Author: Uri Blass
Date: 15:22:57 06/08/03
Go up one level in this thread
On June 08, 2003 at 17:16:48, Drexel,Michael wrote:
>On June 08, 2003 at 16:28:14, Uri Blass wrote:
>
>>I want to find the mistakes of movei against
>>Sos in Leo's tournament and I may use them as test positions for future
>>versions.
>>
>>Note that the problem of finding the losing mistake is often hard to find and it
>>is hard to know the exact point that the position became lost instead of
>>inferior.
>>
>>I decided to try to find the point that movei lost material.
>>I have a version of yace only material and I gave it the position
>>after move 13 of white.
>>
>>Yace(only material) use the following values for material:
>>Queen=10.01,Rook=5,Bishop=3.5,Knight=3.4,Pawn=0.8
>>
>>Yace only finished 46 out of 53 moves to analyze and I will not give it to
>>finish iteration 14 but it seems that already at move 13 black is losing
>>material(yace only material cannot finish depth 14 even after more than 5
>>hours).
>>
>>SOS 99.11.03 - Movei 0.08.068
>>[D]2kr3r/ppp1ppbp/2n3p1/7n/1qbPP3/2N1BNP1/PP3PBP/1R1QR1K1 b - - 0 1
>>
>>Analysis by Yace 0.99.56:
>>
>>13...Qxc3 14.bxc3 Bxa2
>> +- (5.81) Depth: 1 00:00:03
>>13...Nxg3 14.fxg3 Nxd4
>> +- (1.80) Depth: 1 00:00:03
>>13...Nxd4 14.Nxd4 Bxd4 15.Bxd4 Rxd4 16.Qxd4 Qxc3
>> ± (0.80) Depth: 1 00:00:03
>>13...Bf8
>> = (0.00) Depth: 1 00:00:04
>>13...Bf8 14.Bg5
>> = (0.00) Depth: 2 00:00:04
>>13...Bf8 14.Bg5 Be6
>> = (0.00) Depth: 3 00:00:04
>>13...Bf8 14.Bg5 Be6 15.a4
>> = (0.00) Depth: 4 00:00:04
>>13...Bf8 14.Bg5 Be6 15.a4 Qd6
>> = (0.00) Depth: 5 00:00:04 17kN
>>13...Bf8 14.Bh3+ Kb8 15.g4 Nf6 16.e5 Nd5
>> = (0.00) Depth: 6 00:00:04 88kN
>>13...Bf8 14.d5 Kb8 15.a3 Qxa3 16.bxa3 Nf6
>> ² (0.40) Depth: 7 00:00:05 391kN
>>13...Bf8 14.d5 a6 15.Ng5 Ne5 16.f4 Nd3 17.Nxf7
>> ² (0.70) Depth: 7 00:00:05 391kN
>>13...Qb6 14.d5 Qa5 15.Bb6 axb6 16.Na4
>> ² (0.69) Depth: 7 00:00:06 724kN
>>13...Qb6 14.d5 Qa6 15.g4 Bxc3 16.bxc3 Nf6 17.Nd2
>> = (0.10) Depth: 7 00:00:07 724kN
>>13...Qa5 14.d5 Bxc3 15.bxc3 Bxa2 16.dxc6 Rxd1 17.Rbxd1
>> = (0.09) Depth: 7 00:00:07 1042kN
>>13...Qa5 14.a3 Nf6 15.Nd2 Bd3 16.Bh3+ Kb8 17.Nb3
>> = (0.00) Depth: 7 00:00:07 1042kN
>>13...Qa5 14.Nd2 Qa6 15.Nxc4 Qxc4 16.Bh3+ Kb8 17.Bf1 Qe6 18.d5
>> = (0.10) Depth: 8 00:00:09 1369kN
>>13...Na5 14.a3 Qd6 15.Nd2 Bd3 16.Qxh5 gxh5 17.Nb5
>> = (0.09) Depth: 8 00:00:20 5166kN
>>13...Na5 14.a3 Qd6 15.Nd2 Bd3 16.Bh3+ Kb8 17.e5 Qa6
>> = (0.00) Depth: 8 00:00:20 5166kN
>>13...Na5 14.Bd2 Nc6 15.g4 Nf6 16.d5 Na5 17.Ne5 Rdf8 18.Nxc4
>> = (0.10) Depth: 9 00:00:34 9474kN
>>13...e6 14.g4 Nf6 15.Nd2 Nxd4 16.Nxc4 Qxc4 17.Qxd4 Rxd4 18.Bxd4
>> = (0.09) Depth: 9 00:01:26 25042kN
>>13...e6 14.g4 Nf6 15.Bh3 Kb8 16.a3 Qa5 17.e5 Ng8
>> = (0.00) Depth: 9 00:01:27 25501kN
