Tapolov is a sleeper!

[Robert Flesher]

Thanks Ted, this made me laugh!

[Lyudmil Tsvetkov]

Correction: seemingly, white wins this.

After Be1 Rh8 h4 Rd7 Bd2 Bd8, white has to find Bc1, with convincing win by advancing the pawn phalanx, as g5 is protected by the bishop, a move that SF for some reason does not see at all; meaning, it does not take it into consideration at all, but that is the only winning move.

[d]3b1knr/1pqr2p1/p1p3B1/2Pp2P1/3PnP1P/8/PPQB4/1K1R3R w - - 0 4

Why SF does not see Bc1 here, with convincing win? It does not consider the move at all.

h5 on the other hand draws in a line that SF does not prefer to a line that SF likes and that actually loses.

Well, for people that understand, this is the real thriller, and not the Anand-Lagrave game.

We should be grateful to Ted that he unearthed this position: the lines are all very deep, very unconventional, and very multi-faceted, difficult to see both for engines and humans.

It is a pity that no one paid attention to the position I posted above, but it is the main line, with the only winning move Bc1 (or, if you want, first h5 Ne7, and now already necessarily Bc1)

Here SF indeed has very hard time understanding why Bc1 wins, while other lines like Be4 are worse. It has hard time, because the lines are deep, but when you have such gorgeous phalanx/connecting pawns, that are storming pawns at that, you are almost assured of a win. When you look at the diagram above, and see white g5 phalanx storming pawn, you know white has very big advantage. SF, of course, does not have a clue about that, and it needs to calculate something specific, in order to see the win, unfortunately, nothing specific here, just long long lines, with many quiet moves, threatening nothing.

If you have good eval, you know you are winning, if you do not and can not calculate deep consequences, you simply do not know what happens, or get it completely the wrong way. That is why eval matters very much.

Matter of fact is though that the white pawn phalanx simply can not be stopped.

[Lyudmil Tsvetkov]

OK, to make it simpler: first h5 Ne7 are played, after which we arrive at this position:

[d]3b1k1r/1pqrn1p1/p1p3B1/2Pp2PP/3PnP2/8/PPQB4/1K1R3R w - - 0 2

How long for your engines to see Bc1 is the only move that wins?

I hope someone will post with feedback here, as this position is really gorgeous.

[Lyudmil Tsvetkov]

And a second riddle for your engines, arising after the Be4 line instead of the right Bc1:

[d]3b1k1r/1pqrn1p1/p1p4P/2P3P1/3PpP2/8/PPQB4/1K1R3R b - - 0 3

How long for your engines to see gh6 draws, while g6 loses?

if black plays g6, this position arises, which SF prefers most of the time to gh6, giving it better score for black:

[d]3b1k1r/1pqrn3/p1p3pP/2P3P1/3PpP2/8/PPQB4/1K1R3R w - - 0 4

believe it or not, this is simply lost for black

SF thinks for a long time this is leading to a draw, but black loses this.
How could SF know white wins this, as it has no eval about long chains as f4,g5,h6 chain of pawns is? This is an advanced long chain and is very dangerous, but SF does not know that, as long chains do not work on the framework.

Nothing specific to see here, all lines are very deep, although endgame, and the win might arrive at 50 moves time, but it will definitely arrive. If you do not have sufficiently good eval, you simply compute random/unreasonable lines. That position definitely can not be solved without long chain evaluation.

Btw., thinking about it now, did not SF use in its last long chain attempt the connected phalanx bonus to try scoring such chains? And, does not the connected phalanx bonus in SF by chance give more bonus to the 4th than the 5th rank?

In this case, long chains will never work, as rank is very important here. Maybe SF should try with scoring long chains with the main connected pawns bonus, which should give bigger value to 5th than 4th rank. One way or another, black loses this.

[MikeGL]

[pgn]
[Event "Norway Chess"]
[Site "0:13:33-0:09:33"]
[Date "2015.06.21"]
[EventDate "2015.06.15"]
[Round "5"]
[Result "0-1"]
[White "Jon Ludvig Hammer"]
[Black "Veselin Topalov"]
[ECO "D12"]
[WhiteElo "?"]
[BlackElo "?"]
[PlyCount "148"]

1.d4 d5 2.c4 c6 3.Nf3 Nf6 4.e3 Bf5 5.Nc3 e6 6.Nh4 Bg6 7.Nxg6
hxg6 8.Bd2 Nbd7 9.Qc2 a6 10.O-O-O Be7 11.Kb1 Qc7 12.h3 Rd8
13.c5 e5 14.f4 exd4 15.exd4 Nh5 16.Ne2 Nhf6 17.Nc3 Nh5 18.Ne2
Nhf6 19.Ng3 Ng8 20.Bd3 f5 21.Nxf5 gxf5 22.Bxf5 Ndf6 23.Bg6+
Kf8 24.g4 Ne4 25.g5 Rh4 26.Bc1 Qd7 27.Qg2 b6 28.cxb6 Bd6
29.Bxe4 dxe4 30.d5 Ne7 31.dxc6 Qxc6 32.Rhe1 Qxb6 33.Qxe4 Bc5
34.Rxd8+ Qxd8 35.Qc4 Qd4 36.Qxa6 g6 37.Re5 Qd6 38.Qa8+ Kg7
39.Qf3 Bd4 40.Re2 Qd7 41.Qd3 Nc6 42.Rh2 Nb4 43.Qc4 Rxh3 44.Rd2
Qf5+ 45.Ka1 Bc5 46.a3 Nd3 47.Rd1 Nxc1 48.Rxc1 Bf8 49.Qd4+ Kg8
50.Rc7 Qe6 51.Qc4 Qxc4 52.Rxc4 Kf7 53.a4 Rb3 54.a5 Rb5 55.Ra4
Bg7 56.Ra2 Bd4 57.Ra4 Bxb2+ 58.Ka2 Bc1 59.a6 Rb2+ 60.Ka1 Rb8
61.Rc4 Be3 62.Rc7+ Kg8 63.Rc6 Bd4+ 64.Ka2 Kf7 65.Rd6 Bc3
66.Rc6 Bh8 67.Ka3 Bd4 68.Ka4 Rf8 69.Kb5 Kg7 70.Rc7+ Rf7
71.Rxf7+ Kxf7 72.Kc4 Ba7 73.Kd5 Ke7 74.Kc6 Ke6 0-1
[/pgn]

