The strongest passer

[Lyudmil Tsvetkov]

The second diagram strongly favours Black and SF thinks this too:

27 -1.06 177.6M 2:15.59 Ne3 Nb6 Reb1 a4 Bd1 a3 Bxb3 cxb3 Rxb3 Nc4 Qd3 Nxe3 Bxe3 Rfb8 Rxb8+ Rxb8 Bg5 h6 Bxe7 Qxe7 c4 dxc4 Qxc4 Rb2 Qa4 Nb5 d5 exd5 Qa8+ Qf8 Qxd5 Nc3 Qd3 Ne2+ Kf1
26 -1.07 146.8M 1:51.05 Ng5 Nb6 Qc1 a4 Qb1 Bxg5 Bxg5 a3 Bh5 h6 Bh4 Nb5 Re3 a2 Qc1 Na4 Rg3 Kh7 Be2 Rg8 f4 Qa7 Re3 Rgb8 Bf2 Qe7 Rg3 Qa3 Qd2 b2 Nxb2 Nxb2
26 -0.98? 131.3M 1:39.14 Ng5 Nb6?
26 -0.96? 121.8M 1:31.70 Ng5 Nb6?
26 -0.90? 108.0M 1:21.27 Ng5 Nb6?
25 -0.84 101.8M 1:16.47 Ng5 Nb6 Qc1 a4 Qb1 Bxg5 Bxg5 a3 Bh5 h6 Bh4 Nb5 Re3 a2 Qc1 Na4 Rg3 Kh7 Be2 Rg8 f4 Qa7 Re3 Rgb8 Bh5 Rf8 Qd2 Naxc3 Rxc3 Qxd4+ Qxd4 Nxd4
25 -0.85? 100.9M 1:15.69 Ng5 Nb6?
25 -0.77! 99.3M 1:14.47 Ng5!
25 -0.87! 98.6M 1:13.86 Ng5!
25 -0.93! 98.4M 1:13.64 Ng5!
24 -0.99 91.6M 1:08.51 h4 Nb6 h5 a4 Nh2 a3 h6 g6 Ng4 Qd8 Nf6+ Bxf6 exf6 Nb5 Bg5 Na4 Bf3 b2 Nxb2 axb2 Rxa4 Rxa4 Qxb2 Qa5 Bd2
24 -1.03? 84.2M 1:02.34 Ne3 Nb6?
24 -0.97? 72.8M 0:53.84 Ne3 Nb6?
23 -0.91 63.7M 0:47.08 Ne3 Nb6 Reb1 a4 Bd1 a3 Bxb3 cxb3 Rxb3 Nc4 Qc2 Nxe3 Bxe3 Rfb8 Ng5 Bxg5 Bxg5 Rxb3 Qxb3 Nb5 Bd2 Qc6 Ra2 h6 Qb4 g5 Qe7
23 -0.84? 44.9M 0:31.91 Ne3 Nb6?
23 -0.73! 42.0M 0:29.89 Ne3!
23 -0.74! 37.7M 0:26.83 Ne3!
23 -0.81? 31.8M 0:22.67 Ne3 Nb6?
22 -0.74 25.9M 0:18.55 Ne3 Nb6 h4 a4 Ng4 a3 Bg5 Qd8 Bxe7 Qxe7 h5 b2 Ra2 h6 Rb1 Rfb8 Nf6+ gxf6 Qxh6 Qf8 Qf4 fxe5 Qxe5 Nb5
22 -0.70? 21.8M 0:15.80 Ne3 Nb6?
22 -0.60? 18.1M 0:13.24 Ne3 Nb6?
22 -0.54? 15.1M 0:11.16 Ne3 Nb6?
21 -0.48? 11.7M 0:08.83 Ne3 Nb6?
20 -0.41 10.6M 0:08.05 Ne3 Nb6 Ng4 a4 Bg5 Qd8 Bxe7 Qxe7 h4 a3 h5 b2 Ra2 f5 exf6 gxf6 Bf1 Nb5 Be2 Qg7 Nh6+ Kh8
20 -0.31! 10.3M 0:07.77 Ne3!
20 -0.47? 7.97M 0:06.11 Ne3 Nb6?
20 -0.26? 6.32M 0:04.91 Ne3 Nb6?
20 -0.18! 6.15M 0:04.78 Ne3!
20 -0.28! 6.11M 0:04.75 Ne3!
20 -0.34? 5.73M 0:04.49 Ne3 Nb6?
19 -0.28? 4.91M 0:03.94 Ne3 Nb6?
18 -0.22! 4.19M 0:03.41 Ne3!
18 -0.32? 4.14M 0:03.34 Ne3 Nb6?
18 -0.25! 3.93M 0:03.20 Ne3!
18 -0.34! 3.92M 0:03.20 Ne3!
18 -0.40! 3.92M 0:03.20 Ne3!
17 -0.46 3.81M 0:03.10 Ne3 Nb6 Ng4 a4 Bh6 Ne8 Bxg7 Nxg7 Nf6+ Bxf6 exf6 Nf5 g4 Qd8 gxf5 Qxf6 fxe6 fxe6 Qg5+ Qxg5+ Nxg5 a3 Nxe6 Rf5 Nc5
16 +0.14 1.32M 0:01.17 h4 Nb6 h5 a4 h6 g6 Ne3 a3 Ng4 b2 Rad1 Na4 Bg5 a2 Bxe7 Qxe7 Nf6+ Kh8 Ng5
15 +0.04 1.12M 0:01.02 h4 Rfb8 h5 h6 Be3 b2 Rb1 Rb3 Bxh6 gxh6 Qxh6 Nb5 Ng5 Bxg5 Qxg5+ Kh7 Ne3 Qd8
14 +0.20 834219 0:00.81 h4 Rfb8 Bg5 b2 Rb1 Bxg5 hxg5 Rb3 Nxb2 Nxc3 Ra1 Nxe2+ Qxe2 a4 Nd1 Qd8
13 +0.24 437780 0:00.53 h4 Rfb8 Bg5 b2 Rb1 Bxg5 Nxg5 h6 Nf3 Nb5 h5 Qb7 Rxb2 Nxb2 Nxb2
12 +0.44 256672 0:00.36 h4 Rfb8 h5 h6 Be3 b2 Rb1 Rb3 Bxh6 Nb5 Bf4 Nbxc3 Nxc3
11 +0.77 115748 0:00.17 h4 Nb6 h5 h6 Bxh6 gxh6 Qxh6 a4 Ng5 Bxg5 Qxg5+ Kh7 h6 f5 exf6
10 +0.42 96661 0:00.14 Bg5 Bxg5 Qxg5 b2 Rb1 Rab8 Qd2 Rb3 Rxb2 Nxb2
9 +0.64 25675 0:00.05 g3 Rfb8 h4 h6 Nb2 Nxb2 Qxb2 a4 Nd2
8 +0.82 10890 0:00.02 h4 Rfb8 Ne3 b2 Rad1 h6 g4 Ba3
7 +0.70 8221 0:00.02 h4 Rfb8 Nb2 Nxb2 Qxb2 a4 Red1
6 +0.91 2039 0:00.00 Ne3 Nb6 Bg5 Bxg5 Nxg5 h6
5 +0.93 1460 0:00.00 Ne3 Nb6 Nf1 a4 Ne3
4 +1.05 679 0:00.00 Rb1 h6 Nb2 Nxb2
3 +1.27 423 0:00.00 Nb2 Nxb2 Qxb2
2 +1.52 186 0:00.00 Rb1 f6
1 +1.27 39 0:00.00 Nb2 Nxb2
0 #

