Shakmaty Bereolos - The Official Chess Site of Peter Bereolos

5/29/20 - TCEC Season 17 Superfinal, Games 9-10: Leningrad Dutch

Games 9 and 10 in the TCEC Spuperfinal were both drawn. The starting position was a sideline of the Leningrad Dutch. 1.d4 f5 2.g3 Nf6 3.Bg2 g6 4.Nf3 Bg7 5.0-0 0-0 6.c4 d6 7.Nc3 e6

This is quite uncommon compared to 7...Qe8, 7...c6, and 7...Nc6 Sandipan and Reindermann seem to be the main practitioneers. However, the move does have historical pedigree. Botvinnik used it (with neither side yet castled) trailing by two points late in his first match with Tal. Although Tal won the game, he did not consider it to be a bad continuation in his book on the match. 8.Re1 I don't know if this move was still part of the proscribed opening moves or if both engines selected it. Tal played 8.Qc2 with the idea of preventing Ne4. 8...Ne4 When Stockfish was Black, it opted for the immediate 8...d59.Qc2 d5 Sandipan and Reindermann both played 9...Nxc3 in this position. I think the opening debate can be closed here. If transposing to a Stonewall is the best Black has, then the idea doesn't seem very promising. Black would never play g6 in a Stonewall and he has wasted a move by playing d5 in two moves.

There were a few other positions of note later in the game that I think show the engines still have some weaknesses in closed positions. 10.Bf4 c6 11.e3 Nd7 12.Rac1 Qe7 13.b3 Re8 14.Be5 Nxe5 15.Nxe5 Bd7 16.Ne2 Rac8 17.c5 Ng5 18.f4 Nf7 19.Nd3 g5 20.h3 Kh8 21.Qc3 Rg8 22.Bf3 Bf6 23.Kf2 Ra8 24.b4 Rg7 25.a4 Rag8 26.Rh1 Be8

Here, Stockfish uncorked the absurd 27.a5?!Certainly no human would remove the possibility of a queenside breakthrough with b5. Perhaps its evaluation function gave a plus for forcing one more Black pawn on to the light squares. 27...a6 28.Rcg1 Nd8 29.Qb2 Qf8 30.Kf1 Qe7 31.Kf2 Qf8 32.Rh2 Qe7 33.Qa1 h5

Likewise, here it looks like Stockfish again saw the opportunity to fix another Black pawn on light squares with 34.fxg5 Rxg5 35.h4Black's position is somewhat of a Stonewall player's nightmare with the light squared bishop trapped behind the pawns. But White doesn't have any pawn breaks, and the pawn at h5 is easily covered (bad bishops guard good pawns). 35...R5g6 36.Nef4 Rh6 37.Rhg2 Kg7 38.Rf1 Kf8 39.Qe1 Bf7 40.Qd1 Rgh8 41.Kg1 Ke8 42.Ne5 Qc7 43.Nfd3 Kf8 44.Qe2 Kg8 45.Qf2 Be8 46.Be2 Nf7 47.Qf4 Nxe5 48.Nxe5 Qg7 49.Kf2 Bd8 50.Rfg1 Bc7 51.Ke1 R8h7 52.Kd2 Kh8 53.Rf2 Bxe5 54.Qxe5 Qxe5 55.dxe5 Rg7 56.Bf3 Bd7 57.Ke2 Rgg6 58.Rff1 Rh7 59.Kf2 Rg8 60.Rd1 Rg6 61.Rh1 Kg8 62.Kg2 Rh8 63.Be2 Kg7 64.Kf2 Kh6 65.Bf3 Rhg8 66.Rhg1 Be8 67.Rc1 Rh8 68.Rcd1 Bd7 69.Rd4 Ra8 70.Rd2 Rag8 71.Rd3 Rd8 72.Rdd1 Rdg8 73.Be2 R6g7 74.Rd4 Rg6 75.Bd3 Ra8 76.Bf1 Rgg8 77.Rh1 Kg7 78.Be2 Kh6 79.Rdd1 Rac8 80.Ra1 Rcf8 81.Rhd1 Rg6 82.Kg2 Rfg8 83.Kf2 R6g7 84.Rg1 Rg6 85.Ra3 R6g7 86.Rb3 Kg6 87.Kg2 Re7 88.Rd3 Kh6 89.Kf2 Rg6 90.Ke1 Re8 91.Rd4 Rh8 92.Bf3 Rb8 93.Kf2 Rbg8 94.Be2 R6g7 95.Rg2 Rg6 96.Bd1 Ra8 97.Kg1 Rgg8 98.Kh2 Rac8 99.Rg1 Rb8 100.Kg2 Rg7 101.Kf2 Rbg8 102.Be2 Be8 103.Bf3 Bd7 104.Rdd1 Rg6

