I usually do a write-up of my participation in the ICFP Contest each year, but this year, Dana did such a good job with the web page I didn’t bother to mention it. I think I was also just being lazy, but eh, who really knows.
Anyway, I had no idea how well we’d do, as is usually the case with ICFP. However, the scores shown so far have been pretty promising.
It seems like we’ve been pretty much middle of the road throughout the heats, rising slightly in the end. Look for Five of Six in the last heat to get a sense of where we stand.
The last heats probably won’t be announced until ICFP actually happens in a few weeks, and I totally don’t expect to win. However, it’d be neat to see us get in the top 10-15%. The code is worth checking out. Jia and Ed had worked out a more reliable mathematical model by the time the contest ended, but we didn’t have time to implement. Our submission actually uses something James, Ed, and Caleb cooked up which is a bit naive, but apparently serving us pretty well!
The real innovation is James’ voting system, which basically allows us to hook up modules and assign them a weight in determining what actions should be taken next. Here’s an example of one of the thoughts I wrote:
Anyway, this post is a bit too little, too late, but seeing that they posted partial scores got me all excited and I needed to at least give it a mention. It’s pretty cool that our team was able to use Ruby this year without very many performance issues and still end up in the top 25%. I hope to see more Ruby hackers join the festivities next year, as the code that comes out of ICFP is always chock full of learning experiences.
Written by Gregory Brown on 2008.08.12 at 23:18 | Responses