Griffith Rees

About

I'm a sociology Phd student working in the CABDyN Complexity Centre at Oxford University. I'm studying the dynamics of a social system, with the help of a bunch of physicists. I also direct plays, play in jazz bands (drums/guitar), and try to go sailing whenever I can. Hopefully I'll finish coding a website for myself before the summer school starts.

I'm actually from the suburbs outside Washington DC. I went to Oxford for my undergrad and still haven't managed to leave.

You can email me at griffith.rees@sociology.ox.ac.uk. Please do, especially if you'd like to work on one of the projects I've suggested!

Main Interests

Generally social system dynamics. How do people move in crowded spaces? What mechanisms explain simple, everyday behaviour cascades? What underlying patterns or mechanisms explain macro-social phenomena?

On a more specific level: How do communities self-organize, especially without an initial, well-defined hierarchy? How do open content communities sustain themselves, and in general how do communities solve free-rider problems?

Expertise

I'm pretty good with python, in particular django, scipy, and networkx, and occasionally the python bindings to igraph. I dabble in R and have recently become obsessed with processingjs: an html5/javascript implementation of the Processing language. Hooray for webstandards! I'm also reasonably adept at various unices and live in gnu screen, zsh, vim and git.

I unlike some of the proper scientists attending, I have a background in sociology, economics and philosophy, which is useful if you want to study social systems (but little else). Somehow physicists seem to be able to apply their work to other disciplines, while Sociology is unlikely to contribute to Physics.

I'm pretty good at data scraping and storage via django's object relational mapper. I've also done some work with geographic data, generally via the excellent geodjango project.

What I hope to get out of CSSS

I'm a sociologist who last took proper math and computer science courses in high school, so basically all the stuff I do now I've had to pick up myself (for a visual explanation see this). I'm really hoping to learn some probability theory and stats, and how best to design stochastic models. I'd also quite like to pick up a bit more programming, especially agent-based modeling and visualizations.

Perhaps more importantly I'd like to see how the natural sciences approach complex systems, in the hope that I can drag some of those methods into the social sciences, or at least my own research.

Perhaps more optimistically, I'm looking for people interested in working on social systems with me, and maybe to write something simple together over the 3 weeks (or at least start!).

Possible Projects

There's so much data out there on social systems waiting to be analyzed! And most of it is better than what sociologists and economists collect themselves.

Sorry the following is a brain dump. If anyone's interested at all do email me.

• Just thought of this today: how do trends in pop culture change and cycle over time? Imagine pulling in a list of the 'top 100 music videos' as defined by MTV over the last 30 years, and looking for cyclical patterns. How soon will music from the 60s be popular again?

• Are there network externalities in social systems, meaning for every new member there is a non-linear increase in activity, perhaps corresponding to Metcalfe's Law or Reed's Law (or perhaps something smaller, see this)? What is the ratio of users to activity in various social systems and is there some pattern and commonality across different types?

• What is the ratio of contributors to users in open content communities like Wikipedia, open source projects and bittorrent, and how does the interaction between users and contributors sustain such communities?

• What is the impact of decentralized source control management (git, mercurial, and bazaar) on the productivity of open source projects and how does the community structure differ between subversion style projects and decentralized projects?

• Develop a visualization package based on processingjs and JSON that allows dynamic data analysis with a simple webserver backend via AJAX. Then you can use it directly on the web and even run in a presentation (if you do presentations in an SVG format anyway), and then plugin whatever server backend you like (be it python, ruby, php etc.) as long as it can serve JSON. I've already got a short package that'll graph a network. If anyone's interested I can stick my code up on github so people can start hacking before we get to SFI.