Test position: One more for the humans (Kurnosov-Najer)

[Albert Silver]

In the 8th round game from the Russian Team championship between Kurnosov and Najer, Najer had been leading the Elo performers (2900+) with Kurnosov right behind. They reached this position:

[D]1rr3k1/3b1pq1/4p2p/1n1pP2P/pp3QP1/8/PPP5/1K1RNB1R b - - 0 29

after which Najer played Rf8. Instead had he played 29...b3 30.axb3 axb3 he would have had a decisive attack. The engines proposed 31.c4 as an attempt to defend, but fail to find the winning shot for Black. At least they did not find it here, though my hardware is not the best.

[D]1rr3k1/3b1pq1/4p2p/1n1pP2P/2P2QP1/1p6/1P6/1K1RNB1R b - c3 0 31

The best move (AFAIK) is 31...Nc3+!! after which 32.bxc3 Qf8 (and Ra8) wins.

[bob]

In the 8th round game from the Russian Team championship between Kurnosov and Najer, Najer had been leading the Elo performers (2900+) with Kurnosov right behind. They reached this position:

[D]1rr3k1/3b1pq1/4p2p/1n1pP2P/pp3QP1/8/PPP5/1K1RNB1R b - - 0 29

after which Najer played Rf8. Instead had he played 29...b3 30.axb3 axb3 he would have had a decisive attack. The engines proposed 31.c4 as an attempt to defend, but fail to find the winning shot for Black. At least they did not find it here, though my hardware is not the best.

[D]1rr3k1/3b1pq1/4p2p/1n1pP2P/2P2QP1/1p6/1P6/1K1RNB1R b - c3 0 31

The best move (AFAIK) is 31...Nc3+!! after which 32.bxc3 Qf8 (and Ra8) wins.

I'm not quite sure how this is "one more for the humans"??? Black misplayed the position and didn't push the attack. Nc3 might not be the best move, although it is part of a potentially winning plan. Ra8 first is also difficult to answer, as it doesn't seem white has time to push the g-pawn to further his own attacking plan. If he does, he could after Nc3 bxc3 also. I let crafty stew on the second position for a half-hour on fast hardware and it seems to like Ra8 first. +0.5 is all it sees however. On the first position, it chooses b3 instantly and doesn't change its mind at all, at least for a couple of minutes...

Seems like the computers might do better than the humans in this one...

[Mike S.]

The following results are from 32-bit singlecore, P4/2.4 GHZ, 128 MB hash only:

Analysis by Fritz 12:

31...Nc7 (...)
31...dxc4 (...)

31...Qf8 32.Rh3 Na3+ 33.bxa3 Ra8 34.Rxb3 Ba4 35.Rg3 Bxd1 36.Kc1 Rxa3 37.Rxa3 Qxa3+ 38.Kxd1 Qa1+ 39.Kd2 dxc4 40.Ke2 Rb8
=/+ (-0.35) Depth: 16/47 00:01:30 74532kN
=/+ (-0.35) Depth: 18/47 00:03:00 156mN

Fritz has a very similar idea here, with Qf8 first followed by Na3+. But why not 31...Nc3+?

[D]1rr3k1/3b1pq1/4p2p/3pP2P/2P2QP1/1pn5/1P6/1K1RNB1R w - - 0 32

32.bxc3 Qf8 33.Bd3 Ra8 34.Rf1 Ra5 35.Bh7+ Kxh7 36.Qxf7+ Qg7 37.g5 hxg5 38.h6 Qxf7 39.Rxf7+ Kxh6 40.Rxd7 Rca8 41.Kb2
+/- (1.27) Depth: 10/28 00:00:00 360kN
+/= (0.57) Depth: 17/45 00:01:09 58256kN

Fritz creates counter-threats with Bd3 and Rf1, but the eval dropped.


Analysis by Critter 1.01 32-bit+GTB:

(...)
31...dxc4 32.Bxc4 Bc6 33.Rh3 Nc7 34.Rxb3 Ra8 35.Rc1 Ra4 36.Nd3 Rca8 37.Rb4 Ra1+ 38.Kc2 Rxc1+ 39.Kxc1 Nd5 40.Bxd5 Bxd5 41.Ra4 Rc8+ 42.Kb1
= (-0.14) Depth: 10/36 00:00:02 704kN
= (-0.12) Depth: 15/36 00:00:15 5538kN
31...Qf8 32.Rh3 Ra8 33.Rxb3 Na3+ 34.Rxa3 Rxa3 35.g5 Rb8 36.Rd2 hxg5 37.Qxg5+ Qg7 38.Qxg7+ Kxg7 39.Nc2 Rc3 40.cxd5 exd5 41.Rxd5 Be6 42.Rd2 Kh6 43.Kc1
= (-0.21 ++) Depth: 15/37 00:00:23 8343kN
-/+ (-0.78 --) Depth: 16/52 00:02:54 65775kN

Analysis by Critter 1.01 32-bit+GTB after 31...Nc3+:

32.bxc3 Qf8 33.Rh2 Ra8 34.Rf2 Be8 35.Qc1 dxc4 36.Bg2 Qc5 37.Rd4 Ra5 38.Qxh6 Qa3 39.Qg5+ Kf8 40.Qh6+
+/- (0.94) Depth: 6/14 00:00:00 13kN
-/+ (-0.77 --) Depth: 16/68 00:03:58 85886kN

"Endless" 16th ply... Fritz had Rh1-h2-f2 too but switched to Bd3 and Rh1-f1, which seems more convincing.