Human players would have arranged to repeat the position 3 times long before getting here. I didn't really try to check to see if one side or the other was avoiding the repetition. If Stockfish's contempt setting was high, I think it would avoid repetition as long as it didn't significant change its evaluation. So if it saw two moves both at 0.00 it would chose one that did not repeat for a third time. On the other side, from what I understand of the AI engines, they evauate the position in terms of win probability. So Lc0 might be seeing the position as 99.99% drawn with 0.01% chance of winning. The option that would repeat the position for a third time would make it 100% drawn, so it chooses something else to keep the miniscule winning chance alive. In this position, Stockfish showed that it may have a sense of humor as the move it selected to enforce the 50 move rule was 105.Be4 1/2-1/2

5/27/20 - BCE-349a

Just a short BCE post this week. Fine seems to have made up this position in order to illustrate a stalemate defense, but the defending side has an alternative. Benko decided to completely remove this flawed example from the revised edition. He could have considered using the tragicomedy the Dvoretsky showed in his Endgame Manual from the game between Joerg Hickl and Stephen Solomon in the West Germany-Australia match at the 1988 Olympiad. After 63.Kg4

Black indirectly defended his g-pawn with 63...Rb3? The right idea, but he needed to move the rook further with 63...Rb1 or 63...Rb2. Now, the White king is short of squares 64.Rg8+! Kf6 65.Rf8+? 65.Rg6+! is the stalemate trick. The game was adjourned here. Black sealed 65...Kg6? Black is winning after 65...Kg7, but decided to repeat moves and work the win out at leisure with his teammates. He was quite fortunate that his opponent resigned without resumption! 0-1?

5/24/20 - Bereolos-Moradiabadi, Land of the Sky XXX

GM Elshan Moradiabadi has become a fixture at the annual Land of the Sky tournament winning or tying for first in all but one of the last seven editions. The year he didn't come out on top was 2017, when he was nicked for 3 draws. I almost made it 4, but lost a long ending where the assesment changed from draw to win several times.

5/20/20 - BCE-224d

Endings with a bishop and two connected passed pawns are generally won for the attacking side unless the defender can set up a blockade. Without a blockade, about the only chance is if the attacker has a rook's pawn that queens on the opposite color of the bishop. However, with care, as shown in the correction link, the attacker can avoid allowing the knight to sacrifice itself for the knight's pawn.

In practice, even strong players can get it wrong. Former US Champion Lubomir Kavalek managed to hold the ending twice. First, at Las Palmas 1974 against Guillermo Garcia, after 51...Nxc5

52.h4 Nd3 53.Bb5 Nf4 54.Bc4 Nh5+ 55.Kf3! 55.gxh5? Kf6! and Black will get to the corner. The presence of the extra h-pawn makes no difference. 55...Nf4 56.g5 Nh5 57.Bf7 Nf6 58.g6? 58.Bg6 takes away f5 from the Black king and White is ready to advance the h-pawn. 58...Kf5! 59.g7 White might have thought that this position was zugzwang. 59...Ng8! 1/2-1/2 The bishop blocks the pawn's advance after 60.Bxg8 Kg6

