Easy for humans, very hard for engines

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

Dann Corbit
Posts: 12615
Joined: Wed Mar 08, 2006 8:57 pm
Location: Redmond, WA USA

Easy for humans, very hard for engines

Post by Dann Corbit »

[d]7k/8/pp2p1p1/bN1p2pb/1p4p1/1P6/6P1/4K2R w - - bm g3; id "ACT4.051";
It is clear that by pushing the pawn to g3, both black bishops are in a bullet proof box.
Pushing the rook to f1 will keep the king in a 4 cell prison.
Anywhere on rank 7 would be as good, keeping the king on his baseline.
Meanwhile, the white king can clear the center pawns (possibly with rook help) and then execute a king/rook mate algorithm.
Of course, the optimal path will probably be much better than that simple plan.
I see that chess engines want to save the knight from the pawn, but that is instantly counter productive because it allows the bishop to escape.
I wonder why, in a case like this that is so easy for humans, chess engines don't see the prison cells.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
peter
Posts: 3206
Joined: Sat Feb 16, 2008 7:38 am
Full name: Peter Martan

Re: Easy for humans, very hard for engines

Post by peter »

BTW, it's 3 moves cut off of the study from J. Timman, that's like that in HHdb:

[pgn][Event "sp.c ARVES AT"]
[Site "?"]
[Date "2008.??.??"]
[Round "?"]
[White "Timman=J"]
[Black "(+0171.38e1h8) TT"]
[Result "1-0"]
[SetUp "1"]
[FEN "7k/6p1/pp2p3/b2p1ppB/1p4b1/NP4P1/6P1/4KR2 w - - 0 1"]
[PlyCount "9"]
[EventDate "2008.??.??"]

{(EG#17229).} 1. Nb5 $1 Bxh5 2. Rh1 g6 3. g4 $1 fxg4 4. g3 $1 Kg7 5. Rf1 $1 1-0
[/pgn]

Some output of Lc0 with 3070ti GPU and net 6147500:

[d]7k/8/pp2p1p1/bN1p2pb/1p4p1/1P6/6P1/4K2R w - - 0 1

Analysis by Lc0 v0.32.0-dag+git.4cfe7426:

4.g3 Kg7 5.Tf1 e5 6.Ke2 Kg8 7.Kd3 Kg7 8.Sc7 b5 9.Sxd5 Ld8 10.Ke4 a5 11.Kxe5 a4 12.Ke6 axb3 13.Tb1 La5 14.Txb3 Kf8 15.Tb1 b3
Tiefe: 14/41 00:01:13 391kN
Weiß steht klar auf Gewinn.
+- (11.59)

Started newly 4. (1.) g3 is found at once each time out of three ones tried.

Huntsman is on g3 at once too, that's again started with empty hash of 8G with 30 threads of 16x3.5GHz- CPU:

7k/8/pp2p1p1/bN1p2pb/1p4p1/1P6/6P1/4K2R w - - 0 1

Analysis by The Huntsman 1:

4.g3 Kg7 5.Tf1 Kg8 6.Ke2 Kg7 7.Kd3 e5 8.Tf2 Kg8 9.Te2 e4+ 10.Kd4 e3 11.Txe3 Kf7 12.Kxd5 Kf6 13.Kd6 Kf7 14.Te7+ Kg8 15.Ke5 Kf8 16.Kf6 axb5 17.Td7 Kg8 18.Td8+ Kh7 19.Tc8 Kh6 20.Th8#
Tiefe: 107/34 00:01:04 2959MN
Weiß setzt Matt.
+- (#17)
Peter.
Viz
Posts: 109
Joined: Tue Apr 09, 2024 6:24 am
Full name: Michael Chaly

Re: Easy for humans, very hard for engines

Post by Viz »

This is quite literally not chess.
Botvinnik was right - completely artificial compositions are just bad for your improvement as a chess player. We can see it with a good examples of engines - once that solve it easily completely suck in actual play compared to those who don't. Because latter are optimized towards actual chess game, and not some positions with double tripled pawns + blocked bishop setup.
Dann Corbit
Posts: 12615
Joined: Wed Mar 08, 2006 8:57 pm
Location: Redmond, WA USA

Re: Easy for humans, very hard for engines

Post by Dann Corbit »

LC0 is the second strongest engine.
Admittedly, Stockfish is the strongest engine and struggles mightily with this position.
I also think that compositions are important, because they are fun to solve.
Just because a position never happens in game play does not mean that it is not interesting.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
Uri Blass
Posts: 10424
Joined: Thu Mar 09, 2006 12:37 am
Location: Tel-Aviv Israel

Re: Easy for humans, very hard for engines

Post by Uri Blass »

