Complex Systems Summer School 2000

Project: A Three Part Survey

My primary goal here in CSSS is to be a "scout" for Sun Microsystems in the complexity area; to ferret out if and how complex systems will soon be of use for us, either in products and services, or in how we manage the company. This includes looking into whether or not Sun labs should budget for a small complexity project of some sort.

To focus a bit, I want to limit the investigation to software issues, even though hardware concerns are clearly important for us as well. In addition, I would like to restrict myself to Java based tools and implementations so as to be multiplatform and to possibly uncover weaknesses/issues with Java for complexity usage.

Due to the breadth of this goal, I've decided upon a three part survey.

1 - Java Programming

First, directly implement in Java one or two modest programs which deal with complexity. Possibilities are:

Simple agent system.
Simulate and analyze a dynamic system: dual pendulum or damped driven single pendulum.
Neural Net: Do the XOR problem.
Genetic Algorithm

The key issues here are whether or not Java can be effectively used for complexity investigations and are there improvements needed for Java, either in the language or classes. Possibly I should send out a questionnaire to the students/faculty mail lists to get suggestions as well.

2 - Tools and Utilities

Second, use existing multiplatform (hopefully Java based) tools to investigate complexity

StarLogo: Do several of Ronnie's homework examples.
JNet: Build a simple hidden nodes net.
<other packages?>

Key issues here are just how mature are packaged solutions and how multiplatform are they.

3 - Complexity Topics in Computers and Networking

Finally, survey the field to see where computer and networking complexity topics lurk, and define/begin a complexity-in-computing project. This may result in the final goal above, a project for the lab.

Autonomous computing: routing, name serving, file serving, and so on, services without the need of servers. The latest Gnutella for example. Or Napster (which does, however, use a server for coordination only)
Computers as community: using tools like Small Worlds and 6-degrees of separation as a tool for understanding both computer users and communities, and sets/communities of computers.
Cooperating autonomous servers: Cache and proxy servers collaborating with very simple rules to provide robust services. Example: LOCKSS

http://highwire.stanford.edu/~vreich/LOCKSSusenixpaper.pdf

In particular, the "Small Worlds" paradigm offers an interesting way to analyze some of these. The goal here is to gather and understand the ways in which complexity applies to computing and networking itself, and to define and start on a project.