Even a World Championship Candidate was not immune. In Portisch-Kavalek, Montreal 1979 after 57.Kxe4

57...Nf6+ 58.Kd4 Nd7 59.Bd6 Kf5 60.Bc7 Ke6 61.Bxa5 Kd6 62.b4? 62.Bd8 Nb8 63.Kc4 Nc6 64.Bb6+- is similar to the winning variation in the Garcia game, the a-pawn is ready to roll 62...Nb8! The White pieces are on unfortunate squares, Black threaten the fork Nc6+ followed by the capture of the b-pawn. 63.Kc4 63.b5 doesn't stop it 63...Nc6+! 64.bxc6 Kxc6!= 63...Nc6! 64.Kb5 Nxb4! 65.Kb6 The king reaches the corner after 65.Bxb4+ Kc7 66.Bd6+ Kb7!= 65...Nd3 1/2-1/2 In order to stop the Black king from getting to the corner and not allow the knight to capture the last pawn, the White king gets stuck in front of the pawn. 66.Bc3 Kd7 67.a5 (67.Kb7 Nc5+) 67...Kc8 68.Ka7 Nc5 69.Bd4 Na6 70.Be5 Nc7=

5/13/20 - BCE-62a, Breyer-Nyholm Baden, 1914

This week's BCE position is a pawn ending from the gambit tournament held in Baden 1914. All games were required to begin with a gambit (not including the Queens Gambit). Under such conditions it isn't surprising that the famed attacking player Rudolph Spielmann emerged on top with 12.5/18.

Guyla Breyer is well known to chess players for his contributions to the hypermodern school. His opponent in this game, Gustaf Nyholm, was one of the first great Swedish players, winning matches for the national championship six times between 1917 and 1921. The subject game started as an Evans Gambit, although Nyholm declined it. After an exchange of queens with 36.fxe4

Here, Nyholm made a very risky pawn sacrifice 36...Ke7 This move was severely criticized in Gregor Marco's Viennese chess journal Wiener Schachzeitung where it was given no less than 3 question marks. I think that is a bit extreme given that Black can still hold the draw after the text. Nevertheless, I agree that it is a high risk move that doesn't have much chance of reward, so the clearer path to the draw would be the line 36...c5 37.Nd3 c4 38.Nc5 Ke7 39.e5 f6 40.exf6+ Kxf6 41.Kf2 Bd5 42.g3 Ke5 37.Nxc6+ Kd6 38.Nd4 Ke5 39.Nxe6 fxe6 40.Kf2 Kxe4 41.Ke2 Kd4 42.Kd2 The starting position of BCE-62a. Black has regained his sacrificed pawn and has the more active king, but the outside passed c-pawn gives White the winning chances. 42...g5 42...h5 is the subject of the BCE correction. Marco indicates the text as the losing move, but that comes one move later. 43.c3+ Kc4? The queen ending should be drawn after 43...Ke4! 44.g4 Kf3 45.c4! e5 46.c5! Kf2! 47.c6 e4! 48.c7 e3+! 49.Kd3 e2! 50.c8Q e1Q! 44.g4 Kd5 45.Kd3 Kc5 46.Ke4 1-0

5/9/20 - TCEC Season 17 Superfinal, Games 7-8: Frankenstein-Dracula Variation

Stockfish took the lead with a win in Game 7 of the TCEC Superfinal. Games 7 and 8 featured a complex exchange sacrifice in the Vienna game. 1.e4 e5 2.Nc3 Nf6 3.Bc4 Nxe4 4.Qh5 Nd6 5.Bb3 White can avoid the coming madness with 5.Qxe5+ 5...Nc6 Likewise, Black can bail out with 5...Be7 and 6...0-0 6.Nb5 g6 7.Qf3 f5 8.Qd5