Dann Corbit wrote: Sat May 25, 2024 10:33 am LC0 is the second strongest engine.
Admittedly, Stockfish is the strongest engine and struggles mightily with this position.
I also think that compositions are important, because they are fun to solve.
Just because a position never happens in game play does not mean that it is not interesting.
I am not sure about the claim that lc0 is the second strongest engine when she does not use the same hardware as stockfish in tests.
User avatar
towforce
Posts: 11751
Joined: Thu Mar 09, 2006 12:57 am
Location: Birmingham UK

Re: Easy for humans, very hard for engines

Post by towforce »

Dann Corbit wrote: Fri May 24, 2024 9:58 pm [d]7k/8/pp2p1p1/bN1p2pb/1p4p1/1P6/6P1/4K2R w - - bm g3; id "ACT4.051";
It is clear that by pushing the pawn to g3, both black bishops are in a bullet proof box.
Pushing the rook to f1 will keep the king in a 4 cell prison.
Anywhere on rank 7 would be as good, keeping the king on his baseline.
Meanwhile, the white king can clear the center pawns (possibly with rook help) and then execute a king/rook mate algorithm.
Of course, the optimal path will probably be much better than that simple plan.
I see that chess engines want to save the knight from the pawn, but that is instantly counter productive because it allows the bishop to escape.
I wonder why, in a case like this that is so easy for humans, chess engines don't see the prison cells.

This is an issue of "understanding" vs "generating good moves". Humans are better at the former, and engines are better at the latter. What you've shown here is a contrived position in which "understanding" is strongly rewarded.

The game value of understanding was demonstrated brilliantly by the team that worked out how to enable a good (as opposed to "very strong") Go player to beat an engine that was previously thought to be invincible to humans: they uncovered the fact that the engine did not understand a basic pattern in the game of Go - a group of stones!

Broken record time: yes - I'm going to say it yet again - today's big NNs are encoding a large number of shallow (simple) patterns, when a smaller number of deep (complex) patterns would be better. I think this is why it's proving to be so difficult to get computers to drive cars well - something that even many stupid humans can do.
The simple reveals itself after the complex has been exhausted.
chessica
Posts: 710
Joined: Thu Aug 11, 2022 11:30 pm
Full name: Esmeralda Pinto

Re: Easy for humans, very hard for engines

Post by chessica »

[d]7k/8/pp2p1p1/bN1p2pb/1p4p1/1P6/6P1/4K2R w - - 0 1

Blue Marlin 15.7:
Found 270 tablebases
classical evaluation enabled
...
33/11 00:00,757 2.378k 3.142k 0,00 Sb5-c7 b6-b5 Sc7xe6 g4-g3 Th1-f1 Lh5-g4 Tf1-f8+ Kh8-h7 Tf8-f7+ Kh7-h8
34/33 00:01,224 4.200k 3.431k +0,86 g2-g3 a6xb5 Th1-f1 Kh8-g7 Ke1-e2 e6-e5 Ke2-d3 Kg7-h7 Tf1-f6 Kh7-g7 Tf6-d6 Kg7-f7 Td6xd5 Kf7-e7 Kd3-e4 Ke7-e6 Td5-d2 Ke6-f6 Td2-d6+ Kf6-g7 Ke4xe5 Kg7-f7 Td6-d4 Kf7-e7
35/33 00:01,837 6.751k 3.675k +M35 g2-g3 e6-e5 Ke1-e2 Kh8-g7 Th1-f1 Kg7-g8 Ke2-e3 Kg8-g7 Ke3-d3 Kg7-h8 Tf1-f6 d5-d4 ...
74/38 00:06,861 26.400k 3.848k +M19 g2-g3 Kh8-g8 Th1-f1 Kg8-h8 Ke1-e2 Kh8-g7 Ke2-e3 e6-e5 Tf1-f2 a6xb5 Tf2-f1 e5-e4 Tf1-d1 Kg7-f7 Td1xd5 Kf7-f6 Ke3xe4 Kf6-e6 Td5-d4 Ke6-f6 Td4-d6+ Kf6-f7 Ke4-e5 Kf7-e7 Td6-d1 Ke7-f7 Td1-d7+ Kf7-e8 Ke5-e6 Ke8-f8 Ke6-f6 Kf8-g8 Td7-d8+ Kg8-h7 Td8-b8 Kh7-h6 Tb8-h8+
74/38 00:06,862 26.400k 3.847k +M18 g2-g3 Kh8-g8 Th1-f1 Kg8-h8 Ke1-e2 Kh8-g7 Ke2-d3 e6-e5 Kd3-e3 a6xb5 Ke3-d3 Kg7-g8 Tf1-e1 e5-e4+ Kd3-d4 Kg8-g7 Kd4xd5 Kg7-f7 Te1xe4 Kf7-f6 Te4-e6+ Kf6-g7 Te6-e7+ Kg7-h6 Te7-e8 Kh6-h7 Kd5-e6 Kh7-g7 Te8-d8 Kg7-h6 Ke6-f6 Kh6-h7 Td8-f8 Kh7-h6 Tf8-h8+
Bester Zug: g2-g3, Wert: +Matt in 18, Tiefe: 74/38, Dauer: 00:07,089, 26.400.486 Knoten, 3.847k
User avatar
towforce
Posts: 11751
Joined: Thu Mar 09, 2006 12:57 am
Location: Birmingham UK

