Presentation Materials

In this section you can find the slides from several of the presentations I've given over the years. In addition to these presentations you can go to my presentation section of ComplexityBlog for my lectures on building ABMs presented the ICPSR Complex Systems Computer Modeling Tutorials (from 2005 to 2009). Slides from my presenations will be added here as they are created.

Title: Famed Santa Fe Presentation
Date: July 2005
Objective: Santa Fe Institute Summer School
Description: I worked on a project about power law distributed group sizes in hunter-gatherer societies but I mistakenly chose to implement it in (an early verson of) Netlogo. As a result I didn't have enough time to complete modeling the phenomenon I was working on. I became an expert in Netlogo in the process, so that was good, and I also created this hilarious presentation. Without the monologue I'm sure it loses some of it's magic, but it's still funny to look through as is.
Presentation:  PDF Slideshow

Title: Frontiers of Agent-Based Modeling
Date: February 2006
Objective: CSAAW Presentation
Description: This presentation is a call to arms given to a group of graduate students already self-described as complex systems people. So no convincing was necessary, but since the field does not enjoy widespread acceptance and is underused as a result I provide a roadmap for agent-based modelers to vindicate the field. Some points highlight the weaknesses and provide guidance for improving on them. Other points demonstrate its strengths and how to focus on them. And still other points aim to narrow the perceived gap between traditional modeling and ABMs.
Presentation:  PDF Slideshow

Title: World of Complex Systems v1
Date: December 2006
Objective: Oberlin Modeling Workshop
Description: This presentation was to a group of undergraduates and professors at Oberlin College as part of their initiative to start an intensive modeling program there. Three of us came down from UM to give a series of talks and this one is my general introduction to agent-based models and complex systems.
Presentation: HTML Presentation

Title: World of Complex Systems v2
Date: July 2007
Objective: Lockheed Martin Tech Talk
Description: Expanding on the previous talk, I include more material for a longer presentation to a more technically sophisticated group of mostly professional engineers and computer scientists. The main thrust is the same: introduce complex systems thinking and agent-based modeling to a group of people not already familiar with it. Further refinements have been made and a new slicker talk on similar material is upcoming.
Presentation: HTML Presentation

Title: Introduction to Genetic Algorithms
Date: July 2007
Objective: Lockheed Martin Tech Talk
Description: This is another talk given to a group of engineers and computer scientists. I was asked to give a talk on this subject, but not given much time to prepare it. Nevertheless the presentation has some value and may be of interest to others who need to initiate a group into the benefits of this style of research.
Presentation: HTML Presentation

Title: Causation Presentation v1
Date: April 2008
Objective: Lockheed Martin Tech Talk
Description: I wrote a philosophy paper on Models of Science and the Role of Causation and it has several implications for science, modeling, and interpretation. I highlighted these aspects of the paper and presented them at work. It was quite a bit different than what people there were used to hearing presentations about, but I received several excellent questions and some valuable feedback.
Presentation:  PDF Slideshow

Title: Causation Presentation v2
Date: October 2008
Objective: Philosophy Graduate Student Workshop
Description: Based on the same Models of Science and the Role of Causation paper as the above presentation, but with a very different focus, I presented my ideas to a group of philosophy graduate students who tore it up and left me with lots of great ideas and things to work on. That paper has now been split into three projects, and two of them are still active (one is dead). I might soon run out of opportunities to give philosophy talks unless I begin submitting to conferences, so I'm polishing this materal for conference submission and (maybe) some kind of publication.
Presentation:  PDF Slideshow

Title: Measures of Tipping Points, Robustness, and Path Dependence
Date: May 2009
Objective: CSAAW Conference, Physics Grad Student Workshop, Swarmfest,
AAAI Fall Symposium, Chinese Academy of Math and System Science
Description: My most recent and certainly best piece of methodology work is presented here in sketch. The technique itself is too detailed to explain thoroughly in a 20 minute talk (and these slides are really for a 40-minute talk), but an introduction to the technique is accomplished as well as providing some of the most useful definitions. This paper will also be divided into several smaller papers, but since they build off each other it will be difficult to present the more advanced techniques without first presenting the foundations every time. Still, more experience presenting this material will result in more efficient presentations for the future.
Presentation:  PDF Slideshow

Title: Introduction to ABMing with Netlogo
Date: July 2009
Objective: EITM Workshop Presentation
Description: This presentation introduces Thomas Schelling's famous segregation (aka tipping) model and demonstrates how to extend the model in various way. It is meant to demonstrate how easy it is to make useful qualitative changes to existing research with just a little Netlogo knowledge. Each extension is motivated, demonstrated, and the code as well as the results are provided. In additoin to this slideshow there is an accompanying HTML file that includes an applet of the base model, the documentation, and links to download each extended version (with extensive commenting to describe the changes and spur students for more).
Presentation:  PDF Slideshow