[Albert Silver]

In the 8th round game from the Russian Team championship between Kurnosov and Najer, Najer had been leading the Elo performers (2900+) with Kurnosov right behind. They reached this position:

[D]1rr3k1/3b1pq1/4p2p/1n1pP2P/pp3QP1/8/PPP5/1K1RNB1R b - - 0 29

after which Najer played Rf8. Instead had he played 29...b3 30.axb3 axb3 he would have had a decisive attack. The engines proposed 31.c4 as an attempt to defend, but fail to find the winning shot for Black. At least they did not find it here, though my hardware is not the best.

[D]1rr3k1/3b1pq1/4p2p/1n1pP2P/2P2QP1/1p6/1P6/1K1RNB1R b - c3 0 31

The best move (AFAIK) is 31...Nc3+!! after which 32.bxc3 Qf8 (and Ra8) wins.

I'm not quite sure how this is "one more for the humans"??? Black misplayed the position and didn't push the attack. Nc3 might not be the best move, although it is part of a potentially winning plan. Ra8 first is also difficult to answer, as it doesn't seem white has time to push the g-pawn to further his own attacking plan. If he does, he could after Nc3 bxc3 also. I let crafty stew on the second position for a half-hour on fast hardware and it seems to like Ra8 first. +0.5 is all it sees however. On the first position, it chooses b3 instantly and doesn't change its mind at all, at least for a couple of minutes...

Seems like the computers might do better than the humans in this one...

Nc3 is really the best move in the second position. I only gave the first as a means of explaining how it was reached. Ra8 and Qf8 don't really cut it though they do give Black an edge.

Since I found Nc3 and the engines could not, and I consider myself human... The engines realize this when they try to disprove Nc3 and you compare the resulting positions to the Ra8 and Qf8 they choose. Nc3 leads to +2.xx (minimum) in all lines.

[Terry McCracken]

In the 8th round game from the Russian Team championship between Kurnosov and Najer, Najer had been leading the Elo performers (2900+) with Kurnosov right behind. They reached this position:

[D]1rr3k1/3b1pq1/4p2p/1n1pP2P/pp3QP1/8/PPP5/1K1RNB1R b - - 0 29

after which Najer played Rf8. Instead had he played 29...b3 30.axb3 axb3 he would have had a decisive attack. The engines proposed 31.c4 as an attempt to defend, but fail to find the winning shot for Black. At least they did not find it here, though my hardware is not the best.

[D]1rr3k1/3b1pq1/4p2p/1n1pP2P/2P2QP1/1p6/1P6/1K1RNB1R b - c3 0 31

The best move (AFAIK) is 31...Nc3+!! after which 32.bxc3 Qf8 (and Ra8) wins.

I'm not quite sure how this is "one more for the humans"??? Black misplayed the position and didn't push the attack. Nc3 might not be the best move, although it is part of a potentially winning plan. Ra8 first is also difficult to answer, as it doesn't seem white has time to push the g-pawn to further his own attacking plan. If he does, he could after Nc3 bxc3 also. I let crafty stew on the second position for a half-hour on fast hardware and it seems to like Ra8 first. +0.5 is all it sees however. On the first position, it chooses b3 instantly and doesn't change its mind at all, at least for a couple of minutes...

Seems like the computers might do better than the humans in this one...

Nc3 is really the best move in the second position. I only gave the first as a means of explaining how it was reached. Ra8 and Qf8 don't really cut it though they do give Black an edge.

Since I found Nc3 and the engines could not, and I consider myself human... The engines realize this when they try to disprove Nc3 and you compare the resulting positions to the Ra8 and Qf8 they choose. Nc3 leads to +2.xx (minimum) in all lines.

It's a great position, thanks!

Would you post the full game as well? Thanks.

[Albert Silver]

The following results are from 32-bit singlecore, P4/2.4 GHZ, 128 MB hash only:

Analysis by Fritz 12:

31...Nc7 (...)
31...dxc4 (...)

31...Qf8 32.Rh3 Na3+ 33.bxa3 Ra8 34.Rxb3 Ba4 35.Rg3 Bxd1 36.Kc1 Rxa3 37.Rxa3 Qxa3+ 38.Kxd1 Qa1+ 39.Kd2 dxc4 40.Ke2 Rb8
=/+ (-0.35) Depth: 16/47 00:01:30 74532kN
=/+ (-0.35) Depth: 18/47 00:03:00 156mN

Fritz has a very similar idea here, with Qf8 first followed by Na3+. But why not 31...Nc3+?

[D]1rr3k1/3b1pq1/4p2p/3pP2P/2P2QP1/1pn5/1P6/1K1RNB1R w - - 0 32

32.bxc3 Qf8 33.Bd3 Ra8 34.Rf1 Ra5 35.Bh7+ Kxh7 36.Qxf7+ Qg7 37.g5 hxg5 38.h6 Qxf7 39.Rxf7+ Kxh6 40.Rxd7 Rca8 41.Kb2
+/- (1.27) Depth: 10/28 00:00:00 360kN
+/= (0.57) Depth: 17/45 00:01:09 58256kN

Fritz creates counter-threats with Bd3 and Rf1, but the eval dropped.