I saw at chessbase that there's a drawing line at the end of this game.
74.Kc6?? was the losing move which is the last move made by white.
Other moves such as 74.Ke5 or 74.f5 draws. Lucky tourn for Topalov.

[Lyudmil Tsvetkov]

[pgn]
[Event "Norway Chess"]
[Site "0:13:33-0:09:33"]
[Date "2015.06.21"]
[EventDate "2015.06.15"]
[Round "5"]
[Result "0-1"]
[White "Jon Ludvig Hammer"]
[Black "Veselin Topalov"]
[ECO "D12"]
[WhiteElo "?"]
[BlackElo "?"]
[PlyCount "148"]

1.d4 d5 2.c4 c6 3.Nf3 Nf6 4.e3 Bf5 5.Nc3 e6 6.Nh4 Bg6 7.Nxg6
hxg6 8.Bd2 Nbd7 9.Qc2 a6 10.O-O-O Be7 11.Kb1 Qc7 12.h3 Rd8
13.c5 e5 14.f4 exd4 15.exd4 Nh5 16.Ne2 Nhf6 17.Nc3 Nh5 18.Ne2
Nhf6 19.Ng3 Ng8 20.Bd3 f5 21.Nxf5 gxf5 22.Bxf5 Ndf6 23.Bg6+
Kf8 24.g4 Ne4 25.g5 Rh4 26.Bc1 Qd7 27.Qg2 b6 28.cxb6 Bd6
29.Bxe4 dxe4 30.d5 Ne7 31.dxc6 Qxc6 32.Rhe1 Qxb6 33.Qxe4 Bc5
34.Rxd8+ Qxd8 35.Qc4 Qd4 36.Qxa6 g6 37.Re5 Qd6 38.Qa8+ Kg7
39.Qf3 Bd4 40.Re2 Qd7 41.Qd3 Nc6 42.Rh2 Nb4 43.Qc4 Rxh3 44.Rd2
Qf5+ 45.Ka1 Bc5 46.a3 Nd3 47.Rd1 Nxc1 48.Rxc1 Bf8 49.Qd4+ Kg8
50.Rc7 Qe6 51.Qc4 Qxc4 52.Rxc4 Kf7 53.a4 Rb3 54.a5 Rb5 55.Ra4
Bg7 56.Ra2 Bd4 57.Ra4 Bxb2+ 58.Ka2 Bc1 59.a6 Rb2+ 60.Ka1 Rb8
61.Rc4 Be3 62.Rc7+ Kg8 63.Rc6 Bd4+ 64.Ka2 Kf7 65.Rd6 Bc3
66.Rc6 Bh8 67.Ka3 Bd4 68.Ka4 Rf8 69.Kb5 Kg7 70.Rc7+ Rf7
71.Rxf7+ Kxf7 72.Kc4 Ba7 73.Kd5 Ke7 74.Kc6 Ke6 0-1
[/pgn]

I saw at chessbase that there's a drawing line at the end of this game.
74.Kc6?? was the losing move which is the last move made by white.
Other moves such as 74.Ke5 or 74.f5 draws. Lucky tourn for Topalov.

Another 3 lucky wins, and he will be at 2840 live rating.

On the other hand, he plays complicated positions, which he understands better and where his opponents have bigger chance to make a mistake. Topalov does not quite understand positional chess, but what concerns complications, he is very good.

[Lyudmil Tsvetkov]

This time Topalov missed to ... scare the opponent.

I do not have the whole pgn, but that is the position I remember from the online translation:

[d]2n5/p4pk1/bp2p1pp/3p4/P2qPN1P/1P4P1/5PBK/2Q5 w - - 0 1

Topalov played here ed5 with immediate draw, while the obvious Qc7, which SF shows immediately, still leaves white with some 50cps advantage.

Of course, this is also a draw, but at least the opponent could have been scared a bit more.

Man, until now I thought SF is the weakest player on earth, as every 3 moves it makes 2 mistakes, but when I saw this online translation of the games Topalov-Caruana and Nakamura-Grishchuk, I realised that there are even weaker players than SF, as all 4 of the above were playing suboptimal moves all the time.

Man, chess is a difficult game, neither SF, nor any human player still have full grasp of it.

[Lyudmil Tsvetkov]

Man, and this GM Ramirez, commenting for chessbase, how is it possible to have it wrong for all the games?

In Lagrave-Carlsen Carlsen had the advantage, but he says the opposite, in Aronian-Anand Aronian had an obvious edge, but he says position was fully equal, in Topalov-Caruana only Topalov could have edged out a win because of his better bishop and better development, but Ramirez says the other way, in Nakamura-Grischuk first Nakamura could take the advantage with precise play, and then Grischuk, but he claims nothing of the sort.

He stresses the points that should not be stressed, and does not pay attention to the really interesting moments. Does anyone read chessbase?