Author: Slater Wold
Date: 23:24:04 09/01/01
Go up one level in this thread
On September 02, 2001 at 02:11:03, F. Jermann wrote: >On September 01, 2001 at 19:46:27, Slater Wold wrote: > >>[D]3r3k/2qNb1r1/1p2pn2/pBpbNp2/Q2P1Pp1/P3P3/1PR3P1/2R2K2 b - - >> >>What is the correct move here? >> >>Had you asked Deep Thought back in 1989, it would say 33...Be4. >> >>However I could not get Deep Shredder to see this move in any reasonable amount >>of time. (10+ minutes.) > >Gambit Tiger 2.0 aggr. thinks Le4 is good (about -1.17 for black), but Sxd7 Sxd7 >and then Le4 is even better (about -1.7, score at depth 13) from the very >beginning of analysis. Only the first several iterations, Le4 seems to be the >best. I think GT analysis may be correct here! > >> >>But is it correct? Well, considering 33...Be4 leads to 34. Rd2 Nd5 35. Re2, >>then YES, it would be correct. >> >>However, are there other moves to stop this? Well at depth 12, and 5+ minutes >>into evaluation, Deep Shredder would play 34. Rd2 after 33...Be4. That leaves >>us with: >> >>[D]3r3k/2qNb1r1/1p2pn2/pBp1Np2/Q2PbPp1/P3P3/1P1R2P1/2R2K2 b - - >> >>This move seems pretty difficult to get also. Deep Shredder looked at Bd5 up to >>ply 11, and then changed to Rh7 after 2:44 minutes, scoring itself up +0.25. > >GT 2.0: Rh7, -1.33 for black > >>After 4:36 minutes, it scores Rh7 as +0.47. Nd5 secures a little more position, >>and leads to: >> >>[D]3r3k/2qNb1r1/1p2p3/pBpnNp2/Q2PbPp1/P3P3/1P1R2P1/2R2K2 w - - >> >>There are several moves that are possible here. From Qb3, Kg1, Re2, Re1, >>however all are losing. After 3+ minutes, Deep Shredder plays Kg1, scoring >>itself down -0.58. > >GT 2.0: Kg1, eval -0.56. So from GT view Rh7 is much stronger! > >Re2 seems to be worse for white: > > >3r3k/2qNb1r1/1p2p3/pBpnNp2/Q2PbPp1/P3P3/1P2R1P1/2R2K2 b - - 0 1 > >Analysis by Gambit Tiger 2.0: > >35...Th7 36.Ke1 Th1+ 37.Kd2 Txc1 38.Kxc1 cxd4+ 39.Lc4 d3 > µ (-0.80) Tiefe: 7 00:00:00 35kN >35...Th7 36.Ke1 Th1+ 37.Kd2 Txc1 38.Kxc1 cxd4+ 39.Kd1 d3 > µ (-0.76) Tiefe: 7 00:00:00 55kN >35...Th7 36.Ke1 Th1+ 37.Kd2 Txc1 38.Kxc1 cxd4+ 39.Lc4 Kg7 40.exd4 Sxf4 > µ (-0.98) Tiefe: 8 00:00:02 200kN >35...Th7 36.Ke1 Th1+ 37.Kd2 Txc1 38.Kxc1 > µ (-0.98) Tiefe: 9 00:00:05 399kN >35...Th7 36.Tee1 Kg7 37.Ke2 Lxg2 38.Kd2 Sf6 39.Te2 Sxd7 40.Txg2 Sxe5 41.fxe5 > -+ (-1.42) Tiefe: 10 00:00:41 3612kN >35...Th7 36.Tee1 Kg7 37.Ke2 Th2 38.Kd2 Txg2+ 39.Te2 Sxe3 40.dxc5 Lxc5 41.Txg2 >Lxg2 > µ (-1.34) Tiefe: 11 00:02:02 10287kN > > > >>However, the move that was played, was Re2. I will follow >>through with the move that was played originally: >> >>[D]3r3k/2qNb1r1/1p2p3/pBpnNp2/Q2PbPp1/P3P3/1P2R1P1/2R2K2 b - - >> >>Now it becomes clear what moves are best. Deep Shredder scored this +1.41 for >>itself in 0:00 seconds. After 31 seconds, it drops to +1.16. What about what >>Shredder wanted earlier, Kg1, would that have saved it? No, it still plays Rh7, >>and the score is even worse. (Which makes me wonder, how did it think Kg1 was >>better!) >> >>After this, it is a clear win for black. But how did this happen? At move 33 >>Deep Shredder says that Nxd7 is best, and thinks it's winning by a little over >>0.30. Playing a better 33...Be4! and 2 moves later it's +2.00?! With Be4, and >>a forced Rd2, we know that there is no good response to Nd5 or Rh7, and it's a >>clear win for black!. This is only 2 moves away! This seems baffling to me, >>and I hope someone can help explain. 