Mortimer trap test bug in Stockfish 10

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

Moderators: hgm, Dann Corbit, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
ailin
Posts: 13
Joined: Sun Sep 24, 2017 11:25 am
Location: Iran

Re: Mortimer trap test bug in Stockfish 10

Post by ailin » Fri Jun 07, 2019 6:56 pm

[/quote]
This is not a bug, it's just an observation that Stockfish doesn't play perfect chess.
[/quote]

elapa,
This is an awful fact! Why a high end software like stockfish 10 , on high end hardwares with 12 cores can’t choose a move correctly while in 1880 a normal chess player has found it?! 6.Nc4 !?

jp
Posts: 1438
Joined: Mon Apr 23, 2018 5:54 am

Re: Mortimer trap test bug in Stockfish 10

Post by jp » Fri Jun 07, 2019 6:59 pm

ailin wrote:
Fri Jun 07, 2019 6:56 pm
This is not a bug, it's just an observation that Stockfish doesn't play perfect chess.
elapa,
This is an awful fact! Why a high end software like stockfish 10 , on high end hardwares with 12 cores can’t choose a move correctly while in 1880 a normal chess player has found it?! 6.Nc4 !?
What makes you believe Nc4 is so good? Both moves leave White totally lost.

If you needed convincing of that, isn't MikeB's game enough?

ailin
Posts: 13
Joined: Sun Sep 24, 2017 11:25 am
Location: Iran

Re: Mortimer trap test bug in Stockfish 10

Post by ailin » Fri Jun 07, 2019 7:05 pm

Jp,
The eval can change , ok . But from what I observed after 10 moves (with best play for both sides according to Stockfish)
(and by depths like 30 plies ) , Stockfish concludes that his initial choice (6.Nxf7) was wrong ! Now come back to the sixth move and Stockfish still chooses it ! Strange! Something is wrong here.

jp
Posts: 1438
Joined: Mon Apr 23, 2018 5:54 am

Re: Mortimer trap test bug in Stockfish 10

Post by jp » Fri Jun 07, 2019 7:08 pm

ailin wrote:
Fri Jun 07, 2019 7:05 pm
Jp,
The eval can change , ok . But from what I observed after 10 moves (with best play for both sides according to Stockfish)
(and by depths like 30 plies ) , Stockfish concludes that his initial choice (6.Nxf7) was wrong ! Now come back to the sixth move and Stockfish still chooses it ! Strange! Something is wrong here.
It only makes sense to call one move right and another wrong if they are not both completely lost. Do you agree that after both moves White is completely lost?

You appear to have a strange belief that 6.Nc4 is "good", when it leaves White completely lost.

ailin
Posts: 13
Joined: Sun Sep 24, 2017 11:25 am
Location: Iran

Re: Mortimer trap test bug in Stockfish 10

Post by ailin » Fri Jun 07, 2019 7:14 pm

Jp
You missed the point again .
It’s simple. I dont tell 6.Nc4 is so good or winning move. I’m telling it’s better choice than 6.Nxf7. and Stockfish can’t realize this .
Mike’s evaluation is after black’s move (6...Ng6). He didn’t test “best move” for sixth white move .

jp
Posts: 1438
Joined: Mon Apr 23, 2018 5:54 am

Re: Mortimer trap test bug in Stockfish 10

Post by jp » Fri Jun 07, 2019 7:20 pm

ailin wrote:
Fri Jun 07, 2019 7:14 pm
I dont tell 6.Nc4 is so good or winning move. I’m telling it’s better choice than 6.Nxf7. and Stockfish can’t realize this .
Mike’s evaluation is after black’s move (6...Ng6). He didn’t test “best move” for sixth white move .
What makes you sure it's "better"? Do you really think there's great significance in disagreement about which of two losing moves is "better" at some fixed depth of the engine's search?

You appear to have rigid ideas, without objective support, about what are the "best" moves. MikeB played an engine-engine game with the line including 6.Nc4 forced, and Black totally unsurprisingly won.

