Author: blass uri
Date: 07:54:43 04/16/00
Go up one level in this thread
On April 16, 2000 at 08:31:44, Alvaro Rodriguez wrote: >On April 16, 2000 at 08:24:08, blass uri wrote: > >>On April 16, 2000 at 07:08:05, Alvaro Rodriguez wrote: >> >>>On April 16, 2000 at 03:25:01, blass uri wrote: >>> >>>>On April 16, 2000 at 02:11:59, Chessfun wrote: >>>> >>>>> >>>>>Maybe settings should be called CM6K Jorge less confusion at >>>>>a much later date with some CM8K settings. >>>>> >>>>>Anyway I read the post and stopped the current 100 game match >>>>>I had running at 2hrs/40 moves 1 hr rest between CM6000 and CM6666. >>>>> >>>>>I have thus far played 10 games 1 hr/side game. >>>>>Score CM6666 W3 L5 D3 CM8888 W5 L3 D3 looks like CM8888 will also >>>>>win game 11. Just did haha. >>>>> >>>>>For me I have no interest in what is stronger at 15 mins I am more >>>>>interested in longer games. Besides which for it to prove to be stronger >>>>>than CM 6666 would require a lot more games. For it to prove to be stronger >>>>>against other engines is another matter completely, beating CM 6666 don't >>>>>mean it will do better against Fritz and the like. Though I did read it had >>>>>scored well again the number of games still leaves a large margin of error. >>>> >>>>I think that the different values for qhite and black queen is illogical. >>>>It may help against other programs because they do not understand that counting >>>>pawns is not important in queens endgames with passed pawns but I do not expect >>>>it to help against chessmaster personalities. >>>> >>>>Uri >>> >>>I think that the important thing here is to performe well against other >>>programs, not against cm personalities..IMHO >> >>I think that something that does not work well against other personalities may >>not work well against other programs in the future. >> >>>I agree that the queen value seems a little bit illogical, but what if it works? >>>We just have to wait and see.. >>> >>>Alvaro >> >>I explained that the reason that it works may be the fact that other programs do >>not understand that counting pawns is not important in queen endgames with >>passed pawns. >> >>I believe that when programmers will improve their programs it will stop to be a >>good idea. >> >>Uri > >When you say that "when programmers will improve their programs", what _exactly_ >will they improve mattering this setting? I mean, will they tune the program >against this setting? > >Alvaro No I mean they will change their evaluation function to a better evaluation function. I saw many cases when the evaluation function of programs is wrong in queen endgames. programs often do not understand that some KQP vs KQPP endgames are clear advantage for the KQP side because only this side have chances to promote the passed pawn to queen when the chances of the side with the material advantage is only to draw by perpetual check. The same is if the side with the dangerous pawn is 2 pawns down because counting pawns is not important. 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.