Programming & Science

Hi -- very short entry, I'm on my way to San Francisco now. I'm in the Vienna Airport, which has free wireless (like any airport should!) This really is the best place to fly from. Except there are still some smoking zones, so maybe in that respect Schiphol is now leading...

Anyway, I got more results this week on the culture model and it is all very exciting, but now I have to transition from the exciting pilot studies to the meticulous data gathering. For the last three years the main expert on doing real science / experiments in my group has been my PhD student Hagen Lehmann who has a recent degree in psychology in Germany and is very good with statistics. But still, he is not a programmer so hasn't really created a system for doing analysis. One of the goals for this sabbatical for me is to really get a system of taking results out of NetLogo and BOD/MASON and crank them relatively quickly through statistics in R & get pretty graphs. Will is helping massively with this. Next month I will take the most important course of his I need to take, on doing regression. He says if I really understand the whole course back to front then there won't be much more I'll need. But he also helps me make pretty graphs too.

You would think this stuff would all be fixed, but it isn't. Will is complaining about this with respect to his paper he's writing now too. The science of doing science keeps moving, as does the mathematics of statistics. We spend time on methodology as well as basic research for practically every paper. Hopefully I can get things more nailed down in the next few months. I also need to get a faster computer so I can run statistically significant sets of experiments in reasonable amounts of time. The KLI have said they will buy one.

Speaking of things taking forever to "get to market", I had more traffic than usual on the AI part of my web page, so I tracked it down and it turns out someone wrote a nice article about my PhD research on an AI & Games website.

http://aigamedev.com/theory/behavior-oriented-design-modular-agent

Gotta go board now...
0