The starting position. Black has to cover f7 with his queen, which involves the loss of Ra8. Black's compensation is better development and more space 8...Qf6 Theory and practice have both favored 8...Qe7 which was Stockfish's choice as Black in Game 8. However, the database actually shows Black with a slight plus score after Qf6 and it is also the choice of all 3 engines on ChessBase's Let's Check including a version of Stockfish. On today's chess24 broadcast of the Online Nation's Cup, at around 2:20:00 Jan Gutafsson briefly talks about the variation in relation to the Firouzja-Aronian game calling it fun, but better for White. I did notice when he played the moves on the board, he played the queen to f6.

In game 8 after 8...Qf6 9.Nxc7+ Kd8 10.Nxa8 b6 Lc0 also made an unconventional retreat with its queen 11.Qd3 (The more frequently seen moves are 11.Nxb6 and 11.d3) It seems as maybe the engines have determined that White will end up losing time with his queen in any case, so they do it on their own terms. 9.Nxc7+ Kd8 10.Nxa8 b6 11.Qf3 Lane considers the line with Qf6 to be dubious and gives 11.Nxb6 axb6 12.d4 Nxd4 13.Nf3 Bb7 14.Qxd4 exd4 (White can't play into this line with the queen on e7 because this move would be check) 15.Bg5 claiming a more pleasant ending. In the few games in practice, Black has done fine here with 3.5/5, so it isn't that clear cut. Black has two bishops for compensation and there are not open lines for the rooks. The only high level game to reach this position continued in similar fashion, although White didn't spend the tempo to take on b6 11.d4 Nxd4 12.Nf3 Bb7 13.Qxd4 exd4 14.Bg5 Qxg5 15.Nxg5 Bxg2 16.Rg1 Bxa8 17.0-0-0 Bg7 was prematurely agreed drawn in Sulskis-Motylev 2000 Linares Open. 11...Nd4 I don't really grasp the point of this move rather than continuing to develop with 11...Bb7 12.Qh3 Based on how the game eventually continues, one may wonder about immediately sending the queen back home with 12.Qd1 The problem is that the long diagonal is still a bit soft, for example 12...Bb7 13.f3 Qh4+ 14.g3 Qh5 15.Kf2 Ne4+ 16.Kg2 Ng5 and White's position is falling apart. The text defends g2. 12...f4 another strange one. I would again prefer 12...Bb7. Other engines like 12...g5 to try to further harrass the White queen. 13.c3 Nc6 14.d3 h5 15.Qf3 g5 16.Qd1 Now, White has completely undeveloped, but Black has wasted a lot of time with his knight maneuver and still has to pick up Na8. Stockfish ended up converting the extra exchange in a long ending.

5/8/20 - Timmel-Bereolos, 2020 Land of the Sky

I've only been able to play one tournament so far this year, the annual Land of the Sky tournament in Asheville. In the opening round I played a rook ending against John Timmel that shared some similarity to Tsay-Ippolito. After 27.Ne3

