Analyzed game #1621344 Alphongue - Malaschitz Go forum
5 replies. Last post: 2014-06-04
Reply to this topic Return to forumReturn to forum
You have 0 new messages
You have 0 games on move.
You have 0 invitations to game.
5 replies. Last post: 2014-06-04
Reply to this topic Return to forum[game;id:1621344;move:10;title:c5 (0.54) is better than b5 (0.46)]
[game;id:1621344;move:19;title:g5 (0.50) is better than e4 (0.38)]
[game;id:1621344;move:23;title:h3 (0.34) is better than g3 (0.23)]
[game;id:1621344;move:26;title:g6 (0.84) is better than h5 (0.64)]
[game;id:1621344;move:29;title:g8 (0.41) is better than h8 (0.23)]
[game;id:1621344;move:30;title:g8 (0.77) is better than j8 (0.38)]
[game;id:1621344;move:33;title:j5 (0.56) is better than f8 (0.44)]
[game;id:1621344;move:34;title:h3 (0.56) is better than d8 (0.47)]
[game;id:1621344;move:35;title:h3 (0.53) is better than h3 (0.35)]
[game;id:1621344;move:36;title:j5 (0.65) is better than f9 (0.22)]
[game;id:1621344;move:37;title:j5 (0.78) is better than g9 (0.32)]
[game;id:1621344;move:39;title:j5 (0.33) is better than j4 (0.05)]
[game;id:1621344;move:40;title:j5 (0.95) is better than h9 (0.70)]
[game;id:1621344;move:44;title:e1 (0.77) is better than f1 (0.05)]
It seem that the bot uses a wrong notation of the board: recommendation for move 35: h3 (0.53) is better than h3 (0.35).
Could this be fixed by the programmer(s)?
@varcos: Indeed, my original comment was meant to point out a possible bug, but my theory is that it is not a notation problem. I've had the experience with some of my game programs that in certain positions different runs of a MCTS program can give quite different win rate results. I'm suspecting that might be at the heart of the suggestion on move 35, but since we don't know exactly what he is doing, it's hard to know. Petr, are you willing to let us know just what did happen there? :)