Move horizon example?

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

Moderator: Ras

Uri Blass
Posts: 10898
Joined: Thu Mar 09, 2006 12:37 am
Location: Tel-Aviv Israel

Re: Move horizon example?

Post by Uri Blass »

Fguy64 wrote:4r1k1/5p1p/1qNp2p1/p2P4/Pp3P2/1B2r3/6PP/1R1Q3K b - - 0 36

ok this is a position from one of my games. Black hos a nominal material disadvantage.

As I see it, black has an easy forced win starting with 36...Qc5. Given that in some lines checkmate is not achieved for another dozen or so moves, under what conditions might a decent chess engine misevaluate this position?

regards.
This is an easy forced win only for computers.
It is easy to find for humans but not easy to see that it is a forced win.

for computers it is easy to see that this move is winning and even handicapped rybka(single cpu and 32 bits) can see it in 1 seconds at depth 9

[d]4r1k1/5p1p/1qNp2p1/p2P4/Pp3P2/1B2r3/6PP/1R1Q3K b - - 0 1

Analysis by Rybka 3 1-cpu 32-bit :

1...Re3-e4
³ (-0.33) Depth: 2 00:00:00
1...Re3-e4
³ (-0.33) Depth: 3 00:00:00
1...Re3-e4
³ (-0.28) Depth: 4 00:00:00
1...Qb6-c5
³ (-0.30) Depth: 4 00:00:00
1...Qb6-c5
³ (-0.50) Depth: 5 00:00:00
1...Qb6-c5
³ (-0.70) Depth: 5 00:00:00
1...Qb6-c5 2.h2-h4
µ (-0.81) Depth: 5 00:00:00
1...Qb6-c5 2.Qd1-d2 Re8-e4
µ (-1.00) Depth: 6 00:00:00 3kN
1...Qb6-c5 2.h2-h3 Qc5-c3 3.Kh1-h2 Re3-e1
µ (-1.14) Depth: 7 00:00:00 5kN
1...Qb6-c5 2.h2-h3 Qc5-c3 3.Kh1-h2 Re3-e1 4.Qd1-c2 Re1xb1 5.Qc2xb1 Re8-e1
µ (-1.24) Depth: 8 00:00:00 10kN
1...Qb6-c5
-+ (-1.44) Depth: 9 00:00:00 16kN
1...Qb6-c5
-+ (-1.64) Depth: 9 00:00:00 22kN
1...Qb6-c5
-+ (-2.04) Depth: 9 00:00:00 31kN
1...Qb6-c5
-+ (-2.84) Depth: 9 00:00:01 46kN
1...Qb6-c5 2.h2-h3 Qc5-c3 3.Kh1-h2 Re3-e1 4.Qd1-c2 Re1xb1 5.Qc2xb1 Re8-e1 6.Qb1-c2 Qc3-e3
-+ (-3.19) Depth: 9 00:00:01 63kN
1...Qb6-c5
-+ (-3.39) Depth: 10 00:00:03 113kN
1...Qb6-c5
-+ (-3.59) Depth: 10 00:00:04 174kN
1...Qb6-c5 2.Qd1-g1 Qc5-c3
-+ (-3.86) Depth: 10 00:00:07 268kN
1...Qb6-c5 2.Qd1-g1 Qc5-c3 3.Nc6-e5 d6xe5
-+ (-3.86) Depth: 11 00:00:08 322kN
1...Qb6-c5 2.Qd1-g1 Qc5-c3 3.Nc6-e5 d6xe5 4.f4-f5 g6xf5 5.Bb3-c4 Re8-d8 6.Rb1-c1 Qc3-d4 7.Rc1-d1
-+ (-4.05) Depth: 12 00:00:19 741kN
1...Qb6-c5 2.Qd1-g1 Qc5-c3 3.Nc6-e5 d6xe5 4.f4-f5 g6xf5 5.Bb3-c4 Re8-d8 6.Rb1-c1 Qc3-d4 7.Rc1-d1 Qd4-e4 8.d5-d6 f5-f4
-+ (-4.05) Depth: 13 00:00:25 963kN

(so k, 19.05.2009)
Fguy64
Posts: 814
Joined: Sat May 09, 2009 4:51 pm
Location: Toronto

Re: Move horizon example?

Post by Fguy64 »

Thanks Uri. Perhaps not easy for a human to see a forced win from the starting position over the board, but the first few moves are easy enough to find, and th OTB, and then the winning idea becomes more clear.

I would like to quote one particular variation from your output

1...Qb6-c5 2.h2-h3 Qc5-c3 3.Kh1-h2 Re3-e1 4.Qd1-c2 Re1xb1 5.Qc2xb1 Re8-e1 6.Qb1-c2 Qc3-e3 which gives us the following position...

[d]6k1/5p1p/2Np2p1/p2P4/Pp3P2/1B2q2P/2Q3PK/4r3 w - - 3 7

What I have not seen in any of the analyses is the following idea...

after any 7th move by white with the exception of 7.Ne7+ I consider the winning idea to be 7...Qg1+ 8.Kg3 Re3+ followed by 9...Qe1. Now I understand that this does not mean that the program has not considered this line, but I consider this to be a critical idea, and I'm not sure what to make of the fact that I don't see this line listed here in any of the analysis.

edit: please clarify. After one second, Rybkka may find the move, but what eval does it assign after 1 sec? Thank you.
Fguy64
Posts: 814
Joined: Sat May 09, 2009 4:51 pm
Location: Toronto

Re: Move horizon example?

