On 4-jan-07, at 18:53, David Doshay wrote:
I see it as perfectly fair that the bot with the better ability to read, and thus knows it can pass, should be rewarded for that reading skill.
I think you are mistaken for the real reason of the 'second phase', where he who passes has to pay a point. This 'second phase' only comes into effect after both sides have passed. It's to solve disputes in a fair manner. Since capturing dead stones would cost points, how do you resolve a dispute where your opponent claims his stones are not dead? (Think bent-four corner.)The actual proof consists of playing out the sequence that captures the stones. Every time your opponent passes and you continue playing moves to capture the stones you'd lose a point. That's why passing has to be compensated by paying a point. It's not about Go playing skills that should be rewarded but about being able to resolve disputes fairly.
In the case at hand this phase is abused by a player who doesn't contest the status of stones but instead contests the result of the game when it would be counted according to Japanese rules. Personally I think programming your bot to play inside opponents territory when you obviously know it won't affect the outcome under normal circumstances is showing poor mentality. You'd be wasting my time and/ or computing time. Using the rules used as an argument doesn't hold for me. How would you feel if your opponent played out possible all ko-threats at the end of the game? This is possible without punishment under any set of rules.
In my opinion, the fact that we humans feel bad doing something like that should be enough to at least make an effort to make your program avoid such behaviour too. Unfortunately it seems rather frequent that the opposite is true and that some put effort into explicitly programming such bad behaviour. Personally I htink it's a waste of time. It may win you a game occasionally but it won't make your program play any better.
Mark
_______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
