Page 1 of 1

Stockfish 2.0.1 says only 0.68 but GM Roiz resigns!!!

Posted: Tue Feb 01, 2011 9:19 pm
by notyetagm
Here is the link to the chessbomb page containing the Stockfish 2.0.1 analysis:

http://chessbomb.com/o/2011-gibraltar/0 ... chael_ISR/

In the final position GM Roiz (2574) playing Black resigned, but the Stockfish 2.0.1 engine running on chessbomb gives only an eval of 0.68, i.e., White is only slightly better!

What gives? Premature resignation by Roiz? Is Black really completely lost in the final position? Or is the Stockfish eval too low because of the restricted manner in which chessbomb uses it for non-premium members?

Thanks

Re: Stockfish 2.0.1 says only 0.68 but GM Roiz resigns!!!

Posted: Tue Feb 01, 2011 10:01 pm
by Jeremy Bernstein
notyetagm wrote:Here is the link to the chessbomb page containing the Stockfish 2.0.1 analysis:

http://chessbomb.com/o/2011-gibraltar/0 ... chael_ISR/

In the final position GM Roiz (2574) playing Black resigned, but the Stockfish 2.0.1 engine running on chessbomb gives only an eval of 0.68, i.e., White is only slightly better!

What gives? Premature resignation by Roiz? Is Black really completely lost in the final position? Or is the Stockfish eval too low because of the restricted manner in which chessbomb uses it for non-premium members?

Thanks
This looks pretty lost for black, despite what the engines say (although I'm not even remotely at master level, so whatever I say can be taken with a big grain of salt). White has this great outside passed pawn and a solid wall across the center of the board. Black's bishop can't leave the a8-h1 diagonal and will be lost taking that pawn unless the king comes over, at which point Black's pawns are lunch for the White king. Most other engines (incl. Stockfish) that I tried (Critter, HIARCS, ProDeo (more conservative), Shredder) evaluate this after a new minutes as a solid pawn in White's favor which, against an Ivanchuk, is probably just light's out.

Or do I see it wrong?

Jeremy

Re: Stockfish 2.0.1 says only 0.68 but GM Roiz resigns!!!

Posted: Tue Feb 01, 2011 11:02 pm
by notyetagm
Jeremy Bernstein wrote: This looks pretty lost for black, despite what the engines say (although I'm not even remotely at master level, so whatever I say can be taken with a big grain of salt). White has this great outside passed pawn and a solid wall across the center of the board. Black's bishop can't leave the a8-h1 diagonal and will be lost taking that pawn unless the king comes over, at which point Black's pawns are lunch for the White king. Most other engines (incl. Stockfish) that I tried (Critter, HIARCS, ProDeo (more conservative), Shredder) evaluate this after a new minutes as a solid pawn in White's favor which, against an Ivanchuk, is probably just light's out.

Or do I see it wrong?

Jeremy
No, that's the way I saw the position, as lost for Black due to the reasons that you pointed out.

But as I am only a USCF 1948 near-expert, there is a lot that I do not understand.

Re: Stockfish 2.0.1 says only 0.68 but GM Roiz resigns!!!

Posted: Wed Feb 02, 2011 11:28 am
by Swaminathan
In endgame, the general principle is that Knight is slightly better than Bishop is nearly symmetrical position, and definitely better in closed positions. White is pawn up as well...

At any moment, white might play undermining h5 and hunt for the pawns on king side. GM's are able to accurately evaluate the position just like top engines provided the position is not too complicated or overly aggressive.

Still, Black should have continued playing. I believe Houdini would give evaluation as low as +0.3 or 0.4 if Stockfish gave it 0.68.

Best regards,
Swami

Re: Stockfish 2.0.1 says only 0.68 but GM Roiz resigns!!!

Posted: Wed Feb 02, 2011 12:34 pm
by BB+
8/4k2p/1N2p1p1/3b1p2/P2P1P1P/5PK1/8/8 b - - 0 41There could also be extenuating circumstances: Everything went black on The Rock on Monday evening, as play was interrupted for over an hour due to an electricity outage. The participants had to wait by candlelight.

I have not yet completely convinced myself that White can win, but I am reasonably assured. There are a number of problems with the mobility of the kingside pawns, but none seem insuperable.

A line is: 41... Bc6 42. Kf2 Kd6 43. Ke3 Kc7 44. a5, and now White threatens d5 (followed by Kd4 and invasion). If Black plays Bd5, White can switch to Nb6-a4-c5, push the a-pawn so as to allow Nxe6, then win the h-pawn with Nf8 (and Nd7/Nf6 if Black plays Bg8), and I guess finally push the d-pawn. But there are a number of technical difficulties, and Knight has to take some circuitous paths (Nf8-d7-f6-h7-g5-h3-f2).
So 44... Bd5 45. Na4 Bb3 46. Nc5 Bc4 47. Kd2 Kc6 48. Kc3 Kb5 49. a6 Bd5 50. Kd2 Ka5 51. Ke3 Kb6 52. Kd3 Kc6 53. a7 Kb6 54. Nxe6 Kxa7 55. Nf8 Kb6 56. Ke3 Bg8 57. Nd7+ Kb5 58. Nf6 Bb3 59. Nxh7, White wins. Black has some choices on move 52 (like Ka7 or h6), but I don't think any of them work.
And 44... Bb7 45. d5 exd5 46. Kd4 Kd6 47. Na4 Bc6 48. Nc5 Be8 49. a6 Bb5 50. a7 Bc6 51. Na4 Kc7 52. Nc3 Kd6 53. Nb5+ Kd7 54. Kc5 and White wins. Or 44... Bb7 45. d5 exd5 46. Kd4 Kc6 47. Nxd5 Kd6 48. Nf6 and White wins all the pawns on the kingside. These are long sequences (Knight maneuvers are never very fast), and the horizon probably is too far away for computers.