27...Nxb2 When envisioning this position from afar, I had intended 27...b5 trying to keep up the pressure, but when we got here I saw that he could answer with 28.c4 and I didn't think Black had much. So instead, I won the pawn even though it should lead to a draw. At least my knowledge that Ippolito had managed to win his game gave me some confidence that there could be problems for White. 28.Rf2+!? Before capturing the knight, White gives Black a choice on which way to go with his king. 28...Kg8 This decision was partially driven by the fact that I knew from studying Tsay-Ippolito that one of the key ideas for White is to attack the kingside pawns before so Black passer is too far advanced. Instead, 28...Ke8 29.Rxb2 Rxe3 30.Rxb7 Rxc3 31.Rg7 g4 32.Rg5 Rc1+ (32..c4 33.Rxh5 Rc2 34.h4 gxh3 35.Rxh3 c3 36.g4 Kf7 (36...Rc1+ 37.Kg2 c2 38.Rc3 and the Black king has to defend against the g-pawn.) 37.Kf1 Kf6 38.Ke1 Kg5 39.Kd1 and White wins the c-pawn.) 33.Kf2 Rc2+ 34.Ke3 should be fine for White, but as in the game, White should not play 34.Kg1? when 34...c4 would be decisive. 29.Rxb2 Rxe3 30.Rxb7 Rxc3 31.a4 I thought the simplest was 31.Rb6 threating both Rxa6 and Rg6+ which shows that Black's Kg8 didn't really do what he intended. 31...Rc1+ 32.Kf2 Rc2+ 33.Ke3= g4 34.Rxa6 Rxh2 35.Rc6 Rg2 36.Kf4 31...Rc1+!? seeing if I could improve the rook's position before pushing the pawn 32.Kg2 Rc2+ 33.Kg1?! and it paid off. The h-pawn isn't really worth leaving the king trapped on the back rank. 33.Kf3 Rxh2 34.Ra7 picks up the a-pawn and White should draw without trouble. 33...c4 34.Ra7?! This move doesn't really do much. White should immediately get behind the pawn with 34.Rc7 The best try for Black seems to be 34...c3 35.h4 g4 as there are a couple of pitfalls for White 36.Kf1 It should be noted that there is no passive defense with the extra a-pawns on the board. (36.Kh1? Rc1+ 37.Kg2 c2-+) The Black king then comes over to d8 then up to d3 threatening to move the rook. To prevent this White would check on the d-file with his rook, which let's the Black king cross the c-file and pick up the a-pawn. 36...Kf8 (36...Rc1+ 37.Ke2 c2 38.Kd2 Rg1 39.Kxc2 Rxg3 40.Rc5 Ra3 41.Rxh5 Rxa4 42.Rg5+ and Black doesn't have a good way to make progress) 37.Ke1 Ke8 38.Rc5 (The immediate 38.Kd1? gives Black a critical tempo 38...Rd2+ 39.Kc1 Rd3 40.Rc5 Rxg3 41.Rxh5 Rg1+ 42.Kc2 g3 43.Rg5 g2 ) 38...Kd7 39.Kd1 Rd2+ 40.Kc1 Rd3 41.Kc2 Rxg3 42.Rxh5 Rg1 now this isn't check and White has enough time to construct a defense 43.Kxc3 g3 44.Rg5 g2 45.Kb2 34...c3

35.Rxa6? He's left it very late, but the draw was still to be had with a long series of only moves 35.Rc7! Rc1+ 36.Kf2! (36.Kg2? c2-+) 36...c2 37.Ke3! (37.Ke2? Rh1-+) 37...Re1+ 38.Kd2! Rh1 39.Kxc2! Rxh2+ 40.Kd3! Rh3 41.Ke4! Rxg3 42.Kf5! h4 43.Ra7! h3 44.Rxa6! h2 45.Rh6! Rg2 46.a5! g4 47.a6 g3 48.Kg4 Rg1 49.a7! 35...Rd2 0-1Black promotes his pawn with ...c2 and Rd1+

5/6/20 - BCE-453, Capablanca-Lasker, St. Petersburg 1914

This week's BCE correction is from the famous tournament at St. Petersburg 1914. Legend has it that Tsar Nicholas II first bestowed the title Grandmaster on the 5 finalists: Lasker, Capablanca, Alekhine, Tarrasch, and Marshall. The tournament was a great triumph for World Champion Lasker who managed to overcome a 1.5 point deficit to Capablana by scoring 7/8 in the double round robin finals to take home the 1200 ruble prize. From what I could find about conversion rates and inflation, that might be around $15,500 today.