Re: Easy for humans, very hard for engines

Post by towforce »

chessica wrote: Sat May 25, 2024 11:25 amBlue Marlin 15.7:
Found 270 tablebases
classical evaluation enabled
...
33/11 00:00,757 2.378k 3.142k 0,00 Sb5-c7 b6-b5 Sc7xe6 g4-g3 Th1-f1 Lh5-g4 Tf1-f8+ Kh8-h7 Tf8-f7+ Kh7-h8
34/33 00:01,224 4.200k 3.431k +0,86 g2-g3 a6xb5 Th1-f1 Kh8-g7 Ke1-e2 e6-e5 Ke2-d3 Kg7-h7 Tf1-f6 Kh7-g7 Tf6-d6 Kg7-f7 Td6xd5 Kf7-e7 Kd3-e4 Ke7-e6 Td5-d2 Ke6-f6 Td2-d6+ Kf6-g7 Ke4xe5 Kg7-f7 Td6-d4 Kf7-e7
35/33 00:01,837 6.751k 3.675k +M35 g2-g3 e6-e5 Ke1-e2 Kh8-g7 Th1-f1 Kg7-g8 Ke2-e3 Kg8-g7 Ke3-d3 Kg7-h8 Tf1-f6 d5-d4 ...
74/38 00:06,861 26.400k 3.848k +M19 g2-g3 Kh8-g8 Th1-f1 Kg8-h8 Ke1-e2 Kh8-g7 Ke2-e3 e6-e5 Tf1-f2 a6xb5 Tf2-f1 e5-e4 Tf1-d1 Kg7-f7 Td1xd5 Kf7-f6 Ke3xe4 Kf6-e6 Td5-d4 Ke6-f6 Td4-d6+ Kf6-f7 Ke4-e5 Kf7-e7 Td6-d1 Ke7-f7 Td1-d7+ Kf7-e8 Ke5-e6 Ke8-f8 Ke6-f6 Kf8-g8 Td7-d8+ Kg8-h7 Td8-b8 Kh7-h6 Tb8-h8+
74/38 00:06,862 26.400k 3.847k +M18 g2-g3 Kh8-g8 Th1-f1 Kg8-h8 Ke1-e2 Kh8-g7 Ke2-d3 e6-e5 Kd3-e3 a6xb5 Ke3-d3 Kg7-g8 Tf1-e1 e5-e4+ Kd3-d4 Kg8-g7 Kd4xd5 Kg7-f7 Te1xe4 Kf7-f6 Te4-e6+ Kf6-g7 Te6-e7+ Kg7-h6 Te7-e8 Kh6-h7 Kd5-e6 Kh7-g7 Te8-d8 Kg7-h6 Ke6-f6 Kh6-h7 Td8-f8 Kh7-h6 Tf8-h8+
Bester Zug: g2-g3, Wert: +Matt in 18, Tiefe: 74/38, Dauer: 00:07,089, 26.400.486 Knoten, 3.847k

My reading of this: Blue Marlin wanted to save the knight until it found a big advantage through search (and then a checkmate sequence): using the classical eval would have helped here, enabling a deeper search. Credit where it's due, though - it did look for the checkmate sequence in the right place. 8-)
The simple reveals itself after the complex has been exhausted.
peter
Posts: 3206
Joined: Sat Feb 16, 2008 7:38 am
Full name: Peter Martan

Re: Easy for humans, very hard for engines

Post by peter »

Dann Corbit wrote: Sat May 25, 2024 10:33 am Admittedly, Stockfish is the strongest engine and struggles mightily with this position.
It doesn't with less nullmove pruning. Sting, CorChess, ShashChess (with option High Tal checked), Huntsman solve that rather quickly, so does Dragon with Null Move Pruning unchecked, even in free version 1. The named SF- branches solve that from original position 3 moves earlier in reasonable hardware- time too, here's CorChess with 30 threads of 16x3.5GHz CPU and 32 G hash:


