Author: Robert Hyatt
Date: 13:42:38 04/24/00
Go up one level in this thread
On April 24, 2000 at 16:33:18, blass uri wrote: >On April 24, 2000 at 16:04:52, Robert Hyatt wrote: > >>On April 24, 2000 at 14:42:36, A.L.Mourik wrote: >> >>>Hello to everyone, >>> >>>Two days ago I played a Nunn 2 match between Fritz 6 and Crafty 17.10. >>>This whether to check if Crafty as supposed by e.g Jouni Uski was really as good >>>as been told . >>> >>>The result in a Nunn 2 match that I held was disappointing because it ended in a >>>clear 29,5 - 10,5 victory for Fritz6. >>>Timelevel 5min +3 sec. PII 400 each 8 MB HT, pondering off. >>>The only reason that I played a Nunn 2 Match between these two engines, was to >>>TEST which program under equal circumstances, as in this case in the Fritz6 gui, >>>scored the best. This as an indication of the ?chesspower? of that engine. >>>Imho these are the best circumstances with one computer to tell which program >>>produces the best chess. >>> >>>You can?t argue on the fact that the circumstances for the given engines were >>>not the same. >>>As far as I can see it, your only real point is the presumption, that Crafty >>>does a better time saving in the pondering on mode, than Fritz6. >>>So I agree that with testing on two different computers and with both programs >>>pondering on, this presumed negative effect for Crafty would be neutralised. >>>On the other hand I wonder what Morsch or other respectable programmers opinion >>>on these presumption is. >>> >>>I would like to end this discussion with quoting Cock de Gorter who has written >>>in his article on Kasparovchess.com >>> >>>"Crafty is extremely strong in tactics, but it makes positional errors and >>>misplays some endgames. It doesn?t understand piece mobility. Despite its high >>>rating on the Internet, Crafty is a weak program compared with the best >>>commercial chess programs." >>> >>>I agree on that but still I must say that I am very glad to have Crafty for >>>free. >> >> >>That only shows that de Gorter doesn't know his elbow from a potato chip. >>Crafty is actually significantly _worse_ at tactics than commercial programs. >>Check out the tactical scores on _any_ test suite you care to try. > >Crafty may be worse in tactics but test suites cannot tell which program is >better in tactics because they usually position when the right move seem bad and >you can only discover after search that it is good when many tactical mistakes >do not give the opponent a move that seems bad and turn out to be good. > >I think that a better test should be based on games when the target is to >discover more tactical mistakes. > >Uri Win at chess, and ECM weren't constructed to be obscure. They are simply positions chosen to highlight a particular tactical idea or theme. I agree that the anti-null-move positions and such nonsense are useless... but plain tactical tests do give insight into what the engine can see within a given time limit. Mine has never done very well here, other than it sometimes looks good due to super-fast hardware...
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.