Analysis by Critter 1.01 32-bit+GTB:

(...)
31...dxc4 32.Bxc4 Bc6 33.Rh3 Nc7 34.Rxb3 Ra8 35.Rc1 Ra4 36.Nd3 Rca8 37.Rb4 Ra1+ 38.Kc2 Rxc1+ 39.Kxc1 Nd5 40.Bxd5 Bxd5 41.Ra4 Rc8+ 42.Kb1
= (-0.14) Depth: 10/36 00:00:02 704kN
= (-0.12) Depth: 15/36 00:00:15 5538kN
31...Qf8 32.Rh3 Ra8 33.Rxb3 Na3+ 34.Rxa3 Rxa3 35.g5 Rb8 36.Rd2 hxg5 37.Qxg5+ Qg7 38.Qxg7+ Kxg7 39.Nc2 Rc3 40.cxd5 exd5 41.Rxd5 Be6 42.Rd2 Kh6 43.Kc1
= (-0.21 ++) Depth: 15/37 00:00:23 8343kN
-/+ (-0.78 --) Depth: 16/52 00:02:54 65775kN

Analysis by Critter 1.01 32-bit+GTB after 31...Nc3+:

32.bxc3 Qf8 33.Rh2 Ra8 34.Rf2 Be8 35.Qc1 dxc4 36.Bg2 Qc5 37.Rd4 Ra5 38.Qxh6 Qa3 39.Qg5+ Kf8 40.Qh6+
+/- (0.94) Depth: 6/14 00:00:00 13kN
-/+ (-0.77 --) Depth: 16/68 00:03:58 85886kN

"Endless" 16th ply... Fritz had Rh1-h2-f2 too but switched to Bd3 and Rh1-f1, which seems more convincing.

Well, not sure about the endless ply issue, but after 33.Bd3 Ra8 34.Rf1 as suggested, here is what Critter says:


[D]r1r2qk1/3b1p2/4p2p/3pP2P/2P2QP1/1pPB4/8/1K1RNR2 b - - 0 1

Analysis by Critter 1.0 64-bit:

