I've been lurking on the list for a while, thinking about how to make a contribution to computer go in the limited time I have. Perhaps others are in the same situation?
It seems like there is still a fair amount of reinventing the wheel that needs to be done just to get started on a go engine. I downloaded and compiled the libego library and was thinking about what to do next, and it seems like the next hurdle to overcome is writing good regression tests. Playing against other engines lets you know overall strength, but I suspect it's still hard to diagnose bugs. Rather than each person reinventing regression testing in a way that works for his or her platform, maybe it would be useful to have a go problem server specific to computer go. Your program could connect and try a large number of problems taken from the collections that are already freely available, and it would keep stats on how well each program did on each problem. This would serve as a public, shared, platform-independent regression suite, and it would be interesting to see which engines do well on which kinds of problems. Does this sound like a worthwhile project? What else has been done along these lines? - Brian _______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
