Nezhmetdinov - Chernikov - 1962 - Queen Sac

[Lyudmil Tsvetkov]

I do not envy those who would try asking advice from an engine in such position; SF and co are simply unbearbale here. SF wants to draw all the time, while suggesting the worse possible white continuations.

it is difficult to seek advice from engines that are stuck somewhere in the late 17th-early 18th century, that was the way they played back then, captures, captures and more captures, rarely a sound positional moves.

[yanquis1972]

thanks lyudmil -- in the latest position after gxf i havent found a way to make progress w/ white. [d]r4k2/ppr2p1p/3N4/5p2/1P1BP3/2P5/1q4PP/R4RK1 w - - 0 14

[Lyudmil Tsvetkov]

thanks lyudmil -- in the latest position after gxf i havent found a way to make progress w/ white. [d]r4k2/ppr2p1p/3N4/5p2/1P1BP3/2P5/1q4PP/R4RK1 w - - 0 14

who knows, John, maybe Rab1 and then e5 (Qc2 Rc1, not allowing Rc3 sac), this is already too much machine-like, but also very deep, I guess in the end white reaches an imbalance RRBN + 2 pawns vs QRR, almost the same as the previous imbalance, but with fewer pawns, which migth raise a bit drawing chances.

i really do not have time now, very brief check gives still some white plus from SF many moves on, whether it is a win or not, I can not tell you.

but we are missing awful lot of moves along the way, in any case what I see is that in 90% of lines black loses quite convincingly, so this is not a position black should go into.

sorry again, really pressed now, maybe tomorrow, God willing, will be back with more imput.

(and I have the impression that the position hides even more for white, just that those moves require very deep and accurate analysis)

[Lyudmil Tsvetkov]

thanks lyudmil -- in the latest position after gxf i havent found a way to make progress w/ white. [d]r4k2/ppr2p1p/3N4/5p2/1P1BP3/2P5/1q4PP/R4RK1 w - - 0 14

who knows, John, maybe Rab1 and then e5 (Qc2 Rc1, not allowing Rc3 sac), this is already too much machine-like, but also very deep, I guess in the end white reaches an imbalance RRBN + 2 pawns vs QRR, almost the same as the previous imbalance, but with fewer pawns, which migth raise a bit drawing chances.

i really do not have time now, very brief check gives still some white plus from SF many moves on, whether it is a win or not, I can not tell you.

but we are missing awful lot of moves along the way, in any case what I see is that in 90% of lines black loses quite convincingly, so this is not a position black should go into.

sorry again, really pressed now, maybe tomorrow, God willing, will be back with more imput.

(and I have the impression that the position hides even more for white, just that those moves require very deep and accurate analysis)

ok, if John is still around or someone is following, I guess suggested Rb1 and e5 migth still be winning, but, white has even more convincing moves, of course. instead of b4, white can continue with Rab1 (gaining some tempo, so b3 pawn is defended) Qc2 Rfc1 Qa2 and now f6 (avoiding exchanges, lessening pawn material and improving drawing chances) a5 e5:

[d]r4k2/1pr2p1p/3N1Pp1/p3P3/3B4/1PP5/q5PP/1RR3K1 b - - 0 16

and now I am fully certain this is a white win in terms of eval.

[Lyudmil Tsvetkov]

another attacking option is to play h4-h5 instead of f5 first, for example after Rfc8 h4 Rc5 Rab1 Qc2 h5:

[d]r7/pp3pkp/3p1Np1/2r4P/3BPP2/1PP5/2q3P1/1R3RK1 b - - 0 12

I do not know what happens, too complex to analyse, but white is attacking.

[Lyudmil Tsvetkov]

or, even if white captures with the c pawn instead of the a pawn, cb3, after, for example, Rfc8 (Qd2 here is simply bad, because of Bc3) Ne8 Kf8 Nd6 Rc2 e5 (Nc4 is also possible):

[d]r4k2/pp3p1p/3N2p1/q3P3/3B4/1P6/PPr2PPP/R4RK1 b - - 0 10

we get to a position with a lot of pawns and the same imbalance, which only white can win, in spite of SF convincing scores in favour of black. white has a doubled pawn, but the minor pieces will be able to undouble it at some point. here black has also very active rook, but this will be only temporary, though a considerable asset.

believe me or not, even if you remove 2 pawns from white, say e5 and b3, provided black can not create passers tactically, the position is still drawn. so, if white holds with 2 pawns removed, why should not white have good winning chances overall with 2 pawns more? of course, everything is not that simple as tactics matter a lot, but the point is that, even without some pawns in most of the positions white can not lose, while if you remove another pawn from black, white wins almost immediately.

so, those are all very complex lines, but, in most of the lines, if not all, only white could win and is winning. until we have 100% certainty though, for this and many other complex positions we have discussed in the past, we will probably need some 500-1000 elo stronger engine, current SF running on 100 000 000 cores or, the 100 leading GMs spending half a year each on feedback.

not very certain which is the most probable of the above hypotheses, but one way or another, we seemingly do not have other options but to wait until things get more clear.