There is nothing for SF to realize. At different depths, it may order two moves differently.

zullil
Posts: 6442
Joined: Mon Jan 08, 2007 11:31 pm
Location: PA USA
Full name: Louis Zulli

Re: Mortimer trap test bug in Stockfish 10

Post by zullil » Fri Jun 07, 2019 8:34 pm

jp wrote:
Fri Jun 07, 2019 7:20 pm
ailin wrote:
Fri Jun 07, 2019 7:14 pm
I dont tell 6.Nc4 is so good or winning move. I’m telling it’s better choice than 6.Nxf7. and Stockfish can’t realize this .
Mike’s evaluation is after black’s move (6...Ng6). He didn’t test “best move” for sixth white move .
What makes you sure it's "better"? Do you really think there's great significance in disagreement about which of two losing moves is "better" at some fixed depth of the engine's search?

You appear to have rigid ideas, without objective support, about what are the "best" moves. MikeB played an engine-engine game with the line including 6.Nc4 forced, and Black totally unsurprisingly won.

There is nothing for SF to realize. At different depths, it may order two moves differently.
Right. All sixth moves for White lose. Cfish-dev would choose 6. Nxf7. It currently considers 6. Nc4 to be third best.



-1.56 6. Nxf7 Kxf7 7. Ba4 d5 8. Nc3 dxe4 9. Bb3+ Kg6 10. Qd2 h6 11. dxe4 Qxd2+ 12. Bxd2 b5 13. a3 Kh7 14. f4 a5 15. h3 Ng6 16. e5 Nd7 17. Bf7 Nc5 18. g4 Be7 19. O-O Rd8 20. Bxg6+ Kxg6 21. f5+ Kh7 22. Be3 Nd7 23. e6 Nf6 24. Rfd1 Rxd1+ 25. Rxd1 Ba6 26. Bd4 b4 27. axb4 axb4 (depth 42, 0:45:07)

-2.37 6. Nxc6 dxc6 7. Bc4 Ng6 8. O-O Bc5 9. c3 Ne5 10. d4 Nxc4 11. dxc5 Qe7 12. b3 Ne5 13. f3 Qxc5+ 14. Qd4 Qxd4+ 15. cxd4 Ng6 16. Be3 O-O 17. Nc3 h6 18. Rac1 Rd8 19. Rfd1 Be6 20. Kf2 b6 21. g4 Kh7 22. Kg3 b5 23. Ne2 Ne7 24. g5 hxg5 25. Bxg5 a5 26. Nf4 a4 27. Nxe6 fxe6 (depth 41, 0:45:07)

-2.55 6. Nc4 Ng6 7. e5 Nd5 8. Nd6+ Bxd6 9. exd6 cxb5 10. Qf3 Nf6 11. Qe3+ Kf8 12. O-O b6 13. Qd4 Bb7 14. Nc3 h5 15. Bg5 Qc8 16. a4 bxa4 17. Bxf6 gxf6 18. Qxf6 Qd8 19. Qxd8+ Rxd8 20. Rxa4 a5 21. Rc4 Re8 22. f3 Rg8 23. Kf2 Bc6 24. Rb1 h4 25. b4 h3 26. g3 Ne5 27. Rf4 Rg6 (depth 41, 0:45:07)
[+] 6. Nxf7 (suggested move)

ailin
Posts: 13
Joined: Sun Sep 24, 2017 11:25 am
Location: Iran

Re: Mortimer trap test bug in Stockfish 10

Post by ailin » Fri Jun 07, 2019 9:34 pm

Thanks Zullil. Nothing remains to say.
What a strange position.
What’s your system hardware?

