2010 Graduate Workshop in Complexity and Computational Social Science

Student Projects

Each student began a research project during the two-week workshop. Below are brief descriptions of these various projects. These projects will form the basis for dissertation chapters and/or journal articles.

Florian Artinger, Adaptive Behavior and Cognition, Max Planck Institute for Human Development (florian.artinger@gmail.com)

Florian is exploring the implications of price heuristics that arise in an online, used-car market. The data suggest that there is far more price rigidity and dispersion than one would expect, a priori. He models this system assuming that buyers follow simple strategies tied to their buyer values, and that sellers use a step-wise pricing function (inspired by the existing data) driven by strategies that determine the initial offer price and a schedule of price decreases over time. He finds that the various strategies can lead to quite different outcomes in terms of individual performance and aggregate market behavior.

Joslyn Barnhart, Political Science, UCLA (joslyn_b@yahoo.com)

Joslyn is developing a basic framework for investigating the diffusion of traits and institutions in international relations across large time scales. States have multiple attributes and influence one another based on network linkages and various interaction rules. She is using this computational laboratory to explore core issues that arise in the international relations literature, such as the democratic peace, differential power dynamics, and the influence of globalization.

Alexander Funcke, Evolutionary Culture, Stockholm University (funcke@0z.se)

Alexander is modeling the dynamics of systemic corruption. In particular, he wants to understand how corruptive norms arise, and are perpetuated, in societies. He models this system using two peer networks, one of citizens and one of bureaucrats, and also introduces asymmetries into the incentives of the two groups. He finds that the resulting system is rich in behavior and provides a variety of opportunities for linking norms to system behavior.

Thomas Grund, Sociology, Oxford (thomas.grund@sociology.ox.ac.uk)

Thomas is looking at the driving forces of homophily, which is the tendency of similar agents to associate with one other. He develops a dynamic theory of network evolution and homophily. His model is driven by two preferences: 1) a desire to associate with similar others and 2) a desire for triadic closure (a preference to associate with your associate's associates). He finds that the addition of triadic closure has differential effects depending on the state of evolution of the network, and that at certain critical points it serves to amplify the degree of homophily. Moreover, this system has interesting implications for the degree of homophily achieved by new agents that enter an existing network.

Navid Hassanpour, Political Science, Yale (navid.hassanpour@yale.edu)

Navid considers the efficacy of democratic deliberation, investigating insights first elucidated by Rousseau and Condorcet. In his work, he explicitly considers the role of heterogeneity and correlation between voters in a network on the ability of the system to make good decisions. Agents have innate accuracies of making the right decision, and linked agents are able to influence each other's accuracies. He finds that the potential of diversity is closely tied to the type of the deliberative process, and depending on the network links it can either help or hurt group decision making.

Dana Jackman, Natural Resources, Michigan (jackman@umich.edu)

Dana is using artificial agents to explore behavior in public goods games. Although the incentives of such games are such that individuals have no, self-interested, incentive to contribute to the public good, human experiments tend to show positive (but declining over time) contributions to the public good. In her first model, there are different agent types (tied to experimental observations) that shift types depending on the social outcome. In her second model she uses two parameters to define the core giving of an agent and the reaction of an agent to the average behavior of others. She is extending this work toward situations in which agents alter their types in response to endogenous conditions arising in the system.

Jasmin Kominek, Social Science, U. of Hamburg (jasmin_kominek@yahoo.com)

Jasmin is refining our understanding of collective decision making and its connections to path dependency. She uses two notions of collective behavior: 1) swarming, when agents follow other neighbors, and 2) herding, when agents follow the average of the group. She finds that the higher the degree of herding, the quicker the system will lock-in. Moreover, there are key trade-offs to the degree of lock-in, that is, to the number of herding types: too many herders result in rapid lock-in but perhaps to a bad decision, while too few will result in very slow social decision making.

Santiago Olivella, Political Science, Washington University (olivella@wustl.edu)

Santiago wants to understand how parties allocate campaign resources across a given district. He uses the transmission of information through a geographic network as a starting point from which to model this process, and then ties the political behavior in the resulting system to concepts developed in network theory, for example, closeness. This approach results in some unique predictions that will be tested using campaign finance and travel data.

Sasha Romanosky, Information Systems and Public Policy, Carnegie Mellon (sromanos@cmu.edu)

Sasha is interested in understanding policy options designed to mitigate the costs of data breaches and identity theft. The goal of policy in this area is to induce firms to take some degree of care with their data, often via mechanisms that impose various degrees of liability or disclosure requirements on each firm. He finds that social costs are minimized when the firms and consumers both share part of the loss caused by the data breach.

John H. Miller , miller@santafe.edu.