Author: Howard Exner
Date: 07:49:05 12/14/97
Go up one level in this thread
On December 14, 1997 at 09:20:40, Tony Hedlund wrote: >On December 14, 1997 at 07:25:39, Ed Schröder wrote: > >>>Posted by Amir Ban on December 14, 1997 at 04:50:45: >> >>>In Reply to: Re: Rebel8 and Rebel9 on SSDF. posted by Ed Schröder on >>>December 12, 1997 at 13:19:03: >> >>>On December 12, 1997 at 13:19:03, Ed Schröder wrote: >> >>>>>Posted by Tony Hedlund on December 11, 1997 at 14:32:21: >> >>>>>In Reply to: Re: Rebel8 and Rebel9 on SSDF. posted by Ed Schröder on >>>>>December 10, 1997 at 14:32:23: > >>>>In Holland we use the expression "nobility obliges". To me you can't say >>>>in public: "No comments. I haven't time to play through the games." >>>>while >>>>the game in question is a draw but counted as a loss for Rebel. >> >>>>That's not good for the image of SSDF. >> >>>>- Ed Schroder - >> >>>Didn't Rebel resign the game in question ? If so, there's nothing more >>>to do or say. > >I got that impression to. The draw was a human opinion, wasn't it. > >>No, resign is standard set to OFF. >> >>My critic is not about the half point Rebel9 missed but a general critic >>how to deal when errors are found. Similar mistakes could also be made >>with other programs. > >What mistake would that be? 1k6/8/P1K2p2/8/3BP3/6b1/8/8 b The mistake was to not let the two progeams,Junior4 and Rebel9, battle it out from the above difficult position. This position would surely have drawn a protest if it were stopped in the recent WCCC in Paris. I sent this position to rec.games.chess.analysis (as promised) and received some excellent feedback from posters named mig and ikastan. There is a win for white here(as explained by ikastan) 1. Capture the pawn on f6 and then reach the following position: k7/8/P7/2BK4/4P3/6b1/8/8 w This for a computer program would be challenging but not impossible. Now from here the program must play accurately to win. Bd6! Bh4 e5 Ka7 Kc6 Kxa6 Bc7 B moves then from here the sequence e6, Kd7, Bd8 and so on will queen the pawn. The point in all this is to let the programs fight it out. This is not quite as bad as the recent post of Ferret's speed chess win over Rebel 9 where Rebel's operator made what he thought was a better move than Rebel but turned out to be the losing lemon move. >The game above or the rebel.mvs mistake. The >last >one I agree on. > >>I have absolute trust in the intentions of SSDF and their good work but >>they should correct errors when found. > >Yes, but depending on the consequence of the error. > >>>If I remember correctly, Rebel's autoplayer resigns at -5. It also does >>>the "courtesy" of resigning for the opponent at +5, something that may >>>save time, but it really shouldn't do (imagine Rebel playing the other >>>side here, and deciding it's won). >> >>Then you should run Rebel with the "A" parameter. So REBEL.EXE A >>This will bypass the +5.00 / -5.00 check and also will bypass the double >>game check. >> >>If I am right SSDF only test Rebel with the "A" option. > >Yes, this is correct. > >>- Ed - >> >> >>>Amir > >Tony
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.