Dann Corbit
Posts: 11929
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: Mortimer trap test bug in Stockfish 10

Post by Dann Corbit » Fri Jun 07, 2019 10:34 pm

Stockfish explains Mortimer's trap for us... (First it is important to know that Mortimer means 'dead sea' and possibly is derived from a stagnant pond place name):
https://www.ancestry.com/name-origin?surname=mortimer

In the rows that follow, comment field c0 holds the data for the move necessary for Mortimer's trap, as laid out here:
https://en.wikipedia.org/wiki/Ruy_Lopez,_Mortimer_Trap


For the first few moves, computer analysis shows that they only deviate by a few hundredths of a pawn from optimal:

Code: Select all

rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - acd 70; acs 1589745; bm c4; c0 "e4"; c1 "01a"; ce 8; pm e4 {2379495} d4 {1725526} Nf3 {483679} c4 {479713} g3 {39820} f4 {35589} b3 {27764} Nc3 {13010} e3 {2686} b4 {2458} g4 {2450} c3 {1654} d3 {1649} a3 {1150} h3 {717} a4 {552} h4 {425} f3 {363} Nh3 {82} Na3 {28}; pv c4 c5 Nf3 Nf6 g3 d5 d4 dxc4 Qa4 Nc6 dxc5 e6 Bg2 Bxc5 Nc3 Qa5 Qxc4 O-O O-O Bd7 Ne4 Be7 Bd2 Qf5 Nxf6 Bxf6 Bc3 e5 Nd2 Rac8 Qb3 Rc7 Bd5 b5 a3 a6 Rac1 Be7 e3 Rfc8 f4 Be6 Bxe6 Qxe6 Qxe6 fxe6 Ne4 a5 Bxe5 Nxe5 Rxc7 Rxc7 fxe5 Rc4 Rf4 Bc5 Nxc5 Rxc5 Kf1 Rc1 Kg2 Rc5 b4 axb4 axb4 Rc2 Rf2 Rc3 Kf3 Rc4 Rd2 Rxb4 Rd7 g5 g4 Rb2 Rb7 Rxh2 Rxb5; 
rnbqkbnr/pppppppp/8/8/4P3/8/PPPP1PPP/RNBQKBNR b KQkq - acd 48; acs 212911; bm e6; c0 "e5"; c1 "01b"; ce -16; id "es-dc-neutral.0005"; pm c5 {1218905} e5 {412609} e6 {257916} c6 {204175} d5 {78062} d6 {71821} g6 {57457} Nf6 {46929} Nc6 {20542} b6 {4882} a6 {1853} g5 {987} h6 {878} a5 {501} h5 {447} Na6 {410} f6 {336} f5 {276} Nh6 {273} b5 {241}; pv e6 d4 d5 Nd2 c5 Ngf3 Nf6 exd5 exd5 Bb5+ Bd7 Bxd7+ Nbxd7 O-O Be7 dxc5 Nxc5 Nb3 Nce4 Nfd4 O-O Nf5 Re8 Nxe7+ Rxe7 Nd4 Rc8 Re1 Nd6 Rxe7 Qxe7 Bf4 a6 Bxd6 Qxd6 c3 g6 g3 Ne4 Kg2 Qb6 Qb3 Qxb3; 
