Author: Dann Corbit
Date: 12:43:44 10/16/01
Go up one level in this thread
On October 16, 2001 at 12:53:30, Albert Silver wrote:
>I played a game yesterday in which the ending illustrated quite well IMO some of
>the differences between the human and computer approaches to certain positions.
>The game was played in 2 minutes with a 12 second increment.
>
>[D]8/3p1pkp/3Pp1p1/1bR1P3/p7/1r5P/2N2PPK/8 b - - 0 35
>
>I was clearly winning but had to still convert the point and my only worry was
>that he might find a way to get at the d7 pawn. My opponent had just played
>35.Nc2 and was threatening 36.Nd4. This looked tough, and I admit I only saw the
>problem after he played it. Chess blindness had me prepare 35...Bc6?? in my
>mind. Luckily for me I found an interesting resource which I calculated to the
>key 40th move. The funny thing is that no engine will play the 40th move and
>they go on to their usual long sequences, which win also it's true. For this
>reason, they also take quite some time giving 35...Rb2 its due credit. Once you
>see the 40th move though, there is nothing to calculate. It's over. My opponent
>must have missed it too or he might have tried something different.
>
>35...Rb2 {88} 36.Nd4 {5} a3 {3} 37.Nxb5 {23} a2 {1} 38.Rc1 {49}
>Rb1 {2} 39.Nc3 {4} Rxc1 {2} 40.Nxa2 {2}
>
>[D]8/3p1pkp/3Pp1p1/4P3/8/7P/N4PPK/2r5 b - -
>
>40...Rc4 {2} The knight is boxed in and the rook is in a dominating position. I
>left the rook there and my opponent resigned a few moves later.
Talk about hitting the nail on the head! Computer programs seem to like Rc2.
Now, Yace briefly considers Rc4 but abandons it after 3 seconds. I think (at
least against a human) Rc4 is by far more devastating. I hate to resign and
almost always fight on to the bitter end, hoping for a blunder. In this
position, I would resign after Rc4 but would fight on after Rc2 (despite the
obvious immediate loss of a pawn). I even think that in the long term Rc4
really is the better move. There is simply no question that it wins.
Here is the Yace analysis:
Crafty is the magic word, Scid wants to see
Yace (Yet another chess engine) version 0.99.57
Type help for a (very unhelpful) list of commands
Make sure, you have a file yace.ini in the current directory, with a line:
hash 30M
or similar, otherwise Yace will only use a very small hash of 1M.
feature done=0
Could not open "dbbook.bin" in mode rb: Winboard does not like my error message
entries 87381 size 1048572 size_wanted 1048576 tts3 29126
pos_learn = 1
Stored 0 learned positions into hash table
book_learn = 4
