced form representation must have some fidelity to or verisimilitude with the underlying dynamical system. In this talk I demonstrate mathematically that even so-called perfectly aggregated models will in general display a host of pathologies that are a direct consequence of the aggregation process and which have no analog at the microscopic level. First, a macroscopic model that perfectly aggregates a microscopic system will either not exist or be not unique. Second, with respect to the underlying model the macroscopic system can display spurious equilibria, have altered stability properties, exhibit peculiar sensitivity structure, manifest corrupted bifurcation behavior, and present anomalous statistical features. As a result, there is a definite sense in which even the best system dynamics models are at least potentially problematical, if not outright misrepresentations of the systems they purport to describe. From these formal results I conclude that the majority of such models may have little practical utility in decision support environments where the results of models are used to set policies, although they probably have some value as ‘thought experiments’ in which scientists seek to clarify their own thinking about the coarsest features of specific social processes.

Friday, 15th January: 3.00pm

Simulating Heroin Addiction and Drug Market Behavior Using an Ethnographically Informed Agent-Based Model

Lee D. Hoffer*, Georgiy Bobashev**, Robert J. Morris**, and John Heintz*

*Case Western Reserve University, Department of Anthropology

**RTI International

In this talk I will discuss an experimental collaborative project using ethnographic data to inform an agent-based model (ABM) reproducing the operation of a local heroin market. Ethnographic research is the hallmark of cultural anthropology and involves in-depth fieldwork with a social group to generate detailed understandings of beliefs and behaviors. Extracting data from eighteen-months of participant-observation and in-depth interviewing with heroin dealers and customers in Denver, Colorado, the authors designed a multi-agent social simulation of a heroin market. Incorporating different types of agents buying, selling, and using heroin, the markets adaptability to intervention, the importance of specific market roles, and how the market influenced agent addictions are explored in the model. Although data from the model will be presented, this talk will emphasize the design and development considerations, advantages and limitations in combining ethnography and ABM. To conclude, lessons learned from these efforts, as well as current work integrating ethnographic and quantitative data to build simulations of illegal drug distribution and use will be outlined.

Friday, 22nd January: 3.00pm

Evaluating Plausible Futures of the NextGen in the Air Transportation System

Guillermo Calderón-Meza Graduate Student, CATSR

The Air Transportation System of the United States (NATS) is a complex system that suffers from problems like saturation and limitation in the possibilities of growth. The stakeholders of the NATS have proposed NextGen, i.e. a group of new technologies and procedural changes, intended to solve the problems, make the NATS more flexible, and allow it to grow without saturating. Most of the NextGen concepts have been proposed by experienced people based on their knowledge. Some concepts have been preliminarily researched, but in most cases, the consequences of NextGen are not well understood.

I am presenting results of the preliminary evaluation of tradeoffs that could result from the implementation of NextGen. The first set of results is about the tradeoff between route utilization and cost for the airlines. The Pareto surface was computed using Evolutionary Computation Techniques. It shows that cost and utilization can be controlled and balanced, in a simple system, leading to low costs and acceptable levels of utilization. It was also observed that some strategies (for the airlines) emerge from the experiment that are allow efficient operations in the system.The second set of results is about the interdependence of distance flown, number of conflicts detected, delays, and number of sectors and centers used by the flights. In this case, significant benefits were observed for the airlines and the environment by the reduction of the average distance flown when flights use direct paths and not flight plans. Controllers and safety also benefit with the use of direct routes by the reduction of conflicts. Passengers benefit mildly by a small reductions in the average flight delays. There is still work to do, especially the second set of results could be obtained from a more realistic experiment: on including more flights, and more sources of stochastic behavior.

Friday, 29th January: 3.00pm

Measuring Indirect Costs of Diseases in Economies with Heterogeneous Production Technologies

Bogumil Kaminski*, Maciej Latek** and Michal Jakubczyk*

*Decision Analysis and Support Division, The Warsaw School of Economics

**Department of Computational Social Science, George Mason University

We propose a multiagent model for assessing the impact of illness-related job absenteeism on competitive markets. The model directly measures changes in outputs, prices, technological mix and employment levels that are induced by interactions among diseases, a population of workers and a set of profit-maximizing firms.