34...Be8 35.c5 Qxc5 36.Qf2 Qxc3 37.Qb2 Qa5 38.g5 Qb4 39.Qd2 Qa4 40.Qb2 hxg5 41.Rg1 g4 42.Be2 f5 43.exf6 Bxh5 44.Rd2 Kf8 45.Bd3 Qf4 46.Ng2 Qg5 47.Rc1 Rxc1+ 48.Kxc1 g3 49.Qc3 Kf7 50.Qxb3 Kxf6 51.Qb4 Qe5 52.Qh4+ Kf7 53.Qf4+ Qxf4 54.Nxf4
-+ (-1.91 ++) Depth: 18/65 00:00:46 155mN
34...Be8 35.c5 Qxc5 36.Qf2 Qxc3 37.Qb2 Qa5 38.g5 Qb4 39.Qd2 Qa4 40.Qb2 hxg5 41.Rg1 g4 42.Be2 f5 43.exf6 Bxh5 44.Rd2 Kf8 45.Bd3 Qf4 46.Ng2 Qg5 47.Rc1 Rxc1+ 48.Kxc1 g3 49.Qc3 Kf7 50.Qxb3 Kxf6 51.Qb4 Qe5 52.Qh4+ Kf7 53.Qf4+ Qxf4 54.Nxf4
-+ (-2.01 ++) Depth: 18/65 00:00:54 178mN
34...Be8 35.c5 Qxc5 36.Qf2 Qxc3 37.Qb2 Qa5 38.g5 Qb4 39.Qd2 Qa4 40.Qb2 hxg5 41.Rg1 g4 42.Be2 f5 43.exf6 Bxh5 44.Rd2 Kf8 45.Bd3 Qf4 46.Ng2 Qg5 47.Rc1 Rxc1+ 48.Kxc1 g3 49.Qc3 Kf7 50.Qxb3 Kxf6 51.Qb4 Qe5 52.Qh4+ Kf7 53.Qf4+ Qxf4 54.Nxf4
-+ (-2.18 ++) Depth: 18/65 00:01:01 200mN
34...Be8 35.c5 Qxc5 36.Qf2 Qxc3 37.Qb2 Qa5 38.g5 Qb4 39.Qd2 Qa4 40.Qb2 hxg5 41.Rg1 g4 42.Be2 f5 43.exf6 Bxh5 44.Rd2 Kf8 45.Bd3 Qf4 46.Ng2 Qg5 47.Rc1 Rxc1+ 48.Kxc1 g3 49.Qc3 Kf7 50.Qxb3 Kxf6 51.Qb4 Qe5 52.Qh4+ Kf7 53.Qf4+ Qxf4 54.Nxf4
-+ (-2.42 ++) Depth: 18/65 00:01:17 251mN
34...Be8 35.c5 Qxc5 36.Qf2 Qxc3 37.Qb2 Qa5 38.g5 Qb4 39.Qd2 Qa4 40.Qb2 hxg5 41.Rg1 g4 42.Be2 f5 43.exf6 Bxh5 44.Rd2 Kf8 45.Bd3 Qf4 46.Ng2 Qg5 47.Rc1 Rxc1+ 48.Kxc1 g3 49.Qc3 Kf7 50.Qxb3 Kxf6 51.Qb4 Qe5 52.Qh4+ Kf7 53.Qf4+ Qxf4 54.Nxf4
-+ (-2.79 ++) Depth: 18/65 00:01:49 355mN
34...Be8 35.Qh2 dxc4 36.Bg6 Ra2 37.Rf2 Qa3 38.Rxa2 bxa2+ 39.Ka1 Ba4 40.Bxf7+ Kg7 41.Qb2 Qxb2+ 42.Kxb2 Bxd1 43.Bxe6 Rc5 44.Ng2 Rxe5 45.Bxc4 Bxg4 46.Kxa2 Bxh5 47.Nf4 Be8 48.Kb3 h5 49.Bd5 h4 50.Kc4 Rf5 51.Ne6+ Kf6 52.Nd4 Rf1 53.Bg2 Rg1 54.Bh3
-+ (-3.04) Depth: 18/65 00:02:36 510mN
34...Be8 35.Qc1 b2 36.Qxb2 Rcb8 37.Nc2 Rxb2+ 38.Kxb2 Rb8+ 39.Kc1 dxc4 40.Bxc4 Qc5 41.Rd4 Qxe5 42.Rfd1 Qg5+ 43.R1d2 Qa5 44.R2d3 e5 45.Rd6 Qc5 46.Rd8 Rxd8 47.Rxd8 Kf8 48.Bd3 Qg1+ 49.Kd2 Qxg4 50.Ne3 Qxh5 51.Be4 Qg5
-+ (-3.19) Depth: 19/65 00:03:09 615mN
34...Be8 35.Qc1 b2 36.Qxb2 Rcb8 37.Nc2 Rxb2+ 38.Kxb2 Rb8+ 39.Kc1 dxc4 40.Bxc4 Qc5 41.Rd4 Qxe5 42.Rfd1 Qg5+ 43.R1d2 Qa5 44.R2d3 e5 45.Rd6 Qc5 46.Rd8 Rxd8 47.Rxd8 Kf8 48.Bd3 Qg1+ 49.Kd2 Qxg4 50.Ne3 Qh4 51.Ra8 Qf2+ 52.Be2 f5 53.Nd5 f4 54.Nf6 Qe3+ 55.Kd1 Qg1+ 56.Kc2
-+ (-3.39 ++) Depth: 20/72 00:05:34 1108mN
34...Be8 35.Qc1 b2 36.Qxb2 Rcb8 37.Nc2 Rxb2+ 38.Kxb2 Rb8+ 39.Kc1 dxc4 40.Bxc4 Qc5 41.Rd4 Qxe5 42.Rfd1 Qg5+ 43.R1d2 Qa5 44.R2d3 Bc6 45.Re3 Kg7 46.Be2 Qg5 47.Rc4 Bd5 48.Rd4 Rb3 49.Rd2 Bc6 50.Bc4 Rb8 51.Be2 Ra8 52.Red3 Qc5 53.Kb2 Qa7 54.Kc1 Bd5 55.Rxd5 exd5 56.Rxd5 Qb7
-+ (-3.58 ++) Depth: 20/72 00:08:01 1600mN
34...Be8 35.Qc1 b2 36.Qxb2 Rcb8 37.Nc2 Rxb2+ 38.Kxb2 Rb8+ 39.Kc1 dxc4 40.Bxc4 Qc5 41.Rd4 Qxe5 42.Rfd1 Qg5+ 43.R1d2 Qa5 44.R2d3 Bc6 45.Re3 Kg7 46.Be2 Qg5 47.Rc4 Bd5 48.Rd4 Rb3 49.Rd2 Bc6 50.Bc4 Rb8 51.Be2 Ra8 52.Red3 Qc5 53.Kb2 Qa7 54.Kc1 Bd5 55.Rxd5 exd5 56.Rxd5 Qb7
-+ (-3.88 ++) Depth: 20/78 00:10:38 2113mN
34...Be8 35.Qc1 b2 36.Qxb2 Rab8 37.Nc2 Rxb2+ 38.Kxb2 dxc4 39.Be4 Rb8+ 40.Kc1 Qg7 41.Rfe1 Qg5+ 42.Rd2 Rb3 43.Re3 Ba4 44.Bg2 Rb8 45.Rde2 Bxc2 46.Kxc2 Qxg4 47.Rf2 Qxh5 48.Be4 Rd8 49.Ree2 Kf8 50.Rh2 Qxe5 51.Rxh6 Ke7 52.Rh3 f5 53.Bf3 Qd6 54.Rg2 f4 55.Rh7+
-+ (-4.07) Depth: 20/82 00:13:23 2650mN

[Albert Silver]

In the 8th round game from the Russian Team championship between Kurnosov and Najer, Najer had been leading the Elo performers (2900+) with Kurnosov right behind. They reached this position:

[D]1rr3k1/3b1pq1/4p2p/1n1pP2P/pp3QP1/8/PPP5/1K1RNB1R b - - 0 29

after which Najer played Rf8. Instead had he played 29...b3 30.axb3 axb3 he would have had a decisive attack. The engines proposed 31.c4 as an attempt to defend, but fail to find the winning shot for Black. At least they did not find it here, though my hardware is not the best.

[D]1rr3k1/3b1pq1/4p2p/1n1pP2P/2P2QP1/1p6/1P6/1K1RNB1R b - c3 0 31

The best move (AFAIK) is 31...Nc3+!! after which 32.bxc3 Qf8 (and Ra8) wins.

I'm not quite sure how this is "one more for the humans"??? Black misplayed the position and didn't push the attack. Nc3 might not be the best move, although it is part of a potentially winning plan. Ra8 first is also difficult to answer, as it doesn't seem white has time to push the g-pawn to further his own attacking plan. If he does, he could after Nc3 bxc3 also. I let crafty stew on the second position for a half-hour on fast hardware and it seems to like Ra8 first. +0.5 is all it sees however. On the first position, it chooses b3 instantly and doesn't change its mind at all, at least for a couple of minutes...

Seems like the computers might do better than the humans in this one...

Nc3 is really the best move in the second position. I only gave the first as a means of explaining how it was reached. Ra8 and Qf8 don't really cut it though they do give Black an edge.