Could not open "dblearn.bin" in mode rb: Winboard does not like my error message
hash size wanted 65000000
Stored 0 learned positions into hash table
entries 5416665 size 64999980 size_wanted 65000000 tts3 1805554
5-men endgame table bases found
white ( 1): hash 256M
hash size wanted 256000000
Stored 0 learned positions into hash table
entries 21333333 size 255999996 size_wanted 256000000 tts3 7111110
white ( 1): st 999
white ( 1): test n.ci
title "Albert Silver, CCC post 193195"
Stored 0 learned positions into hash table
solution Rc4
usetime = 998.00, mintime = 998.00 maxtime = 998.00 tl 998.00 ml 0
2 0.006 -3.61 1t 1...h5 2.Nxc1 {-340}
26 0.007 -3.29 1t 1...f6 2.exf6+ Kxf6 3.Nxc1 {-340}
27 0.007 1.79 1t 1...Rc2 {160}
45 0.007 1.79 1. 1...Rc2 {160}
69 0.008 2.55 2++ 1...Rc2 2.Nb4 Rxf2 {240}
82 0.008 2.55 2t 1...Rc2 2.Nb4 {160}
193 0.008 2.55 2. 1...Rc2 2.Nb4 {160}
257 0.008 2.55 3t 1...Rc2 2.Nb4 Rxf2 {240}
440 0.009 2.55 3. 1...Rc2 2.Nb4 Rxf2 {240}
702 0.010 2.46 4t 1...Rc2 2.Nb4 Rxf2 3.Kg3 {240}
1575 0.011 2.46 4. 1...Rc2 2.Nb4 Rxf2 3.Kg3 {240}
2058 0.012 2.58 5t 1...Rc2 2.Nb4 Rxf2 3.Kg3 Rb2 {240}
4027 0.014 2.58 5. 1...Rc2 2.Nb4 Rxf2 3.Kg3 Rb2 {240}
5344 0.030 2.52 6t 1...Rc2 2.Nb4 Rxf2 3.Kg3 Rb2 4.Nd3 {240}
13275 0.042 2.52 6. 1...Rc2 2.Nb4 Rxf2 3.Kg3 Rb2 4.Nd3 {240}
17393 0.049 2.63 7t 1...Rc2 2.Nb4 Rxf2 3.Na6 Re2 4.Nb8 Rxe5 {320}
33216 0.075 2.63 7. 1...Rc2 2.Nb4 Rxf2 3.Na6 Re2 4.Nb8 Rxe5 {320}
42150 0.089 2.63 8t 1...Rc2 2.Nb4 Rxf2 3.Na6 Re2 4.Nb8 Rxe5 5.Nxd7
{240}
71641 0.137 2.64 8t+ 1...Rc4 2.f4 Rxf4 3.Nc1 Rc4 4.Kg1
118836 0.218 2.97 8t 1...Rc4 2.f4 f6 3.Kg3 fxe5 4.fxe5 Re4H 5.Kf3H
Rxe5H {240}
125262 0.233 2.97 8. 1...Rc4 2.f4 f6 3.Kg3 fxe5 4.fxe5 Re4 5.Kf3
Rxe5 {240}
184350 0.347 2.88 9t 1...Rc4 2.Kg3 f6 3.f4 fxe5 4.fxe5H Re4H 5.Kf3H
Rxe5H 6.Nc3H {240}
270072 0.483 2.88 9. 1...Rc4 2.Kg3 f6 3.f4 fxe5 4.fxe5 Re4 5.Kf3
Rxe5 6.Nc3 {240}
358813 0.635 3.01 10t 1...Rc4 2.Kg3 f6 3.f4 fxe5 4.fxe5 Re4 5.Kf3
Rxe5 6.g4 Kf6 {240}
499857 0.846 3.01 10. 1...Rc4 2.Kg3 f6 3.f4 fxe5 4.fxe5 Re4 5.Kf3
Rxe5 6.g4 Kf6 {240}
727482 1.218 3.12 11t 1...Rc4 2.Kg3 f6 3.f4 fxe5 4.fxe5 Re4 5.Kf3
Rxe5 6.Nc3 Rf5+ 7.Ke4 Kf6H {240}
1022041 1.651 3.12 11. 1...Rc4 2.Kg3 f6 3.f4 fxe5 4.fxe5 Re4 5.Kf3
Rxe5 6.Nc3 Rf5+ 7.Ke4 Kf6 {240}
1878685 3.086 3.24 12t 1...Rc4 2.Kg3 g5 3.Kf3 f6 4.exf6+ Kxf6 5.Ke3
Ke5 6.Kd3 Rd4+ 7.Ke3 Rxd6 8.Nc3 Rc6 {240}
1972033 3.245 3.25 12t+ 1...Rc2 2.Nb4 Rxf2 3.Nd3 Re2 4.h4 Re3 5.h5
1987067 3.272 3.52 12++ 1...Rc2 2.Nb4 Rxf2 3.Nd3 Re2 4.h4 Re3 5.Nc5
Rxe5 6.h5 Rxh5+ 7.Kg3
2294120 3.797 3.71 12t 1...Rc2 2.Nb4 Rxf2 3.Kg3 Re2 4.Kf4 g5+ 5.Kxg5
Rxe5+ 6.Kf4 Rb5 7.Nd3 Rd5 8.Ne5 Rxd6 {320}
