Author: enrico carrisco
Date: 15:15:26 02/05/06
Go up one level in this thread
I guess when something is the NEW greatest thing, it is easy to exaggerate (i.e. "I caught a fish that was 20ft long" -- in all the excitement of a 4ft catch...) I think the "SUPER" evaluation strength is imagined because the tactics are stronger than anything else. Somehow, in the midst of all the madness, the eval gets the credit. Let’s take a look… For example: [POSITION 1]: [D]r4rk1/2q2ppp/N4n2/4nb2/4p3/P1P1P3/1B2BPPP/R2Q1RK1 b - - 0 18 HIARCS 10 finds Rxa6 instantly and sticks to it, eval (-0.81) just like Rybka. Analysis by Hiarcs 10: 1...Rxa6 ² (0.38) Depth: 9/25 00:00:00 1...Rxa6 2.Bxa6 Nfg4 3.g3 Qb6 4.Qe2 Nxh2 5.Rfd1 Nef3+ ² (0.34) Depth: 9/25 00:00:00 160kN 1...Rxa6 2.Bxa6 Nfg4 3.g3 Qb6 4.Qe2 Nxh2 5.Rfb1 Nef3+ 6.Kh1 Qh6 = (0.22) Depth: 10/25 00:00:00 309kN 1...Rxa6 = (-0.03) Depth: 11/25 00:00:01 517kN 1...Rxa6 2.Bxa6 Nfg4 3.g3 Qb6 4.Be2 Qxb2 5.c4 Qc3 6.Bxg4 ³ (-0.48) Depth: 11/25 00:00:02 732kN 1...Rxa6 2.Bxa6 Nfg4 3.g3 Qb6 4.Be2 Qxb2 5.c4 Nf6 6.c5 Bh3 ³ (-0.48) Depth: 12/31 00:00:03 1336kN 1...Rxa6 2.Bxa6 Nfg4 3.g3 Qb6 4.Be2 Qxb2 5.h3 Nf6 6.g4 Be6 ³ (-0.44) Depth: 13/37 00:00:07 2773kN 1...Rxa6 ³ (-0.69) Depth: 14/37 00:00:16 5941kN 1...Rxa6 2.Bxa6 Nfg4 3.g3 Nf3+ 4.Kg2 Qb6 5.h3 Nxe3+ 6.fxe3 Bxh3+ 7.Kf2 Bxf1 8.Bxf1 Qxb2+ 9.Be2 Qxc3 10.Bxf3 exf3 11.Kxf3 µ (-0.81) Depth: 14/37 00:00:20 7321kN [POSITION 2]: [D]5rk1/2q2ppp/B4n2/4nb2/4p3/P1P1P3/1B3PPP/R2Q1RK1 b - - 0 19 HIARCS 10 finds Nfg4 at 1 ply and sticks to it, again with good eval for black (-1.42) and HIARCS 10 sees the whole line of play (even 22...Rd8) This is all tactical king attacks based on search... Analysis by Hiarcs 10: 1...Nfg4 2.g3 Qb6 3.Qe2 Nf3+ 4.Kh1 Ngxh2 5.Rfd1 Qh6 ² (0.30) Depth: 9/22 00:00:00 1...Nfg4 2.g3 Qb6 3.Qe2 Nxh2 4.Rfd1 Rb8 = (0.17) Depth: 9/22 00:00:00 157kN 1...Nfg4 = (-0.08) Depth: 10/24 00:00:00 281kN 1...Nfg4 2.g3 Qb6 3.Be2 Qxb2 4.c4 Rc8 ³ (-0.57) Depth: 10/24 00:00:01 400kN 1...Nfg4 2.g3 Qb6 3.Be2 Qxb2 4.h3 Nf6 5.g4 ³ (-0.44) Depth: 11/27 00:00:01 690kN 1...Nfg4 2.g3 Qb6 3.Be2 Qxb2 4.c4 Rc8 5.a4 Nf6 6.Rb1 ³ (-0.39) Depth: 12/31 00:00:03 1354kN 1...Nfg4 ³ (-0.64) Depth: 13/32 00:00:08 3114kN 1...Nfg4 2.g3 Nf3+ 3.Kg2 Qb6 4.Be2 Qxb2 5.h3 Nge5 6.g4 Be6 7.Bxf3 exf3+ 8.Kg3 Qxc3 µ (-1.09) Depth: 13/32 00:00:11 4534kN 1...Nfg4 2.g3 Nf3+ 3.Kg2 Qb6 4.Be2 Qxb2 5.h3 Nge5 6.g4 Bd7 7.Bxf3 exf3+ 8.Kg3 Qxc3 µ (-1.01) Depth: 14/36 00:00:19 7571kN 1...Nfg4 µ (-1.26) Depth: 15/36 00:00:32 12894kN 1...Nfg4 2.g3 Nf3+ 3.Kg2 Qb6 4.Be2 Rd8 5.Qa4 Qxb2 6.Qa6 Qxc3 7.Rad1 Rb8 8.Qd6 Qb2 -+ (-1.48) Depth: 15/44 00:00:49 19722kN [POSITION 3]: [D]2rq1rk1/1p1nbp1p/6pB/p4b2/N1Pp4/PP6/4BPPP/2QR1RK1 b - - 0 21 Again HIARCS 10 finds b5 