rnbqkbnr/pppp1ppp/8/4p3/4P3/8/PPPP1PPP/RNBQKBNR w KQkq - acd 45; acs 2500; bm Nf3; c0 "Nf3"; c1 "02a"; ce 25; id "es-dc-neutral.0003"; pm Nf3 {353642} Bc4 {17694} Nc3 {17350} f4 {14760} d4 {3628} b3 {3002} g3 {509} a3 {392} d3 {250} Ne2 {224} Bb5 {222} c4 {214} c3 {199} h3 {93} Be2 {87} Qh5 {75} f3 {71} Qf3 {49} a4 {46} Bd3 {30} b4 {18} h4 {16} g4 {15} Nh3 {13} Qe2 {9} Na3 {3} Ba6 {1} Ke2 {1} Qg4 {1}; pv Nf3 Nc6 Bc4 Nf6 d3 Bc5 c3 O-O O-O d6 Re1 Ng4 Re2 Bb6 a4 h6 h3 Nf6 Nbd2 a5 Bb3 Re8 Nc4 Ba7 Na3 Bd7 Bc4 Bb6 Qb3 Rf8 Be3 Bxe3 Rxe3 Nh5 d4 Nf4 Bb5 exd4 cxd4 Nb4 Bxd7 Qxd7 Nc4 Rfe8 Rae1 c6 e5; 
rnbqkbnr/pppp1ppp/8/4p3/4P3/5N2/PPPP1PPP/RNBQKB1R b KQkq - acd 40; acs 3572; bm Nc6; c0 "Nc6"; c1 "02b"; ce -22; pm Nc6 {305788} Nf6 {35603} d6 {7699} f5 {2080} d5 {1387} Qe7 {589} Qf6 {250} Bd6 {55} Bc5 {38} f6 {35} a6 {29} c6 {25} b6 {18} Be7 {16} g6 {14} Bb4 {10} a5 {8} c5 {8} Ne7 {3} h5 {3} h6 {3} Ke7 {1} Na6 {1} Nh6 {1} Qh4 {1} b5 {1}; pv Nc6 Bc4 Nf6 d3 Bc5 O-O O-O c3 d6 Bg5 h6 Bh4 g5 Bg3 a5 Nbd2 Rb8 Bb3 Kg7 Nc4 b5 Ne3 Bb6 Bc2 Re8 a4 b4 h3 Bd7 Re1 Ba7 Bb3 Nh5 Nxe5 Nxg3 Nxd7 Qxd7 fxg3 Bxe3+ Rxe3 f5 Re1 bxc3 bxc3 fxe4; 
r1bqkbnr/pppp1ppp/2n5/4p3/4P3/5N2/PPPP1PPP/RNBQKB1R w KQkq - acd 39; acs 1919; bm Bb5; c0 "Bb5"; c1 "03a"; ce 16; pm Bb5 {211404} Bc4 {44650} d4 {31948} Nc3 {20016} c3 {3337} Be2 {286} a3 {198} b3 {190} c4 {178} d3 {177} h3 {112} g3 {58} b4 {7} Bd3 {6} Nd4 {6} Ng1 {3} Nxe5 {3} Qe2 {3} Ng5 {1} g4 {1}; pv Bb5 Nf6; 
r1bqkbnr/pppp1ppp/2n5/1B2p3/4P3/5N2/PPPP1PPP/RNBQK2R b KQkq - acd 39; acs 2630; bm a6; c0 "Nf6"; c1 "03b"; ce -23; pm a6 {157048} Nf6 {33689} Bc5 {6360} f5 {4324} d6 {3355} g6 {2496} Nge7 {1710} Nd4 {1698} Bb4 {169} Bd6 {165} Be7 {121} Qf6 {74} a5 {63} g5 {54} Qe7 {43} f6 {17} d5 {12} Nce7 {8} h6 {7} b6 {5} Na5 {3} h5 {2} Nb8 {1}; pv a6 Ba4 b5 Bb3 Bb7 O-O Nf6 d3 Bc5 Nc3 d6 a3 O-O Nd5 Nxd5 Bxd5 Qc8 Bb3 a5 a4 b4 c3 bxc3 bxc3 Ne7 Re1 Rb8 d4 exd4 cxd4 Bb4 Bd2 Ba6 Rc1 c6 Ba2 Qc7 Qc2 Bxd2 Qxd2 Rb4 e5 Rfb8 exd6 Qxd6 Rc5 Rxa4; 
here, d3 isn't as popular as short castling, but you are likely to see it in real games and it may be best:
acd 41; acs 1731; bm d3; c0 "d3"; c1 "04a"; ce 50; pm O-O {23103} d3 {5170} Nc3 {3302} Qe2 {1151} d4 {641} Bxc6 {312} Ba4 {7} Bd3 {2} c3 {2} Rf1 {1} h3 {1}; pv d3 Bc5 Bxc6 dxc6 Nbd2 O-O O-O Nd7 Nb3 Qe7 Nxc5 Nxc5 b3 a5 a4 Bg4 h3 Bxf3 Qxf3 Ne6 Bb2 Rfd8 Qg4 Qf8 Bxe5 Qc5 Bf6 Qxc2 Rac1 Qxd3 Bxd8 Nxd8;