[d]7k/6p1/pp2p3/b2p1ppB/1p4b1/NP4P1/6P1/4KR2 w - - 0 1

Analysis by CorChess 4 dev-20240509-84f039ff:

Code: Select all

  
1.Sc4 Lxh5 2.Sxa5 bxa5 
  Schwarz steht deutlich besser.  
  -+ (-2.35)  Tiefe: 1/7   00:00:04  1221Knoten
...
1.Sc2 Lxh5 2.Th1 g6 3.g4 fxg4 4.Sd4 b5 5.Tf1 Lb6 6.Sxe6 g3 7.Tf8+ Kh7 8.Sxg5+ Kh6 9.Se6 Lf2+ 10.Kd2 Kh7 11.Tf7+ Kh6 12.Tf8 
  Die Stellung ist ausgeglichen.  
  = (0.00)  Tiefe: 50/23   00:00:55  503MN
1.Sb5 
  Die Stellung ist ausgeglichen.  
  = (0.03 ++)  Tiefe: 51/24   00:01:15  730MN
...
1.Sb5 
  Die Stellung ist ausgeglichen.  
  = (0.22 ++)  Tiefe: 51/28   00:01:16  749MN
1.Sb5 
  Weiß steht etwas besser.  
  +/= (0.31 ++)  Tiefe: 51/28   00:01:16  755MN
...
1.Sb5 
  Weiß hat entscheidenden Vorteil.  
  +- (3.38 ++)  Tiefe: 51/35   00:01:58  2086MN
1.Sb5 Lxh5 2.Th1 g6 3.g4 Kg7 4.gxh5 gxh5 5.Sc7 h4 6.Sxe6+ Kf6 7.Sc7 Ke5 8.Ke2 f4 9.Te1 d4 10.Kd3+ Kf5 11.Sd5 b5 12.Kxd4 Kg4 13.Te5 f3 14.gxf3+ Kxf3 15.Tf5+ Kg4 16.Ke4 h3 17.Se3+ Kg3 18.Txg5+ Kf2 19.Th5 Kg3 20.Th6 Lc7 21.Txa6 h2 22.Th6 Ld6 23.Sf5+ Kg2 24.Sxd6 
  Weiß steht auf Gewinn.  
  +- (3.62)  Tiefe: 51/68   00:02:43  2786MN
...
1.Sb5 Lxh5 2.Th1 g6 3.g4 Kg7 4.gxh5 gxh5 5.Sc7 h4 6.Sxe6+ Kf6 7.Sc7 Ke5 8.Ke2 f4 9.Te1 d4 10.Kd3+ Kf5 11.Sd5 b5 12.Kxd4 Kg4 13.Te5 f3 14.gxf3+ Kxf3 15.Tf5+ Kg4 16.Ke4 h3 17.Se3+ Kg3 18.Txg5+ Kf2 19.Th5 Kg3 20.Th6 Lc7 21.Txa6 h2 22.Th6 Ld6 23.Sf5+ Kg2 24.Sxd6 
  Weiß steht auf Gewinn.  
  +- (4.07 ++)  Tiefe: 52/25   00:05:46  6665MN
Started with empty hash again, this time of 8G at the test- position 3 moves later:

[d]7k/8/pp2p1p1/bN1p2pb/1p4p1/1P6/6P1/4K2R w - - 0 1

Analysis by CorChess 4 dev-20240509-84f039ff:

