Watching random players play Go on a 203x203 board at 800 moves/second is mesmerizing.
-
-
Replying to @peterseibel
@peterseibel Where can I find this magic thing?2 replies 0 retweets 0 likes -
Replying to @meangrape
@meangrape Note, also, that this is playing "ancient-style" Go where the goal is to actually fill in the board, not just surround territory.1 reply 0 retweets 0 likes -
Replying to @peterseibel
@peterseibel@meangrape ah, there's my answer to the scoring question I guess.1 reply 0 retweets 0 likes -
Replying to @avibryant
@avibryant@meangrape Right. Except for a few edge cases, the ancient system and the various modern scoring system give the same result.1 reply 0 retweets 0 likes -
Replying to @peterseibel
@avibryant@meangrape But still no fitness function because I haven't actually gotten around to implementing the critter part.1 reply 0 retweets 0 likes -
Replying to @peterseibel
@peterseibel I have some Monte Carlo Tree Search code (not written for Go) that would be fun to adapt to your board interface at some point.1 reply 0 retweets 0 likes -
Replying to @avibryant
@avibryant Does that require that you have kind of evaluation function for non-terminal positions?1 reply 0 retweets 0 likes -
-
Replying to @avibryant
@avibryant Ah, I must have misunderstood the thing I spent 30 seconds reading. Next time you're in SF we should get together to hack.3 replies 0 retweets 0 likes
@peterseibel you build a sparse tree of all possible moves depth first, and evaluate internal nodes by how many of their leaves won vs lost.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.