This is an awful move, but it is intentional, it is the "trap" move in Mortimer's trap. Bc5 and d6 are clearly better moves, but we are hoping for a blunder
acd 43; acs 1733; bm Bc5; c0 "Ne7?!"; c1 "04b"; ce 0; id "es-dc-neutral.0785"; pm d6 {2846} Bc5 {2169} Ne7 {82} Bd6 {61} a6 {16} Be7 {5} Bb4+ {4} d5 {4} Nd4 {3} Ng4 {1} Qe7 {1} h6 {1}; pv Bc5 Bxc6 dxc6 O-O Nd7 c3 a5 d4 Bd6 Re1 O-O Bg5 f6 Bh4 Qe7 Bg3 a4 Nbd2 b5 Qe2 Re8 h3 Qf7 a3 Ba6 Qe3 Bb7 dxe5 fxe5 Qd3 h6 Red1 Nc5 Qe2 Rad8 Qe3 Nd7 Qe2 Ba6 Nh4 Bc5 Nf5 Bd6;

This move (Nxe5) is the first blunder for white. He goes from a small advantage to a pawn and a half behind by making it:
acd 40; acs 974; bm Bc4; c0 "Nxe5??"; c1 "05a"; ce 17; id "C.A.P. 443186"; pm O-O {39} Nc3 {17} Bc4 {12} d4 {7} Ba4 {2} Nxe5 {2} c3 {2}; pv Bc4 c6 Nc3 d6 O-O h6 d4 Qc7 a4 Ng6 h3 Be7 Qd3 Bd7 Be3 O-O Rfe1 Qc8 Qd1 Qc7 Nd2 exd4 Bxd4 Rad8 Nf1 Be6 Ne3 Kh7 Bb3 Nd7 Ncd5 cxd5 exd5 Nc5 Bc4 Bf6 Bxf6 gxf6 dxe6 fxe6 Bb3 Qg7;

Here c6 is best here (played move agrees with computer analysis):
acd 40; acs 472; bm c6; c0 "c6"; c1 "05b"; ce 146; id "WCSAC.269"; pm c6; pv c6 Nxf7 Kxf7 Ba4 d5 Bb3 g6 Nc3 Bg7 d4 Be6 f3 Rf8 O-O Qd7 e5 Ne8 Ne2 Kg8 c3 Nc7 Bg5 c5 Qd2 c4 Bd1 Bf5 Nf4 h6 Bxe7 Qxe7 g4 Bb1 Rxb1 Qg5 Qe1 Qxf4 b3 cxb3 Bxb3 b6 Qe4 Kh7 Qxf4 Rxf4 Kg2 Rd8 h4 Rff8 Bc2 Ne6;

Here Nc4 is part of the suggested sequence but it is in fact a blunder giving away a pawn in value. White is still in trouble, but not as bad as after Nc4??:
acd 40; acs 554; bm Nxf7; c0 "Nc4??"; c1 "06a"; ce -157; id "C.A.P. 442993"; pm Nxf7; pv Nxf7 Kxf7 Ba4 d5 Bb3 g6 Nc3 Bg7 d4 Be6 e5 Ne8 Bg5 Nc7 Ne2 h6 Be3 Nf5 Bf4 Kg8 c3 Kh7 Bc1 Rf8 Nf4 Bf7 h4 Ne6 h5 Nxf4 Bxf4 g5 Bc2 Be6 Bh2 Kh8 O-O Qb6 Qd3 c5 dxc5 Qxc5 a3;