[yanquis1972]

still here, & appreciate the follow-thru. a late morning for me so i havent been able to examine your suggestions in any kind of depth yet, but just checking the first line i find it quite solid.

the issue for me of course is that i am entirely at the mercy of the engines here, & currently the one i'm checking with still evaluates blacks position as roughly a pawn up. (houdini 5, [d]r4k2/1pr2p1p/3N1Pp1/p3P3/3B4/1PP5/q5PP/1RR3K1 b - - 0 16)

my guess would be in an engine shootout from the position, the result owuld be draw 9/10 (which is to say that black is not actually winning, even by the computers that give an eval suggesting it)

my quick & dirty attempt to progress from this root gives this result, which is concretely drawn:

[pgn][Event "?"]
[Site "?"]
[Date "?"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[FEN "r1b2rk1/pp1ppp1p/5Qp1/q7/3nP3/1BN1B3/PPP2PPP/R4RK1 b - - 0 1"]
[SetUp "1"]

1...Ne2+ 2.Nxe2 exf6 3.Nc3 d6 4.Nd5 Be6 5.Nxf6+ Kg7 6.Bd4 Bxb3 7.axb3 Qd2 8.c3 Qxb2 9.f4 Rfc8 10.f5 Rc5 11.Ne8+ Kf8 12.Nxd6 Rc7 13.Rab1 Qc2 14.Rfc1 Qa2 15.f6 a5 16.e5 Ra6 17.c4 Rd7 18.c5 Qd2 19.Rd1 Qe2 20.Re1 Qg4 21.Bf2 Kg8 22.e6 fxe6 23.Re4 Raxd6 24.Rxg4 Rd1+ 25.Rxd1 Rxd1+ 26.Be1 Rxe1+ 27.Kf2 Rb1 28.Ra4 Rb2+ 29.Ke1 Rxb3 30.Rxa5 Kf7 31.g4 Kxf6 *[/pgn]

ofc i dont think its best, but combined with the immensely skeptical root eval, i'm at least momentarily still unable to demonstrate any + for white.

will look at this further, as well as at your other suggestions, as i wake up

[beram]

thanks lyudmil -- in the latest position after gxf i havent found a way to make progress w/ white. [d]r4k2/ppr2p1p/3N4/5p2/1P1BP3/2P5/1q4PP/R4RK1 w - - 0 14

who knows, John, maybe Rab1 and then e5 (Qc2 Rc1, not allowing Rc3 sac), this is already too much machine-like, but also very deep, I guess in the end white reaches an imbalance RRBN + 2 pawns vs QRR, almost the same as the previous imbalance, but with fewer pawns, which migth raise a bit drawing chances.

i really do not have time now, very brief check gives still some white plus from SF many moves on, whether it is a win or not, I can not tell you.

but we are missing awful lot of moves along the way, in any case what I see is that in 90% of lines black loses quite convincingly, so this is not a position black should go into.

sorry again, really pressed now, maybe tomorrow, God willing, will be back with more imput.

(and I have the impression that the position hides even more for white, just that those moves require very deep and accurate analysis)

ok, if John is still around or someone is following, I guess suggested Rb1 and e5 migth still be winning, but, white has even more convincing moves, of course. instead of b4, white can continue with Rab1 (gaining some tempo, so b3 pawn is defended) Qc2 Rfc1 Qa2 and now f6 (avoiding exchanges, lessening pawn material and improving drawing chances) a5 e5:

[d]r4k2/1pr2p1p/3N1Pp1/p3P3/3B4/1PP5/q5PP/1RR3K1 b - - 0 16

and now I am fully certain this is a white win in terms of eval.