instantly and sticks with it. You say you’re almost positive no other engine would play these moves... Well, at least one other engine does! ;) I am sure Fritz 9 would play them also (though I have not personally checked it.) [Another Position]: [D]1r1q1r1k/7p/2npb3/p2Nppb1/2B5/Q1P1N3/RP3PPP/3R2K1 b - - 0 23 Again HIARCS 10 finds 23...f4 instantly and sticks to it in fact scoring at least .5 pawn (-0.56) more in blacks favor - does this show more positional knowledge than Rybka? ;) Following on in this game HIARCS 10 finds 26...Bxc4 instantly and sticks with it scoring (-2.80) at depth 13 after 16 sec. Rybka thought it was even (-0.01) after the same amount of time. Analysis by Hiarcs 10: 4...Bxc4 5.Nxc4 d5 6.Nfe3 d4 7.Qc5 µ (-0.76) Depth: 9/21 00:00:00 155kN 4...Bxc4 5.Nxc4 d5 6.Nfe3 d4 7.Qc5 Bxe3 8.Nxe3 Rbc8 9.cxd4 exd4 10.Qh5 µ (-0.85) Depth: 10/31 00:00:00 238kN 4...Bxc4 µ (-1.10) Depth: 11/31 00:00:01 461kN 4...Bxc4 5.Nxc4 d5 6.Qc5 Ne7 7.Nce3 µ (-1.11) Depth: 11/31 00:00:02 950kN 4...Bxc4 µ (-1.36) Depth: 12/35 00:00:06 2154kN 4...Bxc4 5.Nxc4 Qh3 6.Nce3 Nb4 7.Qxa5 Nxa2 8.Rxd6 Rf5 9.g4 Rg8 10.Qxa2 Bxe3 11.Nxe3 Rxg4+ 12.Kh1 Rfg5 -+ (-2.16) Depth: 12/35 00:00:07 2738kN 4...Bxc4 -+ (-2.41) Depth: 13/35 00:00:13 4475kN 4...Bxc4 5.Nxc4 Qh3 6.Nce3 Nb4 7.Qxa5 Nxa2 8.Rxd6 Rf5 9.g4 Rg8 -+ (-2.80) Depth: 13/35 00:00:16 5567kN Analysis by Rybka 1.01 Beta 13b 32-bit: 4...Rfd8 ³ (-0.29) Depth: 3 00:00:00 4...Rfd8 = (-0.06) Depth: 4 00:00:00 4...Rbd8 = (-0.17) Depth: 4 00:00:00 4...Rbd8 5.Bxe6 = (0.11) Depth: 5 00:00:00 4...Rfd8 5.Bxe6 = (-0.05) Depth: 5 00:00:00 4...Rfd8 5.Bxe6 Qxe6 6.Nd5 = (-0.05) Depth: 6 00:00:00 10kN 4...Rfd8 5.Bxe6 Qxe6 6.Nd5 Kg7 = (-0.07) Depth: 7 00:00:00 22kN 4...Rfd8 5.Qa4 Kg7 6.Bxe6 Qxe6 7.h4 = (0.07) Depth: 8 00:00:01 88kN 4...Rfd8 5.Qa4 Qc8 6.Bxe6 Qxe6 7.Qc4 Qxc4 = (0.25) Depth: 9 00:00:01 168kN 4...Rbd8 5.Bxe6 Qxe6 6.Qa4 Kg7 7.Qc4 Qd7 = (0.17) Depth: 9 00:00:02 213kN 4...Rbd8 5.Bxe6 Qxe6 6.Qa4 Kg7 7.Qc4 Qd7 8.h4 Bh6 = (0.22) Depth: 10 00:00:03 333kN 4...Bxc4 5.Nxc4 Qh3 6.Nce3 Rbc8 7.b3 Be7 = (0.03) Depth: 10 00:00:04 409kN 4...Bxc4 5.Nxc4 Qh3 6.Nce3 Rbc8 7.b3 Be7 8.Qa4 = (0.09) Depth: 11 00:00:06 579kN 4...Bxc4 5.Nxc4 Qh3 6.Nce3 Rbc8 7.b3 Rfe8 8.Rxd6 Be7 = (0.09) Depth: 12 00:00:10 992kN 4...Bxc4 5.Nxc4 Qh3 6.Nce3 Rbe8 7.Qa4 Re6 8.b4 Bxe3 9.fxe3 Ne7 = (-0.01) Depth: 13 00:00:17 1708kN Again this is not positional, it’s tactical and discovered by search. Sometimes people read what they want into the motives for why moves were selected and "Rybka worship" requires that people think it evaluates these things when it actually searches them. Regards, -elc. (All analysis with an A64 2.85GHz and 512MB hash.)
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.