Since I found Nc3 and the engines could not, and I consider myself human... The engines realize this when they try to disprove Nc3 and you compare the resulting positions to the Ra8 and Qf8 they choose. Nc3 leads to +2.xx (minimum) in all lines.

It's a great position, thanks!

Would you post the full game as well? Thanks.

Here is the game with the notes I published. I focused on the turning point of course, and kept the notes to the essential (readers start rolling their eyes if we put anything too extensive).


[Event "TCh-RUS Premier"]
[Site "Olginka RUS"]
[Date "2011.04.22"]
[Round "10"]
[White "Kurnosov, Igor"]
[Black "Najer, Evgeniy"]
[Result "1-0"]
[ECO "C11"]
[WhiteElo "2653"]
[BlackElo "2638"]
[Annotator "Albert Silver"]
[PlyCount "107"]
[EventDate "2011.04.12"]
[WhiteTeam "Tomsk-400"]
[BlackTeam "ShSM-64"]

1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. e5 Nfd7 5. f4 c5 6. Nf3 Nc6 7. Be3 Be7 8. Qd2
O-O 9. dxc5 Bxc5 10. O-O-O Qa5 11. Bxc5 Nxc5 12. h4 Bd7 13. h5 b5 14. Kb1 b4
15. Ne2 Ne4 16. Qe1 Rab8 17. Nd2 Nxd2+ 18. Qxd2 Qc5 19. Nc1 a5 20. Nd3 Qe7 21.
g4 a4 22. Bh3 Rfc8 23. f5 Nd4 24. f6 Qd8 25. Qg5 Qf8 26. Ne1 Nb5 27. Bf1 h6 28.
fxg7 Qxg7 29. Qf4 $2 {#Najer was doing very well and misses a winning
continuation here.} Rf8 ({Here he might have finished his amazing journey in
style had he played} 29... b3 $1 30. axb3 axb3 {#and both White's main
choices fail.} {The alternative} 31. c4 {also meets a spectacular demise with#}
({The obvious} 31. cxb3 {is punished by} Nc3+ $1 32. bxc3 (32. Kc2 Ne2+ 33. Kd2
Nxf4) 32... Rxb3+ 33. Kc1 Rcxc3+ 34. Kd2 Qf8 {and Black is winning.}) 31...
Nc3+ $3 {Curiously, none of the engines seem to find this move.} 32. bxc3 Qf8
$1 {and though White has a number of choices, they all end up losing
significant material to avoid mate. Ex:} 33. Rh2 (33. Bd3 Ra8 34. Rf1 Be8 {and
White is lost.}) 33... Ra8 34. Rf2 Be8 {Protecting f7 and freeing the queen.}
35. Qc1 Ra4 {followed by Rca8 and White is lost.}) 30. Nf3 f6 31. exf6 Qxf6 32.
Qxf6 Rxf6 33. Ne5 Be8 34. Rg1 Rb7 35. g5 hxg5 36. Bd3 Rg7 $2 (36... Nd6 $1 37.
Rxg5+ Rg7 38. Rxg7+ Kxg7 39. Rg1+ Kh8 {and it is equal.} (39... Kh6 40. Ng4+))
37. Ng4 Rf4 38. h6 Rh7 39. Bxh7+ Kxh7 40. Rdf1 Bg6 41. Rxf4 gxf4 42. Rh1 f3 43.
Kc1 Nd6 44. Nf6+ Kh8 45. Rg1 f2 46. Rf1 Nf7 47. Ng4 Kh7 48. Rxf2 Bf5 49. Nf6+
Kg6 (49... Kxh6 $2 50. Nxd5 exd5 51. Rxf5) 50. Ng8 Nxh6 51. Nxh6 Kxh6 52. Rf4
b3 53. cxb3 axb3 54. a4 1-0

[Terry McCracken]

Thanks Albert! And thanks for the annotation!

[Eelco de Groot]

In the 8th round game from the Russian Team championship between Kurnosov and Najer, Najer had been leading the Elo performers (2900+) with Kurnosov right behind. They reached this position:

[D]1rr3k1/3b1pq1/4p2p/1n1pP2P/pp3QP1/8/PPP5/1K1RNB1R b - - 0 29

after which Najer played Rf8. Instead had he played 29...b3 30.axb3 axb3 he would have had a decisive attack. The engines proposed 31.c4 as an attempt to defend, but fail to find the winning shot for Black. At least they did not find it here, though my hardware is not the best.

[D]1rr3k1/3b1pq1/4p2p/1n1pP2P/2P2QP1/1p6/1P6/1K1RNB1R b - c3 0 31

The best move (AFAIK) is 31...Nc3+!! after which 32.bxc3 Qf8 (and Ra8) wins.

I'm not quite sure how this is "one more for the humans"??? Black misplayed the position and didn't push the attack. Nc3 might not be the best move, although it is part of a potentially winning plan. Ra8 first is also difficult to answer, as it doesn't seem white has time to push the g-pawn to further his own attacking plan. If he does, he could after Nc3 bxc3 also. I let crafty stew on the second position for a half-hour on fast hardware and it seems to like Ra8 first. +0.5 is all it sees however. On the first position, it chooses b3 instantly and doesn't change its mind at all, at least for a couple of minutes...

Seems like the computers might do better than the humans in this one...

Nc3 is really the best move in the second position. I only gave the first as a means of explaining how it was reached. Ra8 and Qf8 don't really cut it though they do give Black an edge.

Since I found Nc3 and the engines could not, and I consider myself human... The engines realize this when they try to disprove Nc3 and you compare the resulting positions to the Ra8 and Qf8 they choose. Nc3 leads to +2.xx (minimum) in all lines.