Today's game is one of Lasker's two draws in the finals. Capablanca pushed for a long time in the ending, but could not overcome Lasker's defense. After 28...Rb2

Capablanca won two pieces for a rook 29.Rxb5 Rxb3 30.Bd2 Bc5+ 31.Rxc5 Nxc5 32.Nxc5 Rb2 33.Be3 The starting position of BCE-453 33...Re2 34.Bf2 f6 35.Kf1 Ra2 36.g4 Kf7 37.Ne4 h6 38.Kg2 Ra3 39.f4 Rb3 40.Ng3 Ra3 41.Nf1 Rd3 42.Ne3 Rc3 43.Kf3 Ra3 44.f5 Ra2 45.Nd5 Rb2 46.Nf4 Ra2 47.h4 Ra5 48.Bd4 Fine now uses the sequence 48.Ne6 Rb5 49.Ke4 Rb2 50.Bd4 Rb4 51.Kd5 Rb1 to skip ahead to the position after Black's 72nd move. 48...Ra3+ 49.Be3 Ra5 50.Nh5 Ra4 51.Ng3 Kg8 52.Ne4 Kf7 53.Bd2 Ra1 54.Bc3 Rf1+ 55.Nf2 Rc1 56.Bd4 Re1 57.Ne4 Rf1+ 58.Bf2 Ra1 59.Kf4 Ra4 60.Bc5 Rc4 61.Kf3 Rc1 62.Bf2 Ra1 63.Kf4 Ra4 64.Kf3 Ra3+ 65.Be3 Ra5 66.Nc5 Ra1 67.Ne6 Ra3 68.Ke4 Ra4+ 69.Bd4 Rb4 70.Kd3 Rb3+ 71.Ke4 Rb4 72.Kd5 Rb1 73.g5 hxg5 74.hxg5 fxg5

75.Nxg5+ 75.Nxg7 is the subject of the BCE correction. Karpov and Zaitsev analyzed the ending for the Encyclopedia of Chess Endings in Position 1007 and do mention the variation 75...g4? 76.Nh5! although they only give it !? 76...Re1 77.Ng3 with clear advantage to White. They attribute this line to Minev. Today, thanks to the tablebases, we know White is winning. 75...Kg8 This position is also analyzed in the Encylopedia as Position 997. It doesn't give an annotator, so I don't know if this was still Karpov and Zaitsev, or one of the editors. There are a few variations of interest in that analysis, which I quote below. 76.Ne6 Rd1 76...Rb7? 77.Bxg7! 77.Ke4 77.Nxg7 Rxd4+! 77...Kf7 78.Ng5+ 78.Bxg7 Re1+! 79.Kd5 Rf1! 78...Kg8 79.Ke5 Re1+ 80.Kf4 Rf1+ 81.Kg4 Rd1 82.Nf3 Rf1 83.Be5 Kf7 84.Kf4 Kg8 85.Ke4 Rd1 86.Ng5 Re1+ 87.Kd5 Rd1+ 88.Ke6 Re1 89.Nh3 Rb1 89...Rh1 90.Nf4 Rg1? 91.Nh5! g6 92.f6 +- 90.Nf4 Rb6+ 91.Ke7 Rb5 92.Ng6 Rb6 93.Bd6 Ra6 94.Ke6 Rb6 95.Ne7+ Kh7 96.Nc8 Ra6 97.Ne7 Rb6 98.Nd5 Ra6 99.Nc3 Kg8 100.Ne4 Rb6 1/2-1/2

5/3/20 - Tsay-Ippolito, 2019 US Masters

A few days ago, I heard the sad news that the 2020 US Masters is the latest event to be cancelled because of the coronavirus. However, we still have games from last year's event to analyze. An example with the rook in front of the extra outside passed pawn where White managed to win occurred in the game between Vincent Tsay and Dean Ippolito. The rook ending began after 34...Rxa2