Code: Select all

4.Sc7 
  Die Stellung ist ausgeglichen.  
  = (0.03)  Tiefe: 1/3   00:00:01  1351Knoten
...
4.g3 
  Die Stellung ist ausgeglichen.  
  = (0.03 ++)  Tiefe: 45/27   00:00:10  92051kN
...
4.g3 
  Weiß steht klar auf Gewinn.  
  +- (4.45 ++)  Tiefe: 45/51   00:00:13  177MN
...
4.g3 
  Weiß steht klar auf Gewinn.  
  +- (59.48 ++)  Tiefe: 45/51   00:00:15  255MN
4.g3 
  Weiß steht klar auf Gewinn.  
  +- (79.30 ++)  Tiefe: 45/51   00:00:15  262MN
4.g3 axb5 5.Kd1 Kg8 6.Tf1 Kh8 7.Kd2 Kg8 8.Kd3 d4 9.Ke4 Kg7 10.Tc1 Kf7 11.Ke5 Ke7 12.Tc6 Kf7 13.Tc7+ Ke8 14.Tc8+ Kd7 
  Weiß setzt Matt.  
  +- (#27)  Tiefe: 45/58   00:00:15  286MN
...
4.g3 e5 5.Tf1 Kg7 6.Ke2 Kg8 7.Kd3 Kg7 8.Ke3 axb5 9.Kd3 Kh8 10.Te1 Kg7 11.Txe5 Kf6 12.Kd4 Kf7 13.Te2 Kg7 14.Kxd5 Kf6 15.Kd6 Kf7 16.Tc2 Kg8 17.Ke6 Kh7 18.Kf6 Kh6 19.Tc8 Kh7 20.Te8 Kh6 21.Th8# 
  Weiß setzt Matt.  
  +- (#18)  Tiefe: 49/36   00:00:20  414MN
4.g3 e5 5.Kd2 Kg7 6.Tf1 Kg8 7.Ke3 Kg7 8.Sc7 b5 9.Sxd5 Lb6+ 10.Sxb6 a5 11.Sd5 a4 12.Ke4 axb3 13.Kxe5 b2 14.Ke6 Kh6 15.Tb1 Kg7 16.Txb2 Kf8 17.Sf6 Kg7 18.Td2 Kf8 19.Td7 b3 20.Tf7# 
  Weiß setzt Matt.  
  +- (#17)  Tiefe: 50/36   00:00:23  458MN
...
4.g3 e5 5.Kd2 Kg7 6.Tf1 Kg8 7.Ke3 Kg7 8.Sc7 b5 9.Sxd5 Lb6+ 10.Sxb6 a5 11.Sd5 a4 12.Ke4 axb3 13.Kxe5 b2 14.Ke6 Kh6 15.Tb1 Kg7 16.Txb2 Kf8 17.Sf6 Kg7 18.Td2 Kf8 19.Td7 b3 20.Tf7# 
  Weiß setzt Matt.  
  +- (#17)  Tiefe: 56/34   00:00:35  686MN
Last edited by peter on Sat May 25, 2024 12:29 pm, edited 2 times in total.
Peter.
chessica
Posts: 710
Joined: Thu Aug 11, 2022 11:30 pm
Full name: Esmeralda Pinto

Re: Easy for humans, very hard for engines

Post by chessica »