It is difficult to compare evaluations if one engine's eval is much more speculative. Maybe Rainbow Serpent's eval is too speculative, but I hope it is not too imprecise here... Nc3+ is second move after a while, well only after 33 minutes, which is not too good, but the variations with Na3 before that get high evals as well. Black promotes the b pawn to quen but still is about 3 pawns down! I hope it is enough. Anyway the eval is apparently much higher than Crafty's


1rr3k1/3b1pq1/4p2p/1n1pP2P/2P2QP1/1p6/1P6/1K1RNB1R b - -

Engine: eXperimental Tommy Cooper - Rainbow Serpent Build 148 Space (100), Pawn Structure (100, 100) (=defaults) (Athlon 2009 MHz, 256 MB)
changes in pawns.cpp, evaluate.cpp, search.cpp
by Tord Romstad, Marco Costalba, Joona Kiiski

1/01 0:00 +6.74 1...Ra8 (2.160) 2

2/04 0:00 +4.64 1...Ra8 2.Bd3 dxc4 3.Be4 (17.988) 22

3/04 0:00 +4.32 1...Ra8 2.cxb5 Ra2 3.Bd3 (87.063) 91

4/06 0:01 +4.16 1...Ra8 2.cxb5 Ra2 3.Bd3 d4 4.Qf6 (129.285) 125

5/08 0:02 +3.19 1...Ra8 2.cxb5 Ra2 3.Bd3 Bxb5 4.Rf1 Bd7
5.Rf2 (828.459) 339

6/14 0:05 +3.35 1...Ra8 2.cxb5 Be8 3.g5 Qh7+ 4.Bd3 f5
5.b6 hxg5 6.Qxg5+ Kf8 7.Qf6+ Bf7
8.b7 (2.061.138) 405

7/14 0:05 +2.66-- 1...Ra8 2.cxb5 Be8 3.g5 Qh7+ 4.Bd3 f5
5.b6 hxg5 6.Qxg5+ Kf8 7.Qf6+ Bf7
8.b7 (2.259.720) 410

7/12 0:05 +2.81 1...Na3+ 2.bxa3 Ra8 3.Bd3 Rxa3 4.Qf6 Qxg4
5.Nf3 Rca8 6.Qxh6 Ra1+ 7.Kb2 (2.402.087) 413

8/12 0:07 +2.10 1...Na3+ 2.bxa3 Qf8 3.Rd2 Qxa3 4.Rf2 f5
5.exf6 b2 6.f7+ Kh7 7.Bd3+ (3.329.772) 426

8/11 0:11 +3.06 1...Nc7 2.Rd2 Ne8 3.Bd3 Ra8 4.Nc2 bxc2+
5.Rxc2 dxc4 6.Be4 Qf8 (5.031.038) 442

9/13 0:12 +3.20 1...Nc7 2.Rd2 Ne8 3.Bd3 dxc4 4.Bxc4 Rb4
5.Rd4 Bb5 6.Bxb5 Rxb5 7.Rf1 Rxe5 (5.528.252) 443

10/12 0:14 +2.82 1...Nc7 2.Rd2 dxc4 3.Rxd7 Nd5 4.Rxd5 exd5
5.Bg2 c3 6.Bxd5 cxb2 7.Qxf7+ (6.338.655) 445

11/08 0:18 +2.30-- 1...Nc7 2.g5 hxg5 3.Qf6 Qxf6 4.exf6 g4
5.Rg1 (8.513.073) 454

11/06 0:21 +1.48-- 1...Nc7 2.Qf6 Qxg4 3.Be2 Qxe2 4.Rg1+ (9.762.428) 461

11/11 0:24 +1.56 1...Ra8 2.cxb5 Qf8 3.Rh3 Ra1+ 4.Kxa1 Rc1+
5.Rxc1 Qa8+ 6.Kb1 Qa2+ (11.646.849) 466

11/15 0:31 +2.94++ 1...Na3+ 2.bxa3 Qf8 3.Rd2 Ra8 4.Rf2 Be8
5.Qc1 Rxa3 6.Rf6 Rca8 7.Qxh6 Ra1+
8.Kb2 Qa3+ (14.701.607) 466

11/15 0:36 +3.47++ 1...Na3+ 2.bxa3 Qf8 3.Rd2 Ra8 4.Rf2 Be8
5.Qc1 Rxa3 6.Rf6 Rca8 7.Qxh6 Ra1+
8.Kb2 Qa3+ (17.214.552) 469

11/15 0:40 +2.43 1...Na3+ 2.bxa3 Qf8 3.Rh2 Ra8 4.Rf2 Be8
5.Qc1 Rxa3 6.Qf4 Rca8 7.Kc1 Ra1+
8.Kd2 R8a2+ (19.227.284) 471

11/11 0:43 +4.52++ 1...Qf8 2.cxb5 Ra8 3.Rh3 Ra1+ 4.Kxa1 Rc1+
5.Rxc1 Qa8+ 6.Kb1 Qa2+ (20.327.533) 472

11/11 0:53 +4.90 1...Qf8 2.Bd3 Na3+ 3.Kc1 Nxc4 4.Bxc4 Rxc4+
5.Qxc4 dxc4 6.Rxd7 Qc5 (25.200.704) 475

12/17 0:55 +4.80 1...Qf8 2.Bd3 Na3+ 3.bxa3 Qxa3 4.Qf2 Ra8
5.Bh7+ Kf8 6.Qb2 Qa7 7.Rh3 Rxc4
8.Rxb3 Bc6 9.Nc2 Re8 (26.276.371) 475

