Author: Uri Blass
Date: 12:11:33 07/13/04
Go up one level in this thread
On July 13, 2004 at 14:41:30, Joachim Rang wrote: >On July 13, 2004 at 13:23:52, Amir Ban wrote: > >>On July 13, 2004 at 05:13:55, Joachim Rang wrote: >> >>>Hi Amir, >>> >>>congratulations for the title and many thanks for your report and opinions about >>>the WCCC. Although I'm a Shredder-Fan I must admit that I am always stunned to >>>see with which style and force Junior manages to win. Although it feels like >>>your playing style is more unbalanced and less complete than that of Shredder >>>(you wrote it feels almost "perfect") I think you have discovered something >>>special, which other programs have not. I'm always impressed how Junior operates >>>with its major pieces against the enemy king especially in the late >>>midgame/beginning endgame. I think Junior knows a lot more how to exploit an >>>exposed king when the position is open and queens are still on board than other >>>programs do. >>> >>>Some questions: Does you focus in your work especially on long time controls on >>>superfast machines? I ask since Junior 8 seems relatively better on longer time >>>controls than on blitz in contrast to Hiarcs 9 for example. Do you think that in >>>_evaluation_ there are things that work differently in blitz and in longer time >>>controls (for search it is obvious that there is a difference)? >>> >> >>I don't know of a reason why things would work out differently at different time >>controls. >> >>Amir > > >thank you for your statement. There were some discussions about that, so I >wanted to here the opinion of the World Champion creator ;-). Of course for >search I find the theoretical explanation most likely that things work >differently on different time controls, since a lot of extensions make your >program tactically stronger in blitz time control but negatively affect the >branching factor, which hurts in longer time controls. > >Good luck with further progress and thanks again > >regards Joachim There is clearly a logical reason that things can work different at different time control. Imagine that you have a type of positions that your program can get 90% from it at long time control but at blitz you get only 50% from the same type of positions. How are you going to evaluate it? If you play blitz then 0.00 may be the best evaluation because chances are equal. If you play long time control then it is logical to evaluate it as a clear advantage for yourself. Uri
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.