to sac or not to sac

[smcracraft]

Thoughts on this position from your program's point of view and analysis anyone?

8/8/4kpp1/3p1b2/p6P/2B5/6P1/6K1 b - - 0 1

The presentation:

https://www.youtube.com/watch?v=4Qput5f_Smw

Seems to me black doesn't have to sac to win.

[ZirconiumX]

[D]8/8/4kpp1/3p1b2/p6P/2B5/6P1/6K1 b - - 0 1

Dorpsgek thinks black is fine not sacrificing here, and also thinks White's king is too passive, but can't decide where to put it:

Code: Select all

18 349 38726 105351902 1. ... a3 2. Kf2 Be4 3. g4 f5 4. Kg3 fxg4 5. Kxg4 a2 6. Bd4 Kd6 7. h5 gxh5+ 8. Kxh5 Bf3+ 9. Kg5 Kc6 10. Kf4 Be4 11. Kg5
19 346 86541 225564226 1. ... a3 2. Kf2 Be4 3. g3 Kf5 4. Ke3 a2 5. Kd2 Kg4 6. Bxf6 Kxg3 7. Kc1 d4 8. Bxd4 Kxh4 9. Kb2 Bd5 10. Be5 Kh3

[pgn]
[Event "Dorpsgek analysis"]
[Site "Scarlatina"]
[Date "2017.07.05"]
[Round "1"]
[White "Dorpsgek"]
[Black "Dorpsgek"]
[Result "*"]
[FEN "8/8/4kpp1/3p1b2/p6P/2B5/6P1/6K1 b - - 0 1"]
1. ... a3 2. Kf2 Be4 3. g4 f5 4. Kg3 fxg4 5. Kxg4 a2 6. Bd4 Kd6 7. h5 gxh5+ 8. Kxh5 Bf3+ 9. Kg5 Kc6 10. Kf4 Be4 11. Kg5 *

[Event "Dorpsgek analysis"]
[Site "Scarlatina"]
[Date "2017.07.05"]
[Round "1"]
[White "Dorpsgek"]
[Black "Dorpsgek"]
[Result "*"]
[FEN "8/8/4kpp1/3p1b2/p6P/2B5/6P1/6K1 b - - 0 1"]
1. ... a3 2. Kf2 Be4 3. g3 Kf5 4. Ke3 a2 5. Kd2 Kg4 6. Bxf6 Kxg3 7. Kc1 d4 8. Bxd4 Kxh4 9. Kb2 Bd5 10. Be5 Kh3 *
[/pgn]

[Vinvin]

Thoughts on this position from your program's point of view and analysis anyone?

8/8/4kpp1/3p1b2/p6P/2B5/6P1/6K1 b - - 0 1

The presentation:

https://www.youtube.com/watch?v=4Qput5f_Smw

Seems to me black doesn't have to sac to win.

Some old threads about this subject :
http://talkchess.com/forum/viewtopic.php?t=54289
http://talkchess.com/forum/viewtopic.php?t=44282
http://talkchess.com/forum/viewtopic.php?t=39391
http://talkchess.com/forum/viewtopic.php?t=36965

[smcracraft]

Looks like the sac was just for flair and jarring the opponent psychologically. He had enough of a lead that it is often the right thing to do with sac'ing and for one's visibility, popularity, etc.

That kind of thing could be, perhaps has, been programmed into a computer chess program and perhaps it would be even wise to do from the perspective of a "psyop" against the other player. Not sure how it would be measured but worth considering if no one has.

[Eelco de Groot]

I think it was genuine sac based on human intuition. Without knowing the exact sequence of moves, he felt that this was his best chance to win. It's not as if Topalov immediately resigned or blundered. That would have spoiled the game.

Worth noting I think that for engines if you play the first move it is not hard at all to show a winning score. Even back in 2001

[smcracraft]

My program doesn't decide to sac.

