Author: Sune Fischer
Date: 03:15:51 11/22/02
Go up one level in this thread
On November 22, 2002 at 06:07:36, Uri Blass wrote: >On November 22, 2002 at 05:07:41, Sune Fischer wrote: > >>On November 22, 2002 at 01:24:44, Robert Hyatt wrote: >> >>>On November 21, 2002 at 23:10:06, Edward Seid wrote: >>> >>>>On November 21, 2002 at 22:32:21, ERIQ wrote: >>>> >>>>>Just got an opening book for Phalanx but no opening book for crafty. played a >>>>>game just to see. Maybe I should set the resign alittle lower for crafty >>>> >>>>The problem with Crafty is that if it's playing another computer, it won't >>>>resign. I'm not sure what's the logic behind that design. So in my WB 90 30 >>>>Ladder, I end up waiting and waiting for Crafty to get mated in hopeless >>>>positions. Perhaps Dr. Hyatt can enlighten us... why not make resigning against >>>>other computers an option with Crafty? >>> >>> >>>The issue is this. The other night I watched Crafty vs a commercial program, >>>crafty ended up with a lone king, the other program had a king rook and pawn. >>>It obviously had the krpk table, but apparently not the kqrk table. It advanced >>>the pawn to the 7th and then moved idly around until the game ended in a 50 move >>>draw. All the while crafty's eval was at something like -Mat09 or something... >>> >>>The moral? Don't resign against a computer, you never know _what_ will >>>happen... >> >>Same with humans, you never know when they blunder. >> >>>That is why I don't resign when playing a machine. >> >>I think that is a bad idea if it is a feature the users would like. >>There are a lot of people doing these automated computers tournaments, and >>engines not resigning is a real waste of time. >> >>It is true that occasionally a point will flip here and there because of some >>bug, but Crafty is at a level where most of these silly bugs should been have >>eliminated long ago. Spending a lot of time hoping for a bug in the opponent is >>pretty lame. >> >>Besides if you don't resign, why should any body else resign? It will all be one >>big waste of time, much better to get on with the next game.... >> >>I guess people should keep a history of what programs refuse to resign, and then >>never resign against them either, to give them a taste of their own medicin. >> >>-S. > >If the target is to save time for operators then I see no reason to prefer to >resign and not to adjudicate. Because then it won't be automated. You can always check if an engine resigned in a win/drawn position and inform the author he has a problem. I always play 50 blitz matches, then look at the score. You play in the same time 2-3 standard games, which is much easier to adjucate. >I plan to add to movei an option to adjudicate games based on evaluation but I >see no reason why to adjudicate it only against itself. I see no reason to adjucate at all :) -S. >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.