35.Rxe6 Rxa4 36.Rc6 Ra2 37.Rxc7 a5 38.Ra7 An important difference from the BCE example is that the defending king is confined on the back rank. Still, I don't think this should be enough to be decisive. 38...a4 As we will see, g4 is a key move in White's defense, so Black could try to prevent it with 38...h5!? If White sits and does nothing, he will lose as in the game, but it appears he can switch to a side defense a la Vancura and hold 39.h3 a4 40.Re7 a3 41.Re3 with the key point being that the Black king is cut off from the queenside so White can just shuffle with Kh1-g1. The only way for Black to make further progress is 41...Ra1+ 42.Kg2 a2, but then 43.Ra3 is an easy draw. 39.Kg1 g5 Black could restrict the White rook with 39...a3 then if the Whtie rook leaves the a-file Black makes a queen with Rb2, a2, and Rb1+ Still White seems to have a path to a draw with 40.g4 g5 (Allowing the white pawns to advance by bringing the king over immediately doesn't seem to help 40...Kf8 41.h4 Ke8 42.Kf1 Kd8 43.Ke1 Kc8 44.Kd1 Kb8 45.Ra4 Kb7 46.Kc1 Kb6 47.h5=) 41.Kf1 Kf8 (The Black rook can check at various points as the White king journey's to the queenside. However, this eliminates the idea of Rb2/a2/Rb1+ so Black needs to follow up with ...a2 after the check in order to keep the White rook tied to the a-file, otherwise White can take the h-pawn with an easy draw. Checking when the king is on f1 doesn't help as the king hides on h3 41...Ra1+ 42.Kg2 a2 43.h3 and White doesn't have any weaknesses for the Black king to attack.) 42.Ke1 Ke8 (Likewise here, 42...Ra1+ 43.Kf2 a2 44.Kg2 and 45.h3) 43.Kd1 by now it is obvious that the White king is getting to the queenside first, so Black doesn't have anything better than trading a for h, but the resulting 2 vs. 1 position doesn't cause White trouble because the Black king is cut off. 43...Ra1+ 44.Kc2! a2 45.Kb2 Rh1 46.Kxa2! Rxh2+ 47.Kb3 Rh4 48.Kc3 h6 49.Ra6 Kf7 50.Kd3 Kg7 51.Ra4 h5 52.Ra7+! Kf6 53.gxh5 Kf5 54.Ke3= 40.h3 Perhaps the first step in the wrong direction. 40.g4 followed by bringing the king across draws similar to the above. Maybe White wasn't sure about the 2 vs. 1 positions so he avoids ...Rxa2 40...a3 41.Kh1? I guess White thought he had a fortress and could defend passively. There was still time for 41.g4 41...h5! Now, Black is winning.

42.Ra5 It's too late for 42.g4 Black fixes the weakness on h3 with 42...h4! 43.Kg1 Kf8 44.Kf1 Ke8 45.Ke1 Ra1+ now this check puts White on the wrong foot as the Black king wlll be close enough to the queenside so Black can keep the ...a2 move in reserve since White isn't in time to capture the g-pawn with his rook 46.Kf2 Kd8 47.Kg2 Kc8 48.Ra5 Kb7 49.Rxg5 Ra2+! 50.Kf3 Rh2! 51.Ra5 Rxh3+ 52.Kg2 Rg3+ 53.Kh2 Kb6 54.Ra8 Kb5 55.g5 Kb4 and Black wins. White can also try running the king to the queenside without pushing a kingside pawn, but thanks to the two tempi White squandered Black can win with very accurate play 42.Kg1 g4 43.hxg4 hxg4 44.Kf1 Ra1+ 45.Kg2 Kf8 46.Ra4 Ra2+! 47.Kf1 Ke7 48.Ke1 Kd6 49.Kd1 Ra1+! 50.Kc2 a2! 51.Kb2 Rg1! 52.Kxa2 Rxg3 53.Ra5 Ke6 54.Kb2 Kf6 55.Kc2 Re3! 56.Kd2 Re7 57.Rb5 g3 58.Rb3 Rg7! 59.Rb1 Kg5! 60.Ke2 Rf7! 61.Rf1 Rf4! -+ 42...Kf7 43.Ra6 Ke7 44.Kg1 Kd7 45.Kh1 This passive defense is hopeless because of the weakness on g3. Black slowly brings the king in to win g3. 45...Kc7 46.Kg1 Kb7 47.Ra4 Kb6 48.Ra8 Kb5 49.Kh1 Kb4 50.Rb8+ Kc4 51.Ra8 Kd3 52.Kg1 Ke4 53.Ra5 g4 54.Ra8 Kf3 55.hxg4 Kxg3 56.Kf1 hxg4 57.Ra5 Ra1+ 58.Ke2 Kg2 59.Ra8 g3 60.Ra7 a2 0-1