33..Be4 only has one response. And we >>know after 34. Rd2 it's a simple win for black. So why can't any programs find >>it? It seems simple. >> >>Nxd7 is not the correct move, as it just exchanges pieces, and gives white a >>better chance to draw. Be4 is winning, clearly. >> >>QUESTION: >> >>In this postiton, who can find 33...Be4! And find it in a tournament time? >>(3-5 minutes) >> >>[D]3r3k/2qNb1r1/1p2pn2/pBpbNp2/Q2P1Pp1/P3P3/1PR3P1/2R2K2 b - - >> >>Here is Deep Shredder eval: >> >>position: >> 10.01 0:01 +0.38 1...Nxd7 2.Bxd7 Qxd7 3.Qxd7 Rxd7 4.Nxd7 cxd4 (474.972) >>467.4 >> 11.01 0:02 +0.38 1...Nxd7 2.Bxd7 Qxd7 3.Qxd7 Bxg2+ 4.Rxg2 Rxd7 5.Nxd7 e5 >>(1.204.841) 478.8 >> 12.01 0:06 +0.38 1...Nxd7 2.Bxd7 Qxd7 3.Nxd7 Rxd7 (3.187.564) 477.7 >> 13.01 0:30 +0.38 1...Nxd7 2.Bxd7 Qxd7 3.Qxd7 Rxd7 4.Nxd7 Bh4 5.dxc5 Bf3 >>6.gxf3 (13.921.049) 460.6 >>best move: Nf6xd7 time: 5:12.688 min n/s: 445.376 CPU 199.3% nodes: >>139.263.904 >> >> >>Thanks. And good luck! >> >> >> >>Slate > >Gambit Tiger continuously thinks it wins with black! I do not agree that Sxd7 is >really bad for black in the position at move 33. GT likes it and all variations >seem to show that is a really interesting attacking (!) move! See: > >Neue Partie - Gambit Tiger 2.0 >3r3k/2qnb1r1/1p2p3/pBpbNp2/Q2P1Pp1/P3P3/1PR3P1/2R2K2 w - - 0 1 > >Analysis by Gambit Tiger 2.0: > >34.Lxd7 Lf6 35.Lc6 Lxe5 36.Lxd5 Lxd4 37.exd4 Txd5 38.dxc5 bxc5 > ³ (-0.58) Tiefe: 6 00:00:00 36kN >34.Lxd7 Th7 35.Le8 Ld6 36.Sg6+ Kg7 37.dxc5 bxc5 > ³ (-0.42) Tiefe: 6 00:00:00 40kN >34.Lxd7 Th7 35.Le8 Ld6 36.Sg6+ Kg7 37.dxc5 bxc5 38.Db5 > ³ (-0.42) Tiefe: 7 00:00:01 74kN >34.Lxd7 > µ (-1.32) Tiefe: 8 00:00:02 173kN >34.Lxd7 Lf6 35.Lb5 Lxe5 36.fxe5 Db7 37.Ke2 Lxg2 38.Kd2 De4 > -+ (-1.60) Tiefe: 9 00:00:03 273kN >34.Sxd7 Le4 35.dxc5 Lxc2 36.Txc2 bxc5 37.b4 axb4 38.axb4 g3 39.bxc5 > -+ (-1.56) Tiefe: 9 00:00:05 430kN >34.Sxd7 Le4 > -+ (-1.56) Tiefe: 10 00:00:05 457kN >34.Sxd7 Le4 35.Se5 Lf6 36.Db3 Lxc2 37.Txc2 Lxe5 38.fxe5 De7 39.Lc4 cxd4 40.Lxe6 > -+ (-1.62) Tiefe: 11 00:00:20 1604kN >34.Sxd7 Le4 35.Se5 Lf6 36.Dc4 Lxc2 37.Txc2 Td5 38.Ke2 Lxe5 39.fxe5 Kg8 40.Kd2 > -+ (-1.60) Tiefe: 12 00:00:48 3681kN >34.Sxd7 Le4 35.Se5 Lf6 36.Dc4 Lxc2 37.Txc2 Td5 38.Db3 Lxe5 39.fxe5 Kg8 40.Ke2 > -+ (-1.76) Tiefe: 13 00:02:14 11316kN > >Frank Gambit Tiger is usually an exception to almost every position. I think a good explanation of what Gambit Tiger likes here, is the fact that it has pieces still pretty local to the king. It's likes to get it's queen on the 7th and keep it there. (I've seen GT 2.0 with a queen on the 7th, and have a positive score in a completly lost position.) I am pretty sure that the exchanges still lead more to a draw, than a win for black though. However, it is really hard to tell. A -1.76 score with Gambit Tiger is basically a draw anyway. :) I would venture to say, that move will fail. And it will move on to another move. This is only a 2 minute evaluation. Give it 20 minutes, and see what it says. Slate
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.