Title: Measuring System Dynamics with a Temporal Web
Date: October 2009
Objective: CSAAW
Description: In order to measure properties of a system�s processes the current methodology captures the dynamics in a temporally extended network. Static network measures applied to this �temporal web� reveal properties of the processes operating in a system, i.e. dynamic properties. Applying various well-known network measures such as betweenness, clustering, and diameter, provides us with information about the process� path dependence, efficiency, and percolation threshold among other things.
Presentation: Can't Find Presentation File Right Now

Title: Measures Of Dynamical Properties
Date: January 2011
Objective: Job Talk
Description: The goal of this project is to formally define and provide measures for several distinct patterns of system dynamics. Complex systems thinking shifts the analytical focus from equilibrium and explaining outcomes to system behavior and explaining processes. However we are still largely incapable of capturing these dynamical features with general, quantitative measures. In response to this need I present a technique to represent system dynamics utilizing a specifically constructed Markov model. This weighted graph representation captures system behavior in its various structural features, patterns, and probability mass flows. First I provide formal definitions of the foundational concepts required (attractor, reach, perimeter, etc.). These are then used to construct definitions for multiple tipping point-related concepts and methods to measure their magnitudes. Building upon the tipping point definitions I formally define and create measures for robustness, resilience, sustainability and similar concepts. The versatility of the Markov model representation combined with the concise definitions fosters comparisons of models at different scales, for disparate applications, and across disciplines.
Presentation:  PDF Slideshow

Title: Testing for Emergence
Date: November 2011
Objective: AAAI CAS Symposium 2011
Description: Complexity science attaches considerable import to the concept of emergence, yet there are no established, unambiguous criteria by which we can determine whether a phenomenon counts as emergent. There exists several loose characterizations, and a few more specific desiderata, but in the current state of understanding describing a phenomenon as ``emergent'' establishes no evaluable claims and imbues no enhancement to our understanding. Rather than attempting to support a single ``correct'' definition of emergence, the goal here is to distinguish various distinct types of emergence. To achieve this end the paper provides a suite of tests by which models can be shown to exhibit (or fail to exhibit) each property indicated by an emergence claim. These exercises will help us sort out the many attributions of emergence and match them to specific, substantive claims -- a process that will demystify emergence and make it safe for science.
Presentation:  PDF Slideshow

Title: Evolution of Prosociality via Preferential Detachment
Date: December 2011
Objective: Dissertation Defense
Description: The current project introduces a general theory and supporting models that offer a plausible explanation and viable mechanism for generating and perpetuating prosocial behavior. The proposed mechanism is preferential detachment and the theory proposed is that agents utilizing preferential detachment will sort themselves into social arrangements such that the agents who contribute a benefit to the members of their group also do better for themselves in the long run. Agents can do this with minimal information about their environment, the other agents, the future, and with minimal cognitive/computational ability. The conclusion is that self-organizing into groups that maintain prosocial behaviors may be simpler and more robust than previously thought. The primary contribution of this research is that a single, simple mechanism operating in different contexts generates the conceptually distinct prosocial behaviors achieved by other models, and in a manner that is more amenable to evolutionary explanations. It also bears importantly on explanations of the evolution of our moral experiences and their connection with prosociality.
Presentation:  PDF Slideshow

Title: Methods for Hypergraphs and k-Partite Graphs
Date: May 2012
Objective: Human Complexity Conference
Description: Network theory has become a common tool for modeling complex systems and exploring complex relational data, yet most of this work utilizes flat unimode dyadic graphs in which the nodes and edges are all of the same type, and each edge connects exactly two nodes. This talk introduces the audience to the tools necessary to utilize the more complicated network forms of k-partite graphs and hypergraphs. After some preliminary material on the various types, variations, and features of network models, this talk presents foundational concerns in extending existing measures and algorithms to be suitable for k-partite graphs and hypergraphs. By making these methods readily available to the broad research community of network modelers and analysts I hope to improve the fidelity of representation, the accuracy of analysis, and the quality of insights gained for the appropriate applications of these more complicated network forms.
Presentation:  PDF Slideshow

Title: Data Analysis for Agent-Based Models
Date: June 2012
Objective: NEH Worskhop
Description: This talk is part of an introduction to agent-based modeling that presents data collection and analysis considerations. First I present how to set the random seed in Netlogo, how to use the Behavior Space tool in Netologo, and how to recreate a run from the stored random seed. Then some general advice on how much data to collect from your model including reasonable limits to data collection. The analysis of simulation data is still in its infancy, but I present some best practices in data analysis including: don't aggregate your data, use non-parametric tests, and measure dynamics rather than just changes across time slices.
Presentation:  PDF Slideshow

This little thing marks the end of the page.