[d]7k/8/pp2p1p1/bN1p2pb/1p4p1/1P6/6P1/4K2R w - - 0 1

John Stanback - McBrain-99-x64-w-static:
---
---------------------------------------------------------------------------
28/40 00:13,801 54.205k 3.928k -1,35 Sb5-d6 Kh8-g8 Th1-f1 g4-g3 Tf1-f7 Lh5-g4 Tf7-d7 e6-e5 Td7-e7 e5-e4 Te7-e5 Lg4-d7 Sd6-b7 g5-g4 Sb7xa5 b6xa5 Te5xd5 Ld7-b5 Td5-e5 a5-a4 b3xa4 Lb5-d3 Ke1-d2 Kg8-f7 Te5-g5 Kf7-f6 Tg5xg4 Kf6-f5 Tg4xg3 g6-g5 Tg3-e3 a6-a5 Te3-e1 Kf5-f4 Te1-a1
28/47 00:13,801 54.205k 3.928k -0,64 Th1-f1 g4-g3 Sb5-c7 Lh5-g4 Tf1-f6 Lg4-f5 Sc7xe6 Kh8-h7 Se6xg5+ Kh7-h6 Sg5-f3 Lf5-e4 Ke1-f1 b6-b5 Tf6xa6 La5-c7 Sf3-d2 Le4-d3+ Kf1-e1 Lc7-e5 Ta6-e6 Le5-f4 Te6-c6 Lf4-e3 Tc6-e6 Le3-f2+ Ke1-d1 d5-d4 Te6-e5 Lf2-e3 Te5-c5 g6-g5 Tc5-c6+ Kh6-h5 Tc6-c5 Kh5-g4 Tc5-d5 Kg4-f4
28/46 00:13,801 54.205k 3.928k 0,00 g2-g3 Kh8-g7 Th1-f1 a6xb5 Ke1-d2 e6-e5 Kd2-e2 e5-e4 Tf1-d1 Kg7-h6 Td1xd5 e4-e3 Ke2xe3 Kh6-h7 Td5-d4 Kh7-g7 Td4-e4 Kg7-f6 Te4-d4
28/11 00:13,801 54.205k 3.928k 0,00 Sb5-c7 b6-b5 Sc7xe6 g4-g3 Th1-f1 Lh5-g4 Tf1-f8+ Kh8-h7 Tf8-f7+ Kh7-h8
28/12 00:13,801 54.205k 3.928k 0,00 Sb5-d4 e6-e5 Sd4-e6 g4-g3 Th1-f1 e5-e4 Tf1-f8+ Kh8-h7 Tf8-f7+ Kh7-g8 Tf7-f8+
---------------------------------------------------------------------------
29/47 00:20,648 81.126k 3.929k -0,29 Th1-f1 g4-g3 Sb5-c7 Lh5-g4 Tf1-f6 Lg4-f5 Sc7xe6 Kh8-h7 Se6xg5+ Kh7-h6 Sg5-f3 Lf5-e4 Tf6-d6 Kh6-h5 Sf3-d2 Le4xg2 Sd2-f1 g6-g5 Sf1xg3+ Kh5-g4 Ke1-f2 Lg2-f3 Sg3-e2 Lf3-e4 Kf2-e3 b6-b5 Ke3-f2 La5-c7 Td6xa6 Lc7-d8 Se2-d4 Kg4-h5 Sd4xb5 g5-g4 Sb5-d4 Ld8-g5 Ta6-e6 Lg5-h4+ Kf2-g1 g4-g3 Te6-d6
29/49 00:20,648 81.126k 3.929k -0,09 Ke1-d2 g4-g3 Sb5-c7 Kh8-g8 Th1-f1 Lh5-g4 Tf1-f6 Kg8-h7 Sc7xe6 Kh7-h6 Se6-d4 Kh6-g7 Tf6-d6 Lg4-f5 Sd4xf5+ g6xf5 Td6xd5 Kg7-f6 Td5-d6+ Kf6-e5 Td6-g6 g5-g4 Kd2-d3 b6-b5 Tg6xa6 La5-d8 Ta6-g6 Ld8-f6 Kd3-e2 Lf6-e7 Ke2-d3 Le7-c5 Tg6-g5 Ke5-f6 Tg5-g8 Lc5-e7 Kd3-e2 Kf6-e5 Tg8-b8
29/22 00:20,648 81.126k 3.929k 0,00 g2-g3 a6xb5 Th1-f1 Kh8-g7 Ke1-d2 e6-e5 Kd2-e2 e5-e4 Tf1-d1 Kg7-h6 Td1xd5 e4-e3 Ke2xe3 Kh6-h7 Td5-d4 Kh7-g7 Td4-e4 Kg7-f6 Te4-d4
29/11 00:20,648 81.126k 3.929k 0,00 Sb5-c7 b6-b5 Sc7xe6 g4-g3 Th1-f1 Lh5-g4 Tf1-f8+ Kh8-h7 Tf8-f7+ Kh7-h8
29/12 00:20,648 81.126k 3.929k 0,00 Sb5-d4 e6-e5 Sd4-e6 g4-g3 Th1-f1 e5-e4 Tf1-f8+ Kh8-h7 Tf8-f7+ Kh7-g8 Tf7-f8+
---------------------------------------------------------------------------