2711667 4.401 3.71 12. 1...Rc2 2.Nb4 Rxf2 3.Kg3 Re2 4.Kf4 g5+ 5.Kxg5
Rxe5+ 6.Kf4 Rb5 7.Nd3 Rd5 8.Ne5 Rxd6 {320}
3376042 5.482 3.77 13t 1...Rc2 2.Nb4 Rxf2 3.Kg3 Re2 4.Kf4 g5+ 5.Kxg5
Rxe5+ 6.Kf4 Rf5+ 7.Ke3 Rb5 8.Nd3 Rd5 9.g4 Rxd6H
{320}
4759636 7.551 3.77 13. 1...Rc2 2.Nb4 Rxf2 3.Kg3 Re2 4.Kf4 g5+ 5.Kxg5
Rxe5+ 6.Kf4 Rf5+ 7.Ke3 Rb5 8.Nd3 Rd5 9.g4 Rxd6
{320}
6347787 10.161 3.86 14t 1...Rc2 2.Nb4 Rxf2 3.Kg3 Rb2 4.Na6 Rb5 5.Nc7
Rxe5 6.Ne8+H Kh6H 7.Nf6H Rc5H {320}
9497685 14.950 3.86 14. 1...Rc2 2.Nb4 Rxf2 3.Kg3 Rb2 4.Na6 Rb5 5.Nc7
Rxe5 6.Ne8+ Kh6 7.Nf6 Rc5 {320}
13190497 20.804 3.98 15t 1...Rc2 2.Nb4 Rxf2 3.Kg3 Rb2 4.Na6 Re2 5.Nc7 g5
6.Ne8+ Kg6 7.Kf3 Rxe5 8.g4 f5H 9.Nc7H {320}
19883574 31.358 3.98 15. 1...Rc2 2.Nb4 Rxf2 3.Kg3 Rb2 4.Na6 Re2 5.Nc7 g5
6.Ne8+ Kg6 7.Kf3 Rxe5 8.g4 f5 9.Nc7 {320}
29553430 47.073 4.00 16t 1...Rc2 2.Nb4 Rxf2 3.Kg3 Rb2 4.Na6 Re2 5.Kf4 h6
6.Nc7 g5+ 7.Kf3 Rxe5H 8.Ne8+H Kg6H 9.g4H f5H
10.Nc7H {320}
58916992 1:34.0 4.00 16. 1...Rc2 2.Nb4 Rxf2 3.Kg3 Rb2 4.Na6 Re2 5.Kf4 h6
6.Nc7 g5+ 7.Kf3 Rxe5 8.Ne8+ Kg6 9.g4 f5 10.Nc7
{320}
79793362 2:08.2 4.06 17t 1...Rc2 2.Nb4 Rxf2 3.Kg3 Rb2 4.Na6 Re2 5.Kf4 h6
6.Nc5 g5+ 7.Kf3 Rxe5 8.Nxd7 Rd5 9.g4 {HT} {240}
141455966 3:48.0 4.06 17. 1...Rc2 2.Nb4 Rxf2 3.Kg3 Rb2 4.Na6 Re2 5.Kf4 h6
6.Nc5 g5+ 7.Kf3 Rxe5 8.Nxd7 Rd5 9.g4 Rxd6H {320}
213258450 5:49.1 4.31 18t 1...Rc2 2.Nb4 Rxf2 3.Nd3 Re2 4.Kg1 Re3 5.Nc5
Rxe5H 6.Nxd7H Rd5H 7.Kf2H Rxd6H 8.Ne5H {320}
247588254 6:45.8 4.31 18. 1...Rc2 2.Nb4 Rxf2 3.Nd3 Re2 4.Kg1 Re3 5.Nc5
Rxe5 6.Nxd7 Rd5 7.Kf2 Rxd6 8.Ne5 {320}
403290513 11:02.4 4.28 19t 1...Rc2 2.Nb4 Rxf2 3.Kg3 Re2 4.Kf4 Rxg2 5.Na6
g5+H 6.Kf3H h5H {HT}
597008442 16:19.4 4.28 19. 1...Rc2 2.Nb4 Rxf2 3.Kg3 Re2 4.Kf4 Rxg2 5.Na6
g5+ 6.Kf3 h5
607934100 16:37.9 4.28 19f. 1...Rc2 2.Nb4 Rxf2 3.Kg3 Re2 4.Kf4 Rxg2 5.Na6
g5+ 6.Kf3 h5
607934100 Nodes, 10.76% Leavenodes, 609157 Nodes/sec
232176913 eval, 94.50% score, 1745734622 genmoves, 4.29% captures le 570/354
ext: pawn 2752149, rcp 2173993, chk 23570046, repchk 1030650, null 15026, prune
492467
htable: 370692759 store, 0 rejected, 604893849 probe, 53.8% f/p, 87.8% f/s
entries 21333333 age 0 renew 0
egtb probes 55, found 55 max_depth 45
solution not found
1 tested, 0 found, 0 mates
607934100 nodes, 65388609 leavenodes in 999.62 sec 608166 n/s
test nodes 607934100 win nodes 0 mate nodes 0
win time 0.00 mate time 0.00 av depth 19.000 (nm 19.000) maxdepth 45, tu 606
log(win_nodes) = 0.00000 log(win_time) = 0.00000
black ( 1):
Crafty never considers Rc4 (except as the second move to analyze):
EPD Kit revision date: 1996.04.21
unable to open book file [e:\crafty\release/books.bin].
