Author: Vasik Rajlich
Date: 03:39:35 10/24/04
Go up one level in this thread
On October 24, 2004 at 03:32:51, Uri Blass wrote: >On October 24, 2004 at 03:18:02, Uri Blass wrote: > >>On October 24, 2004 at 03:06:30, Uri Blass wrote: >> >>>On October 23, 2004 at 16:14:38, Vasik Rajlich wrote: >>> >>>>On October 23, 2004 at 13:40:00, Uri Blass wrote: >>>> >>>>>On October 23, 2004 at 12:33:07, Vasik Rajlich wrote: >>>>> >>>>>>On October 23, 2004 at 12:12:25, Dan Andersson wrote: >>>>>> >>>>>>> According to Ed it's not a Nommen book. I have fond memories of meeting 9. ... >>>>>>>Bd7 Especially when I was up against a wunderkind. He was under the impression >>>>>>>that he was good in a tactical skirmish and that I was a positional player :) >>>>>>> >>>>>>>MvH Dan Andersson >>>>>> >>>>>>I play this move 9. .. Bd7 myself - it seems ok. Sjeng played it horribly though >>>>>>- 12. .. Qa5 is suspicious and then 15. .. Nc4 is just a blunder no human would >>>>>>make - not even blindfold while casually browsing CCC :) ... >>>>>> >>>>>>Vas >>>>> >>>>>I do not understand why do you think that no human will play 15...Nc4 >>>>> >>>>>One of the things that players learn is that pair of bishops is good and if they >>>>>can see that they can trade one of themby mean of Nc4 they may be happy to do >>>>>it. >>>>> >>>>>The only reason that I find not to play this move is some fear from attack >>>>>against the king but I do not know if black has better alternative and >>>>>unexperienced humans may not know about that danger(you can find a lot of humans >>>>>with no exprience or little experience in chess). >>>>> >>>>>I believe that there is no problem to find humans that will play Nc4 if you ask >>>>>them what is the best move. >>>>> >>>>>Uri >>>> >>>>Well - no Dragon-playing human will play 15. .. Nc4. >>>> >>>>In this variation with 9. 0-0-0 Bd7, this knight always stays on e5. If black >>>>ever plays .. Nc4, he'll be either one or two tempi down on the main lines with >>>>9. Bc4 Rc8. >>>> >>>>Vas >>> >>>I do not know the dragon but I checked with yace and the blunder of black was >>>later in the game(Bh5) >>> >>>After Bh5 Yace has no problems to find the right move is tiger's Nf4 but Sjeng >>>even did not expect it. >>> >>>Finding Nf4 is a nice test position and I am going also to test movei in it. >>> >>>Chess Tiger - Deep Sjeng, Dutch Open 2004 2004 >>>[D]2r1r1k1/pp2pp1p/q2p2pQ/3Nb2b/3B3P/1N6/PPP5/1K3R1R w - - 0 1 >>> >>>Analysis by Yace 0.99.87: >>> >>>24.Qxh5 gxh5 >>> -+ (-5.18) Depth: 1 00:00:00 >>>24.Bxe5 dxe5 >>> ± (1.23) Depth: 1 00:00:00 >>>24.Bxe5 dxe5 >>> ± (1.23) Depth: 1 00:00:00 >>>24.Bxe5 dxe5 >>> ± (1.23) Depth: 2 00:00:00 >>>24.Bxe5 dxe5 >>> ± (1.23) Depth: 2 00:00:00 >>>24.Bxe5 dxe5 25.Rhg1 Qc4 >>> ± (1.14) Depth: 3 00:00:00 >>>24.Bxe5 dxe5 25.Rhg1 Qc4 >>> ± (1.14) Depth: 3 00:00:00 >>>24.Bxe5 dxe5 25.Ne3 Be2 26.Rf2 >>> ± (1.07) Depth: 4 00:00:00 >>>24.Bxe5 dxe5 25.Ne3 Be2 26.Rf2 >>> ± (1.07) Depth: 4 00:00:00 >>>24.Bxe5 dxe5 25.Rf2 Qc6 26.Rg1 Qxd5 27.Qxh5 >>> ± (0.92) Depth: 5 00:00:00 >>>24.Rhg1 Qxf1+ 25.Rxf1 Bg7 26.Bxg7 Rxc2 >>> ± (0.93) Depth: 5 00:00:00 >>>24.Rhg1 Bxd4 25.Nxd4 Qc4 26.Qg5 Qxd4 >>> ± (0.95) Depth: 5 00:00:00 >>>24.Rhg1 Bxd4 25.Nxd4 Qc4 26.Qg5 Qxd4 >>> ± (0.95) Depth: 5 00:00:00 >>>24.Rhg1 Bxd4 25.Nxd4 Qc4 26.Qg5 Qxd4 27.Qxh5 >>> ± (0.95) Depth: 6/23 00:00:00 618kN >>>24.Rhg1 Bxd4 25.Nxd4 Qc4 26.Qg5 Qxd4 27.Qxh5 >>> ± (0.95) Depth: 6/25 00:00:01 1437kN >>>24.Rhg1 Qc6 25.Bxe5 dxe5 26.Nc3 Bf3 27.h5 Red8 >>> ± (1.08) Depth: 7/25 00:00:02 3149kN >>>24.Rhg1 Qc6 25.Bxe5 dxe5 26.Nc3 Bf3 27.h5 Red8 >>> ± (1.08) Depth: 7/29 00:00:04 5020kN >>>24.Rhg1 Be2 25.Bxe5 dxe5 26.Re1 f6 27.Nc3 Bc4 28.Nd2 >>> ± (1.01) Depth: 8/29 00:00:06 8253kN >>>24.Ne3 Bxd4 25.Nxd4 Qxf1+ 26.Nxf1 f6 >>> ± (1.02) Depth: 8/29 00:00:09 12841kN >>>24.Ne3 Qb5 25.Bxe5 Qxe5 26.Rf2 f6 27.Rg1 Kh8 28.Nf5 >>> ± (1.02) Depth: 8/29 00:00:10 14289kN >>>24.Ne3 Qb5 25.Bxe5 Qxe5 26.Rf2 f6 27.Rg1 Kh8 28.Nf5 >>> ± (1.02) Depth: 8/29 00:00:12 16720kN >>>24.Ne3 Qb5 25.Rf2 Be2 26.Bxe5 dxe5 27.Re1 Bc4 28.Nd2 Bd5 >>> ± (1.02) Depth: 9/29 00:00:14 19333kN >>>24.Rhg1 Be2 25.Bxe5 dxe5 26.Rxf7 Kxf7 27.Qxh7+ Ke6 28.Nc5+ Rxc5 29.Rxg6+ Kd7 >>>30.Rxa6 Rxc2 >>> ± (1.03) Depth: 9/29 00:00:16 21028kN >>>24.Rhg1 Qc6 25.Nc3 Qc4 26.Bxe5 dxe5 27.Rf5 Kh8 28.Rxe5 Qxh4 >>> ± (1.27) Depth: 9/30 00:00:22 28988kN >>>24.Rhg1 Qc6 25.Nc3 Qc4 26.Bxe5 dxe5 27.Rf5 Kh8 28.Rxe5 Qxh4 >>> ± (1.27) Depth: 9/32 00:00:34 44491kN >>>24.Rhg1 Be2 25.Rf6 Qb5 26.Rxd6 Bh8 27.Bxh8 Kxh8 28.Nd4 Qa4 29.Qg7+ Kxg7 30.Nf5+ >>>Kg8 31.b3 >>> ± (0.88) Depth: 10/33 00:00:54 72094kN >>>24.Ne3 Qb5 >>> ± (0.89) Depth: 10/33 00:00:56 74082kN >>>24.Ne3 Qe2 25.Bxe5 dxe5 26.Rhg1 Bf3 27.Nc1 Qh2 28.Nd3 Be4 29.Nb4 >>> ± (1.02) Depth: 10/33 00:01:08 89137kN >>>24.Nf4 Bg4 25.h5 Bf5 26.hxg6 Bxc2+ 27.Ka1 Bxb3 28.gxf7+ Kxf7 29.Qxh7+ Bg7 >>>30.Qg6+ Kf8 31.Bxg7+ Kg8 >>> ± (1.03) Depth: 10/39 00:01:39 130203kN >>>24.Nf4 Bg4 25.h5 Bf5 26.hxg6 Bxc2+ 27.Ka1 Bxb3 28.gxf7+ Kxf7 29.Qxh7+ Bg7 >>>30.Qg6+ Kf8 31.Bxg7+ Kg8 >>> +- (1.67) Depth: 10/39 00:01:45 135873kN >>>24.Nf4 Bxd4 25.Nxd4 Qa5 26.Nxh5 Qxh5 27.Qxh5 gxh5 28.Rhg1+ Kh8 29.Rxf7 Rc4 >>>30.Nf5 Rg4 >>> +- (2.49) Depth: 10/39 00:01:53 144676kN >>>24.Nf4 Bxd4 25.Nxd4 Qa5 26.Nxh5 Qxh5 27.Qxh5 gxh5 28.Rhg1+ Kh8 29.Rxf7 Rc4 >>>30.Nf5 Rg4 >>> +- (2.49) Depth: 10/39 00:01:57 150016kN >>>24.Nf4 Bxd4 25.Nxd4 Qa5 26.Nh3 Qe5 27.Rxf7 Kxf7 28.Qxh7+ Kf6 29.Rf1+ Bf3 >>>30.Rxf3+ Qf4 31.Nxf4 Ke5 >>> +- (2.89) Depth: 11/39 00:02:28 186993kN >>>24.Nf4 Bxd4 25.Nxd4 Qa5 26.Nh3 Rc5 27.Nb3 Qb6 28.Nxc5 Qxc5 29.Ng5 f6 30.Qxh7+ >>>Kf8 >>> +- (3.45) Depth: 11/42 00:02:50 210979kN >>>24.Nf4 Bxd4 25.Nxd4 Qa5 26.Nh3 Rc5 27.Nb3 Qb6 28.Nxc5 Qxc5 29.Ng5 f6 30.Qxh7+ >>>Kf8 >>> +- (3.45) Depth: 11/42 00:03:30 268205kN >>>24.Nf4 Bxd4 25.Nxd4 Qa5 26.Nh3 Rc5 27.Nb3 Qb6 28.Nxc5 Qxc5 29.Ng5 a6 30.Nxf7 Qc3 >>> +- (3.85) Depth: 12/42 00:04:20 323953kN >>>24.Nf4 Bxd4 25.Nxd4 Qa5 26.Nh3 Qe5 27.Rxf7 Qh8 28.Ng5 e5 29.Nb5 Rf8 30.Rxh7 Qf6 >>>31.Nxa7 >>> +- (4.26) Depth: 12/42 00:05:11 377062kN >>> >>>(, MyTown 24.10.2004) >>> >>>Uri >> >>Here are results of movei(not the first time that it surprise me by being faster >>than Yace inspite of lack of knowledge that yace probably has(movei still does >>not use internal iterative deepening and does not use hash tables for pruning). >> >> >>10 114 2181 13252695 d4e5 d6e5 h1g1 h5e2 f1f7 g8f7 h6h7 f7e6 b3c5 c8c5 g1g6 e6d5 >>g6a6 e2a6 h7f7 e7e6 f7e8 >>10 115 3632 22190503 d5f4 >>10 144 3900 23649160 d5f4 >>10 140 4212 25365120 d5f4 e5d4 b3d4 a6c4 d4e6 f7e6 f4h5 c4d4 c2c3 d4e5 h1g1 e8f8 >>10 140 4257 25652711 d5f4 e5d4 b3d4 a6c4 d4e6 f7e6 f4h5 c4d4 c2c3 d4e5 h1g1 e8f8 >>11 170 5331 31536647 d5f4 >>11 240 7087 41044606 d5f4 >>11 274 9250 52774805 d5f4 e5d4 b3d4 a6a5 f4h3 a5e5 f1f7 e5h8 h1f1 e7e5 d4b5 >>11 274 9911 57082921 d5f4 e5d4 b3d4 a6a5 f4h3 a5e5 f1f7 e5h8 h1f1 e7e5 d4b5 >> >>Uri > >I forgot more disadvantages of movei > >Movei does not have qsearch and it generates all moves only to search the >captures(No miracle that most programs are faster than movei in nodes per >second). > >Movei has a primitive mobility evaluation that is based only on number of legal >move and stage of the game. > >Movei's king safety is also not something special and it only evaluate squares >after the king so if the king is at c3 it does not evaluate attacks against b2 >c2 d2 b3 d3 that can be clearly relevant for king safety. > >Movei does not evaluate backward pawns because my tests did not suggest that it >is productive. >It also does not evaluate knightoutposts for the same reason. > >There are more points that this post is too short to mention. > >If I only fix the bugs and implement the basic knowledge that movei is missing >in a productive way then I guess that movei may get easily above the level of >all the free engines of today. > >Uri Uri, the reason Movei "gets" this problem so much faster than Yace is that it prefers it even with just +1.00 scores, whereas Yace will prefer other moves until the big fail high. So - your evaluation is slightly better for this position. There are many such positions, for example BS2830-1. Many engines get it quickly via eval, while others (like Rybka) which have high weights for the queen have to wait for the big tactical fail high which comes later. Vas
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.