5/2/20 - TCEC Season 17 Superfinal, Games 5-6: Kings Indian Mar del Plata

In games 5 and 6 of the TCEC Superfinal, Lc0 and Stockfish debated one of the big main lines of the Classical Kings Indian. 1.d4 Nf6 2.c4 g6 3.Nc3 Bg7 4.e4 d6 5.Nf3 0-0 6.Be2 e5 7.0-0 Nc6 8.d5 Ne7 9.Ne1 Nd7 10.Nd3 f5 11.Bd2 Nf6 12.f3 f4

The starting position for this pair of games. Back in the early days of computer chess the engines didn't handle these types of closed positions very well. They would often realize the danger in Black's kingside attack far too late. Of course, they have much improved in that area. 13.c5 In the game Stockfish was White, it went for 13.a4 which has only been played at high level in correspondence chess, perhaps showing the influence engines have had there. That game did return to somewhat familar waters after 13...g5 14.b4 h5 15.a5 Now the standard regrouping is Rf7, Bf8, and Rg7. Lc0 instead went for 15...Kh8 16.c5 Rg8 I don't really understand why this should be an improvement as the bishop will still need to move to clear the g-file and from g7 the rook can also help in the defense of the queenside by protecting c7. 13...g5 14.Rc1 Ng6 15.Nb5 Older books by Nunn and Gallagher argued that White had nothing better than to insert 15.cxd6 at this point 15...Rf7 The point of White's previous move is that he can sacrifice a piece after 15...a6 16.cxd6 axb5 17.dxc7 which Gallagher called speculative back in 2004. In practice, White has scored heavily in the database (+12 =4 -1) 16.Ba5 b6 17.cxd6 cxd6 18.Be1 a6 19.Nc3 a5 20.a4 This move was introduced by Giri in his blindfold game against Grischuk at the 2011 Amber tournament. I don't fully understand the point, especially in conjunction with White's next move. Bf8 is generally part of Black's plans and a4 gives him the time to execute it. Compare this to the immediate 20.Nb5 and if we follow the course of the game 20...g4 21.fxg4 (Kotronias only looks at 21.Rc6) 21...Nxe4 22.Rc4 Nf6 23.Bf3 e4 24.Bxe4 Bxg4 25.Bf3 Bxf3 26.Qxf3 Ne5 27.Nxe5 dxe5 28.d6 Rc8 White has 29.Rc7 with some pressure. 20...Bf8 21.Nb5 Giri played 21.Nf2

21...g4 Technically, the novelty in this game, but the idea is known in similar positions. In a couple of games that had reached this point, Black prepared the advance with 21...h5 and was successful in both games 22.fxg4 Nxe4 23.Rc4 Nf6 24.Bf3 e4 25.Bxe4 Bxg4 26.Bf3 Bxf3 27.Qxf3 Ne5 28.Nxe5 dxe5 29.d6 Rc8 30.Rc6 now on 30.Rc7 Black can take on d6 30...Bxd6 with the tactical justification 31.Rxc8 Qxc8 32.Nxd6 Qc5+ 30...Rxc6 31.Qxc6 Nd7 Black has stabilized the queenside situation. Stockfish eventually reached an ending with 2 pieces against a rook that I thought might have some winning chances, but Lc0 held the draw.