hash table memory = 192M bytes.
pawn hash table memory = 80M bytes.
EGTB cache memory = 32M bytes.
draw score set to 0.00 pawns.
choose from book moves randomly (using weights.)
choose from 5 best moves.
book learning enabled
result learning enabled
position learning enabled
threshold set to 9 pawns.
5 piece tablebase files found
19045kb of RAM used for TB indices and decompression tables
Crafty v18.11
White(1): st 999
search time set to 999.00.
White(1): epdpfga \n.epd n.out
PFGA: EPD record: 1 ID: Albert Silver, CCC post 193195
end-game phase
clearing hash tables
time surplus 0.00 time limit 16:39 (16:39)
nss depth time score variation (1)
9-> 0.23 -3.27 1. ... Rc2 2. Nb4 Rxf2 3. Na6 Re2 4.
Nb8 Rxe5 5. Nxd7 Rd5
10 0.29 -3.53 1. ... Rc2 2. Nb4 Rxf2 3. Kg1 Rd2 4.
Na6 f6 5. Nc5 fxe5 6. Nxd7 Rxd6 7.
Nxe5
10-> 0.55 -3.53 1. ... Rc2 2. Nb4 Rxf2 3. Kg1 Rd2 4.
Na6 f6 5. Nc5 fxe5 6. Nxd7 Rxd6 7.
Nxe5
11 0.61 -3.53 1. ... Rc2 2. Nb4 Rxf2 3. Kg1 Rd2 4.
Na6 f6 5. Nc5 fxe5 6. Nxd7 Rxd6 7.
Nxe5
11-> 1.01 -3.53 1. ... Rc2 2. Nb4 Rxf2 3. Kg1 Rd2 4.
Na6 f6 5. Nc5 fxe5 6. Nxd7 Rxd6 7.
Nxe5
12 1.13 ++ 1. ... Rc2!!
12 4.13 -4.07 1. ... Rc2 2. Nb4 Rxf2 3. Kg3 Re2 4.
Kf4 f6 5. Nd3 fxe5+ 6. Kf3 Rd2 7. Ke4
Kf6 8. Nxe5 Rxd6
12-> 5.75 -4.07 1. ... Rc2 2. Nb4 Rxf2 3. Kg3 Re2 4.
Kf4 f6 5. Nd3 fxe5+ 6. Kf3 Rd2 7. Ke4
Kf6 8. Nxe5 Rxd6
13 6.51 -4.00 1. ... Rc2 2. Nb4 Rxf2 3. Kg3 Re2 4.
Kf4 f5 5. Nd3 Rxg2 6. Nc5 g5+ 7. Kf3
Rh2 8. Nxd7 <HT>
13-> 10.99 -4.00 1. ... Rc2 2. Nb4 Rxf2 3. Kg3 Re2 4.
Kf4 f5 5. Nd3 Rxg2 6. Nc5 g5+ 7. Kf3
Rh2 8. Nxd7 <HT>
14 12.93 -4.33 1. ... Rc2 2. Nb4 Rxf2 3. Kg3 Re2 4.
Kf4 Rxg2 5. Na6 Rf2+ 6. Ke4 f5+ 7.
exf6+ Rxf6 8. h4 Rf5 9. Nc7 Kf6 10.
Ne8+ Kf7 11. Nc7
14-> 25.11 -4.33 1. ... Rc2 2. Nb4 Rxf2 3. Kg3 Re2 4.
Kf4 Rxg2 5. Na6 Rf2+ 6. Ke4 f5+ 7.
exf6+ Rxf6 8. h4 Rf5 9. Nc7 Kf6 10.
Ne8+ Kf7 11. Nc7
15 29.32 -4.46 1. ... Rc2 2. Nb4 Rxf2 3. Kg3 Re2 4.
