Author: Laurence Chen
Date: 06:52:30 12/13/98
Go up one level in this thread
On December 13, 1998 at 04:47:38, blass uri wrote: > >On December 12, 1998 at 15:58:39, Laurence Chen wrote: > >>In the past post I have been trying to point that CM uses static features as a >>way to measure and evaluate the position. Some of you scorned at my post saying >>that it is not so, either because you don't know what static features mean. If >>you don't then I recommend you to get a copy of Silman's book How to Reassess >>Your Chess, 3rd Edition, or Simple Chess by Michael Stean. Now the debate about >>that CM is the strongest engine is a open one, I simply don't care to own the >>strongest engine per say, I would be happy with a chess engine which is in the >>top 10 of the SSDF list. I've been using CM since the first one came out, CM >>2000, and one of the things I noticed in that engine is that it uses static >>features of chess position to evaluate the position. There is nothing wrong with >>that, without the exception that not all chess positions can be measured in that >>way. Some of you CM supporters have claimed that CM analyses the position >>accurated because you people believes that CM is the strongest engine and >>concluded that it must be so. Wrong, wrong, wrong. It seems to me that we are >>going back to the time of Tarrasch and Nimzowitch, the classical school vs. the >>dynamic school of thought. CM plays very much like the classical school, and >>that was what I tried to point out that such a style is "boring" and "dry" to >>me. Like everybody else, everyone has its own likes and dislikes. I prefer chess >>positions which are dynamic in nature, and I find that CM is weak in making >>assessments in such types of positions. As an example take a look at the >>following position, I give you the game moves. >>1. d4 d5 2. c4 c6 3. Nf3 Nf6 4. Nc3 e6 5. Bg5 dxc4 6. e4 b5 7. e5 h6 8. Bh4 g5 >>9. Nxg5 hxg5 10. Bxg5 Nbd7 11. exf6 Bb7 12. g3 c5 13. d5 Qb6 14. Bg2 O-O-O 15. >>O-O b4 16. Na4 Qb5 17. a3 Nb8 18. axb4 cxb4 19. Be3 Bxd5 20. Bxd5 Rxd5 21. Qe2 >>Nc6 22. Rfc1 Na5 * >>In this position, CM6000 after analysing the position for 12 hours give the >>following assessment: -0.10 23. Rd1 Rxd1+ 24. Qxd1 Nc6 25. h4 Kc7 26. Bf4 e5 27. >>Be3 Bd6 28. Qg4 Rd8 29. h5 Kb7. >>CM6000 is not able to find the best move in this position, 23. b3 ! >>The reason why it is because it is using static features to evaluate the >>position and it misses the dynamic features in the position. Fritz 5.16 finds >>the best move and give the following assessment: 0.16 23. b3 c3 24. Nxc3 bxc3 >>25. Rxc3+ Kb8 26. Qc2 Bd6 27. b4 Bxb4. > >after 23.Bxa7 Kb7 >Fritz5.16 finds 24.Be3 with the following assessment at depth 14:0.50 24.Be3 Nb3 >25.Nb6 Nxc1 26.Rxc1 c3 27.Qxb5 Rxb5 28.bxc3 b3 > >The assessment is more than 0.16 even at smaller depth (0.25 at depth 9) >Fritz5 is a root proccesor and this is the reason for the change in the >evaluation. > >I think that Fritz5 has an old opinion against some moves like Bxa7 and the >reason for not playing Bxa7 like Junior5 is that it believes that Bxa7 is bad >and it has nothing to do with evaluation of dynamic features in the positon. > >Uri Garry Kasparov in his analyses to the position says that Bxa7 is bad and it gives Black the winning chances. So if you place your faith only in chess engines to do the analysis, then you are really in trouble... Remember that there are tactical positions which chess engines fail to solve.
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.