Re: Stockfish 2.0.1 says only 0.68 but GM Roiz resigns!!!

Posted: Thu Feb 03, 2011 3:43 pm
by Dave Mitchell
I ran this through a "shoot out" with Deep Shredder 12 playing both sides (on a 2 core system that was being lightly used). Ply depths were 20 to 35. White won, but not easily, imo.

[Event "?"]
[Site "?"]
[Date "2011.02.03"]
[Round "?"]
[White "Deep Shredder 12"]
[Black "Deep Shredder 12"]
[Result "1-0"]
[SetUp "1"]
[FEN "8/4k2p/1N2p1p1/3b1p2/P2P1P1P/5PK1/8/8 b - - 0 1"]
[PlyCount "114"]
[EventDate "2011.??.??"]

1... Bc6 {1.11/29 243} 2. a5 {0.99/29 219} Kd6 {1.49/30 832} 3. Kf2 {1.42/30
414} Kc7 {1.05/26 390} 4. Ke3 {1.41/29 473} Bd5 {1.71/30 1047} 5. Na4 {1.71/30
349} Kd6 {1.77/29 237} 6. Nc5 {1.66/29 210} Kc7 {1.92/29 405} 7. a6 {1.85/29
210} Kb6 {1.71/27 71} 8. Kd3 {1.85/29 549} Kb5 {1.90/28 270} 9. a7 {1.96/30 443
} Kb6 {1.89/30 540} 10. Nxe6 {1.61/30 512} Kxa7 {1.78/31 225} 11. Nf8 {1.76/31
287} Kb6 {1.68/31 250} 12. Ke3 {2.53/31 260} Kc6 {2.56/32 701} 13. Nxh7 {2.36/
30 213} Bb3 {2.48/32 287} 14. Kd3 {2.45/30 900} Bd5 {3.00/32 279} 15. Nf8 {3.
25/32 292} Bf7 {3.59/33 416} 16. Ke3 {3.63/30 391} Kd6 {3.91/32 576} 17. Nh7 {
2.49/33 332} Bc4 {3.63/33 310} 18. Ng5 {3.73/32 205} Kc6 {3.82/32 516} 19. Nh3
{3.51/33 449} Kd5 {4.43/33 382} 20. Nf2 {3.69/33 366} Bb3 {3.64/34 235} 21. Nd3
{3.60/33 247} Ba4 {3.69/34 262} 22. Ne5 {4.75/33 331} Be8 {6.21/33 988} 23. Kd3
{3.78/33 300} Bb5+ {5.71/31 307} 24. Kd2 {5.84/32 420} Kxd4 {3.63/32 241} 25.
Nxg6 {4.42/32 341} Bc6 {24.09/36 175:15m} 26. Ke2 {13.09/35 1627} Bb5+ {13.94/
31 854} 27. Kf2 {13.13/24 403} Kd5 {31.04/33 71:50m} 28. h5 {23s} Be8 {127} 29.
Ne7+ {14.01/29 247} Ke6 {22.99/28 223} 30. h6 {28s} Kf7 {14.30/26 198} 31. Ke3
{10s} Bb5 {25.59/24 112} 32. Kd4 {21s} Kf6 {+(15)/28 2:27m} 33. Nxf5 {28s} Kg6
{+(14)/31 1:05m} 34. Ke5 {323} Bc4 {+(12)/29 2:55m} 35. h7 {2078} Kxh7 {+(12)/
32 55s} 36. Kf6 {4s} Bd5 {+(12)/33 2:35m} 37. Nd4 {195} Kh6 {+(10)/29 55s} 38.
f5 {31s} Bb7 {+(9)/29 1:02m} 39. Ke7 {26s} Kg5 {+(7)/28 1:08m} 40. f6 {10s} Kf4
{+(7)/26 1:00m} 41. f7 {32s} Kg3 {+(5)/25 1:07m} 42. f8=Q {6s} Kf2 {+(4)/24 1:
37m} 43. Qg7 {21s} Kf1 {+(4)/22 1:54m} 44. Qg3 {31s} Ba8 {#8/22 108} 45. f4 {9s
} Be4 {#7/21 85} 46. Qh2 {71} Bc6 {#16/18 182} 47. Ke6 {56s} Bf3 {#10/19 179}
48. Ke5 {32s} Bd5 {#9/18 85} 49. Qe2+ {59s} Kg1 {+(1)/10 0s} 50. Nf3+ {13s}
Bxf3 {+(1)/99 0s} 51. Qxf3 {15s} Kh2 {#4/1 0} 52. Qe2+ {65} Kh3 {#4/1 0} 53.
Kf5 {32s} Kg3 {#3/1 0} 54. Kg5 {7s} Kh3 {#4/1 0} 55. Qg4+ {12s} Kh2 {#3/1 0}
56. Kh4 {14s} Kh1 {#2/1 0} 57. Kg3 {23s} Kg1 {#1/1 0} 58. Qd1# {9s} 1-0