But as I wrote in the previous post SF avoids this position. I see that other engines (Komodo) avoid the b3 passer position too.

Static eval at ply 1 - 1.27 in white's favour, static eval at ply 27 - 1.06 in black's favour!

2.5 pawns difference, obviously SF eval for this position is fully off target.

[Lyudmil Tsvetkov]

I tried to look at the position more closely, and came to the conclusion that it is too complicated for me.
White attacks on the kingside with all kinds of possible sacrifices, black threatens to promote on the queenside, a single move might decide.

However, my logic runs as follows:

- I know that at the initial main diagram black has some noticeable advantage
- I also know that, if black does not find counterplay on the queen side, white will use its superior position on the king side to force events, getting advantage
- consequently, black should have a move preventing that, and I see no other move than a piece sacrifice on a4; therefore, the sacrifice should be beneficial for black, no matter how deep and convoluted the respective lines are

Eagerly waiting for longer high-end machine analysis.

[lantonov]

SF gives the following variant on your Black moves:

1... Bxa4 2. bxa4 Nxa4 3. h4 Nb6 4. Ne3 a4 5. Ng4 a3 6. h5 Qa4 7. h6

with some 2.5 pawns advantage for White.

[jorose]

Im not as sure that black has an advantage in the original diagram, the reason SF does not take on a4 is because he feels black is worse if he does so but thinks he sees how to hold the balance otherwise. White would never play c3 if black can reply b3, that would be completely irrational. Whites attack on the Kingside is not to be underestimated and sacrificing a piece does give black counter-play, but also removes certain options as black is a piece down, which might have been useful if the position opens or in order to organize a successful defense.