...
---------------------------------------------------------------------------
40/59 05:45,341 1.140.744k 3.303k -0,48 Th1-f1 g4-g3 Sb5-c7 Lh5-g4 Tf1-f6 Lg4-f5 Sc7xe6 Kh8-h7 Se6xg5+ Kh7-h6 Sg5-f3 Lf5-c2 Sf3-d4 Kh6-g5 Tf6-f7 Lc2-e4 Sd4-e2 Le4xg2 Se2xg3 d5-d4 Tf7-d7 b6-b5 Sg3-e2 d4-d3 Td7xd3 La5-c7 Td3-d7 Lc7-b8 Se2-d4 Kg5-f4 Ke1-f2 Lg2-h1 Kf2-g1 Lh1-e4 Kg1-f2 g6-g5 Td7-f7+ Kf4-g4 Kf2-e3 Le4-b1 Tf7-f1 Lb8-f4+ Ke3-f2 Lb1-e4 Tf1-g1+ Kg4-h5 Sd4-e6 Le4-d5 Se6xf4+ g5xf4 Tg1-a1 Ld5xb3 Ta1xa6 Lb3-c4
40/26 05:45,341 1.140.744k 3.303k 0,00 Ke1-d2 g4-g3 Sb5-c7 Kh8-g8 Th1-f1 Lh5-g4 Tf1-f6 Kg8-h7 Sc7xe6 Lg4-f5 Se6xg5+ Kh7-g7 Tf6-c6 d5-d4 Kd2-e2 Kg7-h6 Sg5-f3 Lf5-e4 Tc6-d6 b6-b5 Td6xd4 Le4-a8 Td4-h4+ Kh6-g7 Th4-d4
40/11 05:45,341 1.140.744k 3.303k 0,00 Sb5-c7 b6-b5 Sc7xe6 g4-g3 Th1-f1 Lh5-g4 Tf1-f8+ Kh8-h7 Tf8-f7+ Kh7-h8
40/12 05:45,341 1.140.744k 3.303k 0,00 Sb5-d4 e6-e5 Sd4-e6 g4-g3 Th1-f1 e5-e4 Tf1-f8+ Kh8-h7 Tf8-f7+ Kh7-g8 Tf7-f8+
40/61 05:45,341 1.140.744k 3.303k +153,70 g2-g3 Kh8-g7 Th1-f1 e6-e5 Ke1-d2 Kg7-g8 Kd2-e3 Kg8-h8 Sb5-c7 d5-d4+ Ke3-d2 b6-b5 Sc7-e6 La5-b6 Tf1-f8+ Kh8-h7 Tf8-f7+ Kh7-g8 Tf7-g7+ Kg8-h8 Tg7-b7 a6-a5 Tb7xb6 a5-a4 b3xa4 b5xa4 Tb6xb4 Kh8-g8 Tb4xa4 Kg8-f7 Se6xg5+ Kf7-f6 Sg5-e4+ Kf6-g7 Ta4-a7+ Kg7-h6 Ta7-a6 Kh6-g7 Se4-f6 g6-g5 Sf6xh5+ Kg7-f7 Kd2-d3 Kf7-e7 Sh5-f6 e5-e4+ Sf6xe4 Ke7-f7 Ta6-g6 Kf7-f8 Kd3xd4 Kf8-e7 Tg6xg5 Ke7-e6 Tg5xg4 Ke6-d7
---------------------------------------------------------------------------
41/67 07:18,248 1.459.167k 3.330k -0,53 Th1-f1 g4-g3 Sb5-c7 Lh5-g4 Tf1-f6 Lg4-f5 Sc7xe6 Kh8-h7 Se6xg5+ Kh7-h6 Sg5-f3 Lf5-c2 Sf3-d4 Lc2-e4 Sd4-f3 Kh6-h5 Ke1-e2 b6-b5 Tf6xa6 La5-c7 Ta6-c6 Lc7-f4 Tc6-b6 Kh5-g4 Sf3-e1 Lf4-e5 Tb6xb5 Le5-c3 Se1-f3 d5-d4 Sf3-e5+ Kg4-g5 Se5-f3+ Kg5-f4 Sf3-e5 Le4xg2 Se5xg6+ Kf4-g4 Sg6-e5+ Kg4-h3 Se5-f3 d4-d3+ Ke2-e3 Lc3-f6 Tb5-d5 Kh3-g4 Sf3-e5+ Lf6xe5 Td5xe5 Lg2-f3 Ke3xd3 g3-g2 Te5-e1 Lf3-d5
41/66 07:18,248 1.459.167k 3.330k 0,00 Ke1-d2 g4-g3 Sb5-c7 b6-b5 Sc7xe6 La5-b6 Th1-c1 Lh5-g4 Tc1-c6 Lb6-f2 Tc6xa6 Kh8-h7 Se6xg5+ Kh7-h6 Sg5-f7+ Kh6-h5 Ta6-d6 d5-d4 Sf7-e5 Lf2-e3+ Kd2-d3 Lg4-f5+ Kd3-e2 Lf5-c2 Td6-b6 Lc2-e4 Se5-f3 Kh5-g4 Sf3-e1 g6-g5 Tb6xb5 Le3-f4 Tb5xb4 d4-d3+ Se1xd3 Le4xg2 Tb4-a4 Lg2-d5 Ke2-f1 Kg4-f3 Sd3-e1+ Kf3-e3 Se1-g2+ Ke3-f3
41/11 07:18,248 1.459.167k 3.330k 0,00 Sb5-c7 b6-b5 Sc7xe6 g4-g3 Th1-f1 Lh5-g4 Tf1-f8+ Kh8-h7 Tf8-f7+ Kh7-h8