Dear Lyudmill,
I have looked into the positions you have presented and my 'guess' is opposite to yours: black is better and white has to play very carefully for a draw

To start with your last position after a5 - e5 (btw a5 is suboptimal b6 is better move for black) Black plays Ra6 and in the best line I have found 17 c4 white can probably hold the draw but not easy

I am certain that you will disagree, but if you want I will provide you the lines.
They are cross examined with the three best engines in the world, playing on al level of approx 3300 ELO (ours are probably 1300 lower)

grts Bram

[yanquis1972]

so, those are all very complex lines, but, in most of the lines, if not all, only white could win and is winning.

i think this is where, for the moment at least, we disagree, & it may simply have to be left at that: while engine eval is clearly wrong/misguided, i'm just not seeing the case that white is objectively winning. where i am more inclined to trust you is that white has good winning chances, but again, with an engine on in the background & me moving the pieces, it seems positively hopeless. even stacking the eval to favor whites pieces from a static situation with komodo's tuning factors like contempt/dynamism/king safety, its a matter of time before the evaluation regresses to 0.00.

i suppose my hope is that we can reach an agreed position where you can demonstrate a winning plan, as my endgame understanding is non-existent, but naturally i understand if we're unable to get there.

[Lyudmil Tsvetkov]

still here, & appreciate the follow-thru. a late morning for me so i havent been able to examine your suggestions in any kind of depth yet, but just checking the first line i find it quite solid.

the issue for me of course is that i am entirely at the mercy of the engines here, & currently the one i'm checking with still evaluates blacks position as roughly a pawn up. (houdini 5, [d]r4k2/1pr2p1p/3N1Pp1/p3P3/3B4/1PP5/q5PP/1RR3K1 b - - 0 16)

my guess would be in an engine shootout from the position, the result owuld be draw 9/10 (which is to say that black is not actually winning, even by the computers that give an eval suggesting it)

my quick & dirty attempt to progress from this root gives this result, which is concretely drawn:

[pgn][Event "?"]
[Site "?"]
[Date "?"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[FEN "r1b2rk1/pp1ppp1p/5Qp1/q7/3nP3/1BN1B3/PPP2PPP/R4RK1 b - - 0 1"]
[SetUp "1"]

1...Ne2+ 2.Nxe2 exf6 3.Nc3 d6 4.Nd5 Be6 5.Nxf6+ Kg7 6.Bd4 Bxb3 7.axb3 Qd2 8.c3 Qxb2 9.f4 Rfc8 10.f5 Rc5 11.Ne8+ Kf8 12.Nxd6 Rc7 13.Rab1 Qc2 14.Rfc1 Qa2 15.f6 a5 16.e5 Ra6 17.c4 Rd7 18.c5 Qd2 19.Rd1 Qe2 20.Re1 Qg4 21.Bf2 Kg8 22.e6 fxe6 23.Re4 Raxd6 24.Rxg4 Rd1+ 25.Rxd1 Rxd1+ 26.Be1 Rxe1+ 27.Kf2 Rb1 28.Ra4 Rb2+ 29.Ke1 Rxb3 30.Rxa5 Kf7 31.g4 Kxf6 *[/pgn]

ofc i dont think its best, but combined with the immensely skeptical root eval, i'm at least momentarily still unable to demonstrate any + for white.

will look at this further, as well as at your other suggestions, as i wake up

I can not demonstrate a 100% win, simply because there are zillions of lines to consider and we are missing moves for both sides on each and every ply. I even do not know if after Ne2 check early in the game, white should not retreat Kh1?! instead of capturing the knigth. (we never checked that)

what I certainly know however are those 3 things:

1. black made 2 mistakes/inaccuracies before the sac: Qa5 (queen is attackable by white pawns, not the best place here), and Bf6 (bishop's place is also not on f6 for certain, attackable by pawns knigth on e4, etc.). apart from that, black plays black, losing tempo. overall, -50cps at least

2. in at least 80% of all positions checked with SF, I get to a white plus, in more than 65% of them to very significant white plus.

3. my assessment of static eval (which of course might be incorrect) is in the range of 50-70cps white edge.

taking only John's latest development, after Qe2 fro example, white migth play h3 instead of Re1, not allowing the black queen to get to g4 and e6:

[d]5k2/1p1r1p1p/r2N1Pp1/p1P1P3/3B4/1P5P/4q1P1/1R1R2K1 b - - 0 4

too many lines to tell you what happens, but:

- f6 pawn is defended pawn on f6, which is much bigger edge than a simple pawn on f6
- c5 and e5 pawns are incredibly strong duo across a file, very nice bonus
- black b7 pawn is backward, can not move forward, as Bd4 controls the b4 square apart from the c5 pawn, and black has no minor pieces to assist the pawn forward
- black king is in extreme danger, white can attack with pawns on the king side, push e5 pawn at some point, etc.
- imagine what happens if on the diagram above, we place a white rook on h8? so white just need to move one rook to h file, open it and mate

no concrete lines, but too many threats and too good eval, for the game not to end with a white win in at least 80% of reasonable moves.