Ng6 would kill him a lot quicker than the suggested d6 (though d6 also wins):
acd 40; acs 1479; bm Ng6; c0 "d6"; c1 "06b"; ce 236; pm Ng6; pv Ng6 e5 Nd5 Qf3 cxb5 Nd6+ Bxd6 exd6 Nf6 h4 O-O h5 Re8+ Kf1 Ne5 Qg3 Qb6 Bg5 Neg4 h6 Re6 Nc3 Qxd6 Qxd6 Rxd6 hxg7 b4 Ne2 Nd5 Nd4 Rg6 Bd2 d6 a3 bxa3 Rxa3 Bd7 c4 Nc7 Rh4 Ne5 Be3 Rxg7 b4 Rg4 Rxg4+ Nxg4 Bd2 Ne5 f3 b5 f4 Nc6 Nxc6 Bxc6;

Every other move is correct from here:

Code: Select all

r1bqkb1r/pp2nppp/2pp1n2/1B6/2N1P3/3P4/PPP2PPP/RNBQK2R w KQkq - acd 36; acs 263; bm Ba4; c0 "Ba4"; c1 "07a"; ce -183; pm Ba4; pv Ba4 b5 Nxd6+ Qxd6 Bb3 Ng6 a4 Be7 axb5 cxb5 Nc3 b4 Nd5 O-O O-O Be6 c4 bxc3 Nxe7+ Qxe7 bxc3 Rfd8 Bxe6 fxe6 Qe2 Qc5 c4 a5 Rd1 a4 Be3 Qe5 f3 Nf4 Bxf4 Qxf4 d4 Nh5 g3 Qg5 c5 a3 c6 Qe7 Kh1 Qf7 f4 Nf6 Qc2 Qc7 Qc4; 
r1bqkb1r/pp2nppp/2pp1n2/8/B1N1P3/3P4/PPP2PPP/RNBQK2R b KQkq - acd 36; acs 246; bm b5; c0 "b5"; c1 "07b"; ce 207; pm b5; pv b5 Nxd6+ Qxd6 Bb3 Ng6 h3 Qd7 O-O Bc5 a4 b4 Nd2 O-O Nc4 Rd8 Qe2 Qc7 Be3 Bxe3 Qxe3 Nh5 e5 Nhf4 Nd6 Be6 Bxe6 Nxe6 d4 c5 c3 Ngf4 Rad1 bxc3 bxc3 cxd4 cxd4 Nd5 Qd2 Rab8 Rc1 Qb6 Rc4 Rxd6 exd6 Qxd6; 
r1bqkb1r/p3nppp/2pp1n2/1p6/B1N1P3/3P4/PPP2PPP/RNBQK2R w KQkq - acd 36; acs 449; bm Nxd6+; c1 "08a"; ce -164; pm Nxd6+; pv Nxd6+ Qxd6 Bb3 Ng6 a4 Be7 h3 Be6 Bxe6 fxe6 O-O e5 Be3 a6 Nd2 O-O g3 Qe6 Kg2 h6 Nb3 Bd6 Qe2 Qf7 h4 Ne7 axb5 axb5 Nc5 Nf5 exf5 Bxc5 Bxc5 Qd5+ f3 Qxc5 c3 Kh8 Rfe1 Rxa1 Rxa1; 
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.

ailin
Posts: 13
Joined: Sun Sep 24, 2017 11:25 am
Location: Iran

Re: Mortimer trap test bug in Stockfish 10

Post by ailin » Fri Jun 07, 2019 11:43 pm

Thanks Dann for such a detailed explanation. Great!

But something to consider :
In the book “Comperhensive chess course “ authors present this as opening trap , and finally conclude that :
There are 2 traps here. first mortimer trap , and then white player sets a COUNTER-TRAP by ignoring to save the white bishop.
They go further and say : there are 2 types of traps , good trap and bad trap. Good trap is that one which if your opponent doesn’t fall in it , at least can’t do a damage to your situation. So they conclude that white’s Trap is a good one !

I think their statement exactly conflicts with yours, ‘cause you say :

"Here Nc4 is part of the suggested sequence but it is in fact a blunder giving away a pawn in value. White is still in trouble, but not as bad as after Nc4?? "

What do you think ? Is the second trap good or bad one?
Thanx

Post Reply