13/16 1:01 +4.16 1...Qf8 2.Bd3 Na3+ 3.bxa3 Qxa3 4.Qf2 Ra8
5.Bh7+ Kf8 6.Qb2 Qa4 7.cxd5 Qxg4
8.Nd3 Qg5 9.Qxb3 (29.135.382) 476

13/21 2:15 +4.36 1...Ra8 2.Rh3 Na3+ 3.bxa3 Rxa3 4.Rf3 Qf8
5.Qxf7+ Qxf7 6.Rxf7 Kxf7 7.Kb2 Rca8
8.Bd3 Bc6 9.Bg6+ Ke7 10.cxd5 Bxd5
11.Bd3 Ra2+ (64.622.224) 477

14/18 2:29 +4.36 1...Ra8 2.Rh3 Na3+ 3.bxa3 Rxa3 4.Rf3 Rca8
5.Kb2 Ra2+ 6.Kc3 Be8 7.cxd5 f5
8.exf6 Qa7 9.Bc4 Rb8 10.Qxh6 (71.831.088) 479

15/18 2:40 +4.08-- 1...Ra8 2.Rh3 Na3+ 3.bxa3 Rxa3 4.Rf3 Rca8
5.Kb2 Ra2+ 6.Kc3 Be8 7.cxd5 f5
8.exf6 Qa7 9.Bc4 Rb8 10.Qxh6 (77.113.023) 479

15/14 2:57 +4.09 1...Ra8 2.Rh3 Na3+ 3.bxa3 Rxa3 4.Rf3 Rca8
5.Kb2 Be8 6.Rb1 Qh7 7.Bd3 Qg7 8.Kc3 (85.202.474) 480

16/21 3:44 +4.33 1...Ra8 2.Rh3 Na3+ 3.bxa3 Rxa3 4.Rf3 Qf8
5.Qxf7+ Qxf7 6.Rxf7 Kxf7 7.cxd5 Rca8
8.dxe6+ Bxe6 9.Kb2 Ra2+ 10.Kc3 Ra1
11.Rxa1 Rxa1 (107.987.466) 480

17/21 3:49 +4.15 1...Ra8 2.Rh3 Na3+ 3.bxa3 Rxa3 4.Rf3 Qf8
5.Qxf7+ Qxf7 6.Rxf7 Kxf7 7.Kb2 Rca8
8.cxd5 Ra2+ 9.Kc3 Ra1 10.Rxa1 Rxa1
11.dxe6+ Bxe6 (110.627.451) 481

18/26 3:58 +4.36 1...Ra8 2.Rh3 Na3+ 3.bxa3 Rxa3 4.Rf3 Qf8
5.Qxf7+ Qxf7 6.Rxf7 Kxf7 7.Kb2 Rca8
8.Bd3 Ra2+ 9.Kc3 Bc6 10.Rb1 Ra1
11.c5 Rxb1 12.Bxb1 Ke7 13.Kxb3 Rb8+
14.Kc2 (114.892.529) 481

19/25 4:23 +4.26 1...Ra8 2.Rh3 Na3+ 3.bxa3 Rxa3 4.Rf3 Qf8
5.Qxf7+ Qxf7 6.Rxf7 Kxf7 7.Kb2 Rca8
8.Bd3 Ra2+ 9.Kc3 Bc6 10.Rb1 Rb8
11.c5 b2 12.Kc2 Rb4 13.g5 hxg5 (126.913.961) 481

20/23 4:47 +4.44 1...Ra8 2.Rh3 Na3+ 3.bxa3 Rxa3 4.Rf3 Qf8
5.Qxf7+ Qxf7 6.Rxf7 Kxf7 7.Kb2 Rca8
8.Bd3 Ra2+ 9.Kc3 Bc6 10.Kxb3 Ba4+
11.Kxa2 Bxd1+ 12.Kb2 Bxg4 (138.481.301) 481

21/24 6:16 +4.16-- 1...Ra8 2.Rh3 Na3+ 3.bxa3 Rxa3 4.Kb2 Qf8
5.Rxb3 Rxb3+ 6.Kxb3 Rb8+ 7.Kc3 Ba4
8.Nc2 Qc5 9.Kd2 Qa5+ 10.Ke2 Bxc2
11.Rd2 Qc3 12.Kf2 Be4 13.Kg1 (182.606.107) 484

21/24 7:21 +3.59-- 1...Ra8 2.Rh3 Na3+ 3.bxa3 Rxa3 4.Kb2 Qf8
5.Rxb3 Rxb3+ 6.Kxb3 Rb8+ 7.Kc3 Ba4
8.Nc2 Qc5 9.Kd2 Qa5+ 10.Ke2 Bxc2
11.Rd2 Qc3 12.Kf2 Be4 13.Kg1 (214.700.967) 486

21/31 11:09 +3.91++ 1...Ra8 2.Rh3 Na3+ 3.bxa3 Rxa3 4.Kb2 Qf8
5.Rxb3 Rca8 6.Nc2 Rxb3+ 7.Kxb3 Ba4+
8.Kc3 Bxc2 9.Kxc2 Ra2+ 10.Kd3 Qa3+
11.Kd4 Qb2+ 12.Ke3 Qb3+ 13.Rd3 Qc2
14.cxd5 Qc1+ (328.545.177) 490

