Author: pete
Date: 14:31:39 06/25/99
Go up one level in this thread
On June 25, 1999 at 15:39:39, Will Singleton wrote: > >On June 25, 1999 at 15:20:30, greg moller wrote: > >>On June 25, 1999 at 13:50:16, Will Singleton wrote: >> >>> >>>On June 25, 1999 at 13:09:04, Lawrence S. Tamarkin wrote: >>> >>>>I thought a fictitious Harvard Cup would involve picking which humans & >>>>computer's would be in it, basing it on the setup that was traditionally used in >>>>past Harvard cup events. >>>> >>>>My picks would be, >>>> >>>>Human's: Playing programs: >>>> >>>>1. Benjamin 1.Shredder (of course) >>>> >>>>2. Rohde 2.Fritz5 >>>> >>>>3. Christiansen 3.Hiarcs7.32 >>>> >>>>4. Yermolinsky 4.Nimzo99 (or 2000) >>>> >>>>5. Fedorowicz 5.CM6000 (of of course) >>>> >>>>6. Gulko 6.CS Tal II for Windows (A real wild card) >>>> >>>> >>>>mrslug - the chess software addict! >>>> >>>> >>>> >>> >>>Re your choice of CSTal, my results indicate it's about as strong as my program, >>>perhaps a bit stronger. This is from ICC blitz play. What indication do you >>>have that it belongs anywhere near the top programs? >>> >>>Will >> >>In Thorsten's _slow_ tournaments Cstal has shown relative strength comparable to >>all the top programs. Of course, due to its provocative style there's no reason >>to think it won't do even better against humans. >> >>ICC blitz play is a poor indicator OTOH, IMO. :) >> >>regards, >>gm > >Probably. But then, how do you explain the fact that of the 4 CStal accounts on >ICC (that I know about), none of them has a high standard score? All of their >scores are less than their blitz scores, way less than top rated computers, and >either about the same as mine or less? (talmoves, master-tal, redbear, cstalx) > >Would also be nice to see the results of some independent testing, perhaps Shep >is doing so? > >Will I think there are several reasons : 1. All Tal-accounts seemed to play nearly everybody and maybe except talmoves never seemed to go for rating points by challenging the popular "chessprog victims " or put a " no computer " which can give you the hell of a rating :) 2. It seems that in the sessions where a strong human master played tal for several games ( IM or GM ) , it was easier for the human to adapt to tal style than to other progs's style IMHO 3. It is _very_ difficult to find strong opponents for standard games . Most good players seem to prefer blitz. 4. In standard games against other chessprogs tal played quite well . I think it gets 30-40 % against the major crafties although there is a significant hardware advantage. 5. Your prog is not _that_ bad and a thrilling opponent to most computer accounts :) Tal plays quite good at really slow time controls ; the speculative functions bring about interesting moves ; some are good and some are bad ; if it gets enough time it usually manages to search out the better ones . But who would play a 2h/40 match over Internet ? I watched several games : sometimes it outplays great human or computer players like a "supermonster" , sometimes it plays like a beginner , you never know :) Pete
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.