1. 1934 0.22 15 70 f5e4
2. 1868 0.22 114 529 g6g5 h4g5 f6g5
3. 1868 0.22 393 1823 g6g5 h4g5 f6g5
4. 1608 0.22 1102 5096 f5e4 g1f2 g6g5 h4g5 f6g5
5. 1620 0.22 4796 22013 f5e4 g1f2 f6f5 c3d4 f5f4
6. 1620 0.22 12079 54049 f5e4 g1f2 f6f5 c3d4 f5f4 g2g3
7. 1644 0.24 58949 243649 f5e4 g1f2 f6f5 c3d4 f5f4 g2g3 f4f3
8. 1574 0.32 179906 558059 f5e4 g1f2 f6f5 c3d4 f5f4 g2g3 f4g3 f2g3 e6f5
9. 1620 0.58 799326 1374671 f5d3 g1f2 d3e4 c3d4 f6f5
10. 1547 1.81 3027224 1668529 g6g5 h4g5 f6g5 c3d4 f5b1 g1f2 e6f5 f2f3 b1e4 f3g3 g5g4
11. 1558 6.95 14487587 2084650 g6g5 h4g5 f6g5 c3d4 f5d3 g1f2 e6f5 f2f3 d3e4
12. 1559 18.65 34311126 1840089 f5e4 g1f2 e6f5 c3a1 g6g5 h4g5 f5g5 a1d4 g5g6 d4b2 g6f5
13. 1559 61.23 135491905 2212977 f5e4 g1f2 e6f5 c3a1 f5e6 g2g3 e4b1 a1d4 b1d3 g3g4 d3c4 g4g5 f6f5
14. 1572 166.97 326382547 1954706 f5e4 g1f2 e6f5 c3d4 g6g5 h4g5 f5g5 d4e3 g5g4 e3c5 g4f5 c5d4 f5e6 g2g3 f6f5
15. 1572 632.99 1436281826 2269048 f5e4 g1f2 e6f5 c3d4 g6g5 h4g5 f5g5 d4e3 g5g4 e3c5 g4f5 c5d4 f5e6 d4b2 f6f5 b2d4

[smcracraft]

My suggestion is that programming into a chess program the idea that when ahead spatially and materially sufficiently to then sac if the complexity of the resulting tree becomes much higher would be a good idea to try in practice against human players. Many human players, as Tal proved, go absolutely whacko when the sacs start pouring in at them. They play within a constrained environment/space of predictability. The sac changes that equation for many players who do not have as much practice playing in extremely unbalanced positions.

[kbhearn]

Your PV arriving at such a large score contains a number of suspect moves including Be3?? that certainly loses on the spot from an otherwise still unclear position. See if it can win against a stronger engine if you want to test your theory at least crudely that Bh3!! is unnecessary. It's not a simple problem at all as there's a large number of opposite bishop fortresses that can arise while up two pawns that still show a significantly +ive score but aren't progress. I know people have done very thorough analysis of the position and the one conclusion i'm sure of is that Bh3 is the clearest winning move if not the only one (and strong engines do eventually settle on it if left to their own devices).

[smcracraft]

I am not looking for perfection. Just something which beats me handily. As I am a patzer, my program is more than sufficient as it is a grandpatzer. But I appreciate your comments and insight.

[Eelco de Groot]

Thoughts on this position from your program's point of view and analysis anyone?

8/8/4kpp1/3p1b2/p6P/2B5/6P1/6K1 b - - 0 1

The presentation:

https://www.youtube.com/watch?v=4Qput5f_Smw

Seems to me black doesn't have to sac to win.

Some old threads about this subject :
http://talkchess.com/forum/viewtopic.php?t=54289
http://talkchess.com/forum/viewtopic.php?t=44282
http://talkchess.com/forum/viewtopic.php?t=39391
http://talkchess.com/forum/viewtopic.php?t=36965

One more here: "Undaunted, the Q5t personality tackles yet another position." Oh wait, that was Rebel...