21/25 11:36 +4.20++ 1...Ra8 2.Rh3 Na3+ 3.bxa3 Rxa3 4.Kb2 Ra2+
5.Kb1 Rca8 6.Rxb3 Ra1+ 7.Kc2 Rxd1
8.Kxd1 Ba4 9.Kc2 Qf8 10.cxd5 exd5
11.Bc4 dxc4 12.Qxc4 Bxb3+ 13.Qxb3 Ra1 (342.905.689) 492

21/25 12:03 +4.41 1...Ra8 2.Rh3 Na3+ 3.bxa3 Rxa3 4.Kb2 Ra2+
5.Kb1 Rca8 6.Rxb3 Ra1+ 7.Kc2 Rxd1
8.Kxd1 Ba4 9.Kc2 Qf8 10.cxd5 Qa3
11.Bc4 Bxb3+ 12.Bxb3 Rc8+ 13.Kd1 Qxb3+ (356.355.527) 492

22/21 13:41 +4.43 1...Ra8 2.Rh3 Na3+ 3.bxa3 Rxa3 4.Rf3 Qf8
5.Bd3 Be8 6.Qd4 Rca8 7.Rd2 Qb4 8.Kc1 Ra1+
9.Bb1 R8a2 10.Qd3 Qa3+ 11.Kd1 Rxd2+ (402.810.762) 490

23/06 33:19 +4.31-- 1...Ra8 2.g5 Nc3+ 3.bxc3 Qf8 4.gxh6 (956.294.285) 478

23/06 38:37 +4.07-- 1...Ra8 2.g5 Nc3+ 3.bxc3 Qf8 4.gxh6 (1.106.418.142) 477

23/06 44:24 +3.59-- 1...Ra8 2.g5 Nc3+ 3.bxc3 Qf8 4.gxh6 (1.270.700.657) 476

23/08 46:36 +3.11++ 1...Ra8 2.g5 Nc3+ 3.bxc3 Qf8 4.gxh6 Ra1+
5.Kxa1 (1.333.669.911) 476

23/29 46:59 +3.23++ 1...Ra8 2.g5 Nc3+ 3.bxc3 Qf8 4.gxh6 Ra1+
5.Kxa1 Qa3+ 6.Kb1 Qa2+ 7.Kc1 b2+
8.Kd2 b1Q+ 9.Ke3 Qa7+ 10.Kd2 Qf5
11.Qxf5 exf5 12.Be2 Ba4 13.h7+ Kh8
14.cxd5 Bxd1 (1.345.323.980) 477

23/29 47:20 +3.47++ 1...Ra8 2.g5 Nc3+ 3.bxc3 Qf8 4.gxh6 Ra1+
5.Kxa1 Qa3+ 6.Kb1 Qa2+ 7.Kc1 b2+
8.Kd2 b1Q+ 9.Ke3 Qa7+ 10.Kd2 Qf5
11.Qxf5 exf5 12.Be2 Ba4 13.h7+ Kh8
14.cxd5 Bxd1 (1.355.533.740) 477

23/32 48:51 +3.65 1...Ra8 2.g5 Nc3+ 3.bxc3 Qf8 4.gxh6 Ra1+
5.Kxa1 Qa3+ 6.Kb1 Qa2+ 7.Kc1 b2+
8.Kd2 b1Q+ 9.Ke3 Qa7+ 10.Kd2 Qf5
11.Qg3+ Kf8 12.Ra1 Qaf2+ 13.Qxf2 Qxf2+
14.Kd1 Qb2 {[D]2r2k2/3b1p2/4p2P/3pP2P/2P5/2P5/1q6/R2KNB1R w - - 0 1} (1.398.005.452) 476

24/30 54:08 +4.08 1...Ra8 2.g5 Nc3+ 3.bxc3 Qf8 4.gxh6 Ra1+
5.Kxa1 Qa3+ 6.Kb1 Qa2+ 7.Kc1 b2+
8.Kd2 b1Q+ 9.Ke3 Qa7+ 10.Kd2 Qf5
11.Qxf5 exf5 12.Nf3 dxc4 13.Rg1+ Kh8
14.h7 Kxh7 {[D]2r5/q2b1p1k/8/4Pp1P/2p5/2P2N2/3K4/3R1BR1 w - - 0 1} (1.547.643.769) 476

25/27 73:37 +4.27 1...Ra8 2.g5 Nc3+ 3.bxc3 Qf8 4.gxh6 Ra1+
5.Kxa1 Qa3+ 6.Kb1 Qa2+ 7.Kc1 b2+
8.Kd2 b1Q+ 9.Ke3 Qa7+ 10.Rd4 Qc1+
11.Kf3 Qxf4+ 12.Rxf4 dxc4 13.Rg1+ Kh8
14.Rd4 Bc6+ {[D]2r4k/q4p2/2b1p2P/4P2P/2pR4/2P2K2/8/4NBR1 w - - 0 1} (2.087.824.083) 472

26/30 143:56 +4.42 1...Ra8 2.g5 Nc3+ 3.bxc3 Qf8 4.gxh6 Ra1+
5.Kxa1 Qa3+ 6.Kb1 Qa2+ 7.Kc1 b2+
8.Kd2 b1Q+ 9.Ke3 Qa7+ 10.Kd2 Qf5
11.Qxf5 exf5 12.Nf3 dxc4 13.Be2 Bc6
14.Nd4 Qa2+ {[D]2r3k1/5p2/2b4P/4Pp1P/2pN4/2P5/q2KB3/3R3R w - - 0 1} (4.068.011.706) 471

best move: Rb8-a8 time: 153:51.953 min n/s: 471.020 nodes: 4.348.422.440