Two key ideas drive our model: representing (a) heterogeneous production technologies, including project teams, and (b) multi-disease epidemiology with a dynamic social network that serves as conduit for contagion. When a single team member falls ill, he not only provides less output, but also impacts team productivity negatively. For contagious diseases, clustering of social networks among coworkers increases the probability of having multi-team failures in a single firm. We quantify the amount by which models omitting both effects underestimate the indirect cost of illness.

During the presentation, we compare results for different types of diseases and markets and show which ones are robust with respect to assumptions about (a) decision making schemes by the companies and (b) labor market regulation.

Friday, March 19, 3.00pm

Trade as a Premise for Social Complexity

Anamaria Berea
Department of Computational Social Science

The argument of this research is that trade is an important causal factor for the origins of social complexity and not just a co-evolutionary process as the current literature suggests. Ancient trade has played a crucial role for the origins of the first state level society (Early Uruk) and was a premise for its social complexity between 9000 and 3000 BC. Due to a comparative advantage for clay and water transportation and to a series of technological innovations (the pottery wheel and the tokens system) that enabled traders to travel even 800-1000 miles away, in the first state level society there emerged a series of endogenous institutions that increased the organizational complexity and the economic development of the polity to that of the first political economy. The research is layered in 3 steps: the theoretical framework of Adam Smith, Friedrich von Hayek and Vincent Ostrom, the historical and anthropological evidence of Algaze, Schmandt-Basserat, Owen and Niessen and the agent based model of a proto-political economy. The agent-based model explores the institutional analysis behind this argument in order to understand if the necessary and sufficiency conditions for trade at the origins for social complexity are being fulfilled.

Friday, 2nd April: 3.00pm

Changing Residence in the City: An Agent-based Model of Residential Mobility in Arlington County

Atesmachew Hailegiorgis, Department of Computational Social Science

Residential mobility is complex dynamic process, which occurs due to individual choice of changing residence location. Agent-based modeling is increasingly being applied to model complex dynamic systems due to its potential to simulate individual decision making processes and local interactions in time and over space. The purpose of this paper is to understand how individual agent mobility behavior and decision-making processes affect the over all composition of the neighborhood in the Arlington county using an agent based model. Residential households are considered as heterogeneous agents who differ in their values, characteristics and decision making processes. The model accounts population growth, immigration and emigration. The decision rule of the household agents’ mobility is determined by individual household characteristics, location factors and social bonds. Agents synthesize the available information and respond to the changes of these factors to determine their satisfaction level of the current location. The allocation of new resident location is entirely on the success of the household agent of getting better residential place than the current location. GIS analysis was applied to generate input data for the model using census data and Arlington County GIS database, which is available for scientific research use.. The model is built in JAVA using the MASON library. The model output captures the change in neighborhood composition, change in population number, mobility frequency of agents and density.

Deriving Optimal Blends of History and Intelligence for Robust Counterterrorism Policy

Maciek Latek
Seyed Mohammad Mussavi Rizi
PhD Candidates
Department of Computational Social Science
George Mason University

Counter-terrorism analysis requires that homeland security organizations determine a suitable blend of evidence on the historical patterns of terrorist behavior with current, and incomplete intelligence on terrorist adversaries in order to predict possible terrorist operations and devise appropriate countermeasures.  This paper models interactions between adaptive and intelligent adversaries embedded in minimally sufficient organizational settings to examine the issue of optimal analytic mixture that should be used to design counterterrorism strategy, expressed as historical memory reach-back and the number of anticipatory scenarios. We show that history is a valuable source of information when the terrorist organization evolves and acquires new capabilities at such a rapid pace that makes optimal strategies put forth by game-theoretic reasoning unlikely to succeed.