Post by Fguy64 »

I bumped this thread cause I am very interested in what people think of my last comment about the line starting with 7...Qg1+
Uri Blass
Posts: 10898
Joined: Thu Mar 09, 2006 12:37 am
Location: Tel-Aviv Israel

Re: Move horizon example?

Post by Uri Blass »

Fguy64 wrote:Thanks Uri. Perhaps not easy for a human to see a forced win from the starting position over the board, but the first few moves are easy enough to find, and th OTB, and then the winning idea becomes more clear.

I would like to quote one particular variation from your output

1...Qb6-c5 2.h2-h3 Qc5-c3 3.Kh1-h2 Re3-e1 4.Qd1-c2 Re1xb1 5.Qc2xb1 Re8-e1 6.Qb1-c2 Qc3-e3 which gives us the following position...

[d]6k1/5p1p/2Np2p1/p2P4/Pp3P2/1B2q2P/2Q3PK/4r3 w - - 3 7

What I have not seen in any of the analyses is the following idea...

after any 7th move by white with the exception of 7.Ne7+ I consider the winning idea to be 7...Qg1+ 8.Kg3 Re3+ followed by 9...Qe1. Now I understand that this does not mean that the program has not considered this line, but I consider this to be a critical idea, and I'm not sure what to make of the fact that I don't see this line listed here in any of the analysis.

edit: please clarify. After one second, Rybkka may find the move, but what eval does it assign after 1 sec? Thank you.
I do not understand the question

after 1 second in which position.

In the root position we have 3.19 pawns for black.
If you follow this line and give rybka 1 second after 6...Qe3 then the score is more than +6 for black.

1...Qb6-c5 2.h2-h3 Qc5-c3 3.Kh1-h2 Re3-e1 4.Qd1-c2 Re1xb1 5.Qc2xb1 Re8-e1 6.Qb1-c2 Qc3-e3
-+ (-3.19) Depth: 9 00:00:01 63kN
Fguy64
Posts: 814
Joined: Sat May 09, 2009 4:51 pm
Location: Toronto

Re: Move horizon example?

Post by Fguy64 »

Uri Blass wrote:
Fguy64 wrote:Thanks Uri. Perhaps not easy for a human to see a forced win from the starting position over the board, but the first few moves are easy enough to find, and th OTB, and then the winning idea becomes more clear.

I would like to quote one particular variation from your output

1...Qb6-c5 2.h2-h3 Qc5-c3 3.Kh1-h2 Re3-e1 4.Qd1-c2 Re1xb1 5.Qc2xb1 Re8-e1 6.Qb1-c2 Qc3-e3 which gives us the following position...

[d]6k1/5p1p/2Np2p1/p2P4/Pp3P2/1B2q2P/2Q3PK/4r3 w - - 3 7

What I have not seen in any of the analyses is the following idea...

after any 7th move by white with the exception of 7.Ne7+ I consider the winning idea to be 7...Qg1+ 8.Kg3 Re3+ followed by 9...Qe1. Now I understand that this does not mean that the program has not considered this line, but I consider this to be a critical idea, and I'm not sure what to make of the fact that I don't see this line listed here in any of the analysis.

edit: please clarify. After one second, Rybkka may find the move, but what eval does it assign after 1 sec? Thank you.
I do not understand the question

after 1 second in which position.

In the root position we have 3.19 pawns for black.
If you follow this line and give rybka 1 second after 6...Qe3 then the score is more than +6 for black.

1...Qb6-c5 2.h2-h3 Qc5-c3 3.Kh1-h2 Re3-e1 4.Qd1-c2 Re1xb1 5.Qc2xb1 Re8-e1 6.Qb1-c2 Qc3-e3
-+ (-3.19) Depth: 9 00:00:01 63kN
OK I see your pont about Rybkka it identifies a clear win.

I guess I was just wondering whether Rybkka ( or Crafty ) specifically identified the line 7...Qg1+ 8.Kg3 Re3+ followed by 9...Qe1. as being optimal. As I look back over the other posts in this thread, I see evidence that this is not the case. For example I saw lines that continued 7.Bc4 Rc1, when it seems to me that 7...Qg1+ is a much more convincing move.
jarkkop
Posts: 198
Joined: Thu Mar 09, 2006 2:44 am
Location: Helsinki, Finland

Re: Move horizon example?

Post by jarkkop »

How do make 7. ..Qg1+ when you are in check by 7. Ne7+
Fguy64
Posts: 814
Joined: Sat May 09, 2009 4:51 pm
Location: Toronto

Re: Move horizon example?

Post by Fguy64 »

jarkkop wrote:How do make 7. ..Qg1+ when you are in check by 7. Ne7+
you are absolutely correct. My point was that 7...Qg1, or the threat of this move, pretty much forces mate or decisive and convincing material loss. Black does not even need to capture with 7...Qxe7. it appears that some king moves by black are also very good, because the checks will run out and Black's queen and rook are still in position for a mating attack.

and we still have the issue of why some of the computer analysis in this thread display. 7.Bc4 Rc1?