That is not to say that you are wrong, just that this is far from clear cut to me.

[Lyudmil Tsvetkov]

SF gives the following variant on your Black moves:

1... Bxa4 2. bxa4 Nxa4 3. h4 Nb6 4. Ne3 a4 5. Ng4 a3 6. h5 Qa4 7. h6

with some 2.5 pawns advantage for White.

I did not say I will play them straight forward.

This is just the plan, meaning, that you should not push a pawn to c3 or b3 first, but instead to a3. You can well do without Qa4, push a3 later, etc.

As said, the position is simply too complicated for quick browsing.
Black can sacrifice another exchange on f8, for example, try to defend with Rb8 and Bf8, etc.

All too complicated for quick browsing.

[lantonov]

I am sure you can prove the machine wrong here, as you have done many other times.

[Lyudmil Tsvetkov]

Im not as sure that black has an advantage in the original diagram, the reason SF does not take on a4 is because he feels black is worse if he does so but thinks he sees how to hold the balance otherwise. White would never play c3 if black can reply b3, that would be completely irrational. Whites attack on the Kingside is not to be underestimated and sacrificing a piece does give black counter-play, but also removes certain options as black is a piece down, which might have been useful if the position opens or in order to organize a successful defense.

That is not to say that you are wrong, just that this is far from clear cut to me.

OK, I see a line like Ba4 ba4 Na4 h4 Nb6 Ne3 Rfb8 Ng4 Ne8, and black should win that

[d]rr2n1k1/3qbppp/1n2p3/p2pP3/1ppP1BNP/5N2/2PQBPP1/R3R1K1 w - - 0 1

The white moves are the moves SF suggests to play in the posted output, but I include Rfb8, to free the f8 square for the black e7 bishop and the rook on b8 helping with pushing forward the queenside passers.

[Lyudmil Tsvetkov]

Or maybem the f8 rook should go to c8 instead of b8, to avoid any tactical complications after Nf6.

[d]r1r1n1k1/3qbppp/1n2p3/p2pP3/1ppP1BNP/5N2/2PQBPP1/R3R1K1 w - - 0 1
So I guess this is one of the positions worth closer look.

[Lyudmil Tsvetkov]

Now, I can not scientifically prove that this is won for black.

However, I played a couple of shootouts with SF 6, indicating here and there a possible good candidate move, and the general trend is that SF score changes from highly convincing in white's favout to straightforward draw, and from minmal white advantage to draw, and then proceeding to negative and won for black.

What is certain is that the score constantly decrease and almost always becomes negative.

[d]4nbk1/1q3p1p/1rn1p1pP/3pP3/rppP2N1/p3QNP1/R1PB1P1K/1R1B4 w - - 0 14
Above you see a diagram of a critical position.
Even here SF can not understand that black has substantial advantage, and only black can win, as the king side is closed, and it is obvious that only black attacks, while white defends all the time, even with a piece more.

The reason is probably too many regrouping moves until a recognizable won situation is reached: you might need to make 20 or 30 preparatory moves, until a break becomes possible and black wins. This is obviously too much for engines.

Finally, we reach a position like that:

[d]2qn2k1/r4p1p/3bpPpP/3p2N1/r1pP2B1/1pP1Q1P1/pR3PK1/R1B5 w - - 0 37
SF does not recognize promptly even here that white is losing, though it already gives a positive score for black.

Now, in such a case, I can do nothing for SF, as it is obvious that the white situation is pitiful in the least.

I know no engine analysis will see black win here, too many preparatory regrouping moves until a winning situation is reached.
But the win is there, and therefore I say that a passer connecting to an own larger chain, especially when the passer is supported by another passer, as a5 in this case, deserves a big additional bonus, of course if you know how to play it.

But even if one might have doubts about particular positions, I think the term is valid in a wide range of situations with different piece placements.

[Lyudmil Tsvetkov]

[d]r1r1n1k1/3qbppp/1n2p3/p2pP3/1ppP1BNP/5N2/2PQBPP1/R3R1K1 w - - 0 1

Actually, black wins even easier here.

For example: Bg5 Bf8, followed by Na4-c3, and black pushes the a passer forward.

White can do nothing to prevent that.

I can't believe there are engines that do not recognize an easy win for black on the above diagram.

So, quite probably, on the original diagram, black has at least 70-80cps advantage in terms of eval at the root.