41/14 07:18,248 1.459.167k 3.330k 0,00 Sb5-d4 e6-e5 Sd4-e6 g4-g3 Th1-f1 Lh5-g4 Tf1-f8+ Kh8-h7 Tf8-f7+ Kh7-g8 Tf7-f8+
41/68 07:18,248 1.459.167k 3.330k +153,70 g2-g3 Kh8-g7 Th1-f1 e6-e5 Ke1-d2 Kg7-g8 Kd2-e3 Kg8-h8 Sb5-c7 d5-d4+ Ke3-d2 b6-b5 Sc7-e6 La5-b6 Tf1-f8+ Kh8-h7 Tf8-f7+ Kh7-g8 Tf7-g7+ Kg8-h8 Tg7-b7 a6-a5 Tb7xb6 a5-a4 b3xa4 b5xa4 Tb6xb4 Kh8-g8 Tb4xa4 Kg8-f7 Se6xg5+ Kf7-f6 Sg5-e4+ Kf6-g7 Ta4-a7+ Kg7-h6 Ta7-a6 Kh6-g7 Se4-f6 g6-g5 Sf6xh5+ Kg7-f7 Kd2-d3 Kf7-e7 Sh5-f6 e5-e4+ Sf6xe4 Ke7-f7 Ta6-g6 Kf7-f8 Kd3xd4 Kf8-e7 Tg6xg5 Ke7-e6 Tg5xg4 Ke6-d7 Tg4-g8 Kd7-c7 g3-g4 Kc7-c6
---------------------------------------------------------------------------
42/70 08:02,663 1.613.990k 3.344k -0,66 Th1-f1 g4-g3 Sb5-c7 Lh5-g4 Tf1-f6 Kh8-g7 Sc7-e8+ Kg7-g8 Se8-c7 Lg4-f5 Sc7xe6 Kg8-h7 Se6xg5+ Kh7-h6 Sg5-f3 Lf5-c2 Sf3-d4 Lc2-e4 Sd4-e2 Kh6-g5 Tf6-f7 d5-d4 Tf7-d7 b6-b5 Ke1-d2 La5-b6 Se2xg3 Le4xg2 Kd2-d3 a6-a5 Sg3-e4+ Kg5-f4 Td7-f7+ Kf4-e5 Tf7-e7+ Ke5-d5 Te7-d7+ Kd5-c6 Td7-d6+ Kc6-c7 Td6xg6 Lg2xe4+ Kd3xe4 a5-a4 Tg6-g7+ Kc7-d6 Tg7-g6+ Kd6-c5 Tg6-g5+ Kc5-c6 Tg5-g6+ Kc6-b7 Ke4-d3 a4-a3 Tg6-g7+ Kb7-a6 Tg7-g2 Ka6-a5 Kd3-e4 Lb6-c5 Ke4-d3 Ka5-b6 Tg2-g6+ Kb6-b7 Tg6-g2 Kb7-c6
42/54 08:02,663 1.613.990k 3.344k 0,00 Ke1-d2 g4-g3 Sb5-c7 b6-b5 Sc7xe6 La5-b6 Th1-c1 Lh5-g4 Tc1-c6 Lb6-f2 Tc6xa6 Kh8-h7 Se6xg5+ Kh7-h6 Sg5-f7+ Kh6-h5 Ta6-d6 d5-d4 Sf7-e5 Lf2-e3+ Kd2-d3 Lg4-f5+ Kd3-e2 Lf5-c2 Td6-b6 Lc2-e4 Se5-f3 Kh5-g4 Sf3-e1 g6-g5 Tb6xb5 Le3-f4 Tb5xb4 d4-d3+ Se1xd3 Le4xg2 Tb4-a4 Lg2-d5 Ke2-f1 Kg4-f3 Sd3-e1+ Kf3-e3 Se1-g2+ Ke3-f3
42/11 08:02,663 1.613.990k 3.344k 0,00 Sb5-c7 b6-b5 Sc7xe6 g4-g3 Th1-f1 Lh5-g4 Tf1-f8+ Kh8-h7 Tf8-f7+ Kh7-h8
42/12 08:02,663 1.613.990k 3.344k 0,00 Sb5-d4 e6-e5 Sd4-e6 g4-g3 Th1-f1 Lh5-g4 Tf1-f8+ Kh8-h7 Tf8-f7+ Kh7-g8 Tf7-f8+
42/58 08:02,663 1.613.990k 3.344k +M29 g2-g3 Kh8-g7 Th1-f1 e6-e5 Ke1-d2 Kg7-h8 Sb5-c7 d5-d4 Sc7-b5 e5-e4 Sb5xd4 b6-b5 Tf1-f8+ Kh8-h7 Tf8-f7+ Kh7-g8 Tf7-b7 La5-d8 Tb7-b8 Kg8-f7 Tb8xd8 a6-a5 Sd4xb5 Kf7-e7 Td8-a8 a5-a4 Ta8xa4 Ke7-d7 Kd2-e3 Kd7-c6 Sb5-d4+ Kc6-b6 Sd4-c2 Kb6-b7 Ke3xe4 Kb7-b6 Sc2xb4 Kb6-c7 Sb4-d5+ Kc7-c6 Ke4-d4 Kc6-b5 Ta4-b4+ Kb5-a6 Tb4-b6+ Ka6-a7 Kd4-c4 Ka7-a8 Tb6-b5 Ka8-a7 Sd5-b4 Ka7-a8 Kc4-c5 Ka8-a7 Sb4-c6+ Ka7-a6 Tb5-b6+
---------------------------------------------------------------------------