>>13...e6 14.g4 Nf6 15.Bh3 Kb8 16.a3 Qa5 17.e5 Ng8 18.Nd2
>> = (0.00) Depth: 10 00:01:36 29516kN
>>13...e6 14.g4 Nf6 15.Bh3 Kb8 16.a3 Qa5 17.e5 Ng8 18.Nd2 Bd3
>> = (0.00) Depth: 11 00:02:38 60258kN
>>13...e6 14.g4 Nf6 15.Bh3 Kb8 16.a3 Qa5 17.e5 Ng8 18.Nd2 Bd3 19.Nb3
>> = (0.00) Depth: 12 00:05:50 153600kN
>>13...e6 14.g4 Nf6 15.Bh3 Kb8 16.a3 Qf8 17.e5 Ng8 18.Qc1 Bd3 19.Ra1 Re8
>> = (0.00) Depth: 13 00:13:26 366031kN
>>13...e6 14.a3 Qf8 15.Qa4 Ba6 16.e5 Bh6 17.b4 Bxe3 18.fxe3 Nf4 19.gxf4 Qxb4
>>20.axb4 Nxb4
>> ² (0.40) Depth: 14 00:28:37 784069kN
>>13...e6 14.a3 Qb3 15.Ng5 Rd7 16.d5 exd5 17.Bh3 Qxd1 18.Bxd7+ Kxd7 19.Nxd1 dxe4
>>20.Nxe4 Nf6 21.Nxf6+ Bxf6
>> ² (0.70) Depth: 14 00:43:24 1065081kN
>>13...Be6 14.d5 Bg4 15.dxc6 Rxd1 16.Nxd1
>> ² (0.69) Depth: 14 03:14:56 109523kN
>>13...Be6 14.d5 Bg4 15.Qc2 Bxf3 16.Bxf3 Ne5 17.Be2 Qa5 18.Nb5
>> ² (0.69) Depth: 14 03:21:57 271323kN
>>
>>(Blass, Tel-Aviv 08.06.2003)
>>
>>In the game movei did not find the best defence
>>and after 13...e6 14.a3 Qf8?(Qb3 was probably better) 15.Qa4 Ba6 16.Red1
>>Ne7?(Nb8 is probably better)
>>17.b4 c6 18.b5 got a clearly lost position.
>
>Maybe Movei saw that he would lose material after 14...Qb3 15.Qxb3 Nd2.
>
>Ne7 was not a mistake, the position is lost anyway.
>
>
>>
>>Here is the full game(the match ended in 2-2)
>>
>>[Event "WBEC4-1st Division"]
>>[Site "ATHLON-MP2200"]
>>[Date "2003.06.07"]
>>[Round "6.1"]
>>[White "SOS 99.11.03"]
>>[Black "Movei 0.08.068"]
>>[Result "1-0"]
>>[PlyCount "67"]
>>[EventDate "2003.??.??"]
>>[TimeControl "40/2400:0"]
>>
>>1. d4 Nf6 2. c4 g6 3. g3 d5 4. cxd5 Qxd5 5. Nf3 Nc6 6. Nc3 Qa5 7. Bg2 Be6 8.
>>O-O O-O-O 9. e4 Bc4 10. Re1 Bg7 11. Bf4 Nh5 12. Be3 Qb4 13. Rb1 e6 14. a3 Qf8
>>15. Qa4 Ba6 16. Red1 Ne7 17. b4 c6 18. b5 cxb5 19. Nxb5 Kb8 20. Nxa7 Rd6 21. e5
>>Nd5 22. exd6 Kxa7 23. Ne5 Qxd6 24. Rxb7+ Kxb7 25. Rb1+ Ka7 26. Bxd5 Rb8 27.
>>Rxb8 exd5 28. Rb1 Nxg3 29. Nc6+ Qxc6 30. Qxc6 Ne2+ 31. Kh1 Bb5 32. Qxb5 Ng3+
>>33. hxg3 h6 34. Qa4# {Movei->White mates} 1-0
>>
>>
>>
>>Uri
>
>4...Qxd5 is probably an inferior line.
>5...Nc6 is a mistake, black should play Bg7 to keep all options.
>7...Be6 is probably already the losing mistake.
>
>IMO you can not expect Movei to play the opening well. Even a commercial program
>like Chessmaster plays often very bad moves in the opening when out of book.
>Pawn sacrifices which are completely unsound for example.
>
>Michael
I think that it is possible to learn from mistakes of programs in the opening.
The fact that commercial programs do mistakes does not mean that it is
impossible to do better.
I would like to have a list of positions when movei fails based on popularity
because maybe I can find patterns that can help me to avoid big majority of the
mistakes.
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.