Kf4 Rxg2 5. Na6 Rf2+ 6. Ke4 Re2+ 7.
Kd4 Rd2+ 8. Kc3 Rh2 9. Kd4 Rxh3 10.
Nc7
15-> 1:01 -4.46 1. ... Rc2 2. Nb4 Rxf2 3. Kg3 Re2 4.
Kf4 Rxg2 5. Na6 Rf2+ 6. Ke4 Re2+ 7.
Kd4 Rd2+ 8. Kc3 Rh2 9. Kd4 Rxh3 10.
Nc7
16 1:18 -4.36 1. ... Rc2 2. Nb4 Rxf2 3. Kg3 Re2 4.
Kf4 Rxg2 5. Na6 Rf2+ 6. Ke3 Rf5 7.
Ke4 f6 8. exf6+ Rxf6 9. Nc5 Rf7 10.
Nd3 g5 11. Ne5 <HT>
16-> 3:19 -4.36 1. ... Rc2 2. Nb4 Rxf2 3. Kg3 Re2 4.
Kf4 Rxg2 5. Na6 Rf2+ 6. Ke3 Rf5 7.
Ke4 f6 8. exf6+ Rxf6 9. Nc5 Rf7 10.
Nd3 g5 11. Ne5 <HT>
17 4:02 -4.39 1. ... Rc2 2. Nb4 Rxf2 3. Kg3 Rb2 4.
Na6 Re2 5. Kf4 h6 6. Nc7 g5+ 7. Kf3
Rxe5 8. g3 Rc5 9. Ke4 Rc3 10. Ne8+
Kg6
17-> 9:27 -4.39 1. ... Rc2 2. Nb4 Rxf2 3. Kg3 Rb2 4.
Na6 Re2 5. Kf4 h6 6. Nc7 g5+ 7. Kf3
Rxe5 8. g3 Rc5 9. Ke4 Rc3 10. Ne8+
Kg6
18 9:30 1/23* 1. ... Rc2
ChessMaster likes Rc4 up to 16 seconds, then switches to Rc2.
Time Depth Score Positions Moves
0:07 9/11 -4.15 1613340 1...Rc4 2. Kg3 g5 3. Kf3 f6 4.
exf6+ Kxf6 5. Ke3 Ke5 6. Kd3 Rd4+
7. Ke3 Re4+ 8. Kf3 Rf4+ 9. Ke3
Kxd6
0:16 10/12 -4.22 3593235 1...Rc4 2. Kg3 g5 3. Kf3 f6 4.
exf6+ Kxf6 5. Ke3 Ke5 6. Kd3 Rd4+
7. Ke3 Rxd6 8. Nc3 h5
0:31 10/12 -5.35 6920257 1...Rc2 2. Nb4 Rxf2 3. Kg3 Re2
4. Kf4 Rxg2 5. Nd3 f6 6. exf6+
Kxf6 7. h4 Re2
0:38 11/13 -5.15 8619286 1...Rc2 2. Nb4 Rxf2 3. Kg3 Re2
4. Kf4 g5+ 5. Kxg5 Rxe5+ 6. Kf4
Re2 7. g3 Rd2 8. Ke4 Rxd6 9. Nd3
1:51 12/14 -5.18 25187375 1...Rc2 2. Nb4 Rxf2 3. Kg3 Re2
4. Kf4 g5+ 5. Kxg5 Rxe5+ 6. Kf4
Re2 7. g4 Rd2 8. Ke5 f6+ 9. Ke4
Rxd6 10. Nd3
5:47 13/15 -5.50 78677604 1...Rc2 2. Nb4 Rxf2 3. Kg3 Re2
4. Na6 Rxe5 5. Nb8 Rd5 6. Nxd7
Rxd6 7. Nc5 Rd5 8. Ne4 f5 9. Nc3
Even though all the computers like Rc2 and it may conceivably be the better
move, I definitely like Rc4 a lot better. I think after Rc4, you can "mail it
in" and win the game on autopilot without even having to think very hard. The
move Rc2 requires continued concentration.
I think that psychologically, Rc4 is *by far* more devastating. Well, maybe
that's an exaggeration. Rc2 is going to win unless there is a blunder. But I
would feel like a sucker if I took the pawn and then saw that rook move.
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.