Analysis
The game in detail
GM Sebastian Bogner (SUI, 2542) vs GM Andrei Sokolov (FRA, 2448) · 1-0
Board 2, Genève vs ASK Réti · Queen’s Indian Defence, Kasparov-Petrosian Variation (E12) · 90 minutes for 40 moves, then 30 minutes, plus 30 seconds per move
By Bogner AI Coach AI Coach . This analysis was written by a Bogner Chess AI coach.
Replay the game on LichessBoard two, Genève against ASK Réti, round nine and the last of the season, all five matches of the round in one hall in Geneva. Across the board sits Andrei Sokolov, a grandmaster since 1984, seven years before the man opposite him was born, 2448 on the list today. Réti went into this final weekend fourth and lost round eight; four and a half board points win the match, and on six of the eight boards the higher-rated player sits opposite. Board two is one of the two where it runs the other way: a board the team is not waiting for a half point from.
It became a whole point in forty-nine moves, and the road there was not straight: after Black’s twenty-second move White stood two pawns better, three moves later it was two thirds of a pawn, and from move thirty everything hung on one single move. After two rounds with White that each produced a half point when there had been more in them, this is the first whole one of the weekend.
Sixteen moves, and twenty-eight minutes in hand
. Sixteen moves of the Queen’s Indian, and not one of them cost more than an eighth of a pawn. After ten moves Bogner had 1:32:05 on his clock, more than he started with; after it read 1:29:15 against 1:01:32.
is the decision worth keeping: three quarters of the players at this strength recapture with the queen, and that costs a good third of a pawn here, because the rook on d3 is the piece that later reaches the black king along the third rank and the g-file. On move twenty-four it does exactly that. and belong together: the bishop leaves g5, where it was holding nothing, and steps in behind the pawn that wants to go to h5. cost twelve minutes forty-one and was the best move. puts the knight on the road to g4, or to e3 by way of f1.
Then the position grew from three quarters of a pawn to two pawns in two moves, without Sokolov making a move anyone would have to call an error. cost two thirds of a pawn, another whole one, and is entirely human: the king leaves the square where a rook check on g3 would catch it after a sacrifice on h6. His best move would have allowed that sacrifice, and allowed it on purpose: , where White has given a bishop for a pawn and stands four fifths of a pawn better. There was no more in the position than that. Space in the centre is not a thing you have to cash in; it is permission to work on a wing while the opponent has to answer a question every move. After sixteen quiet moves Sokolov had spent twenty-eight minutes more, and that is the same advantage in a different currency.
GM Andrei Sokolov (FRA, 2448)
GM Sebastian Bogner (SUI, 2542)
Twenty-six minutes for 23.Qg4, six for 25.Qxg6
cost twenty-six minutes fifty-two, with 28:01 left after it, and it is the longest think of the game. It puts the queen on the square the knight wanted. The road was : the knight arrives on e3, the pawn on f3 takes the square away from the black queen, and White stands a good two pawns better. is almost as good. costs about a pawn and leaves a good one. Half the remaining thinking time went into one decision, and it landed on the wrong square.
then came in five seconds and is the best move; it had been part of the calculation already. Sokolov needed eleven minutes six for and four more for , and with he offers the queen trade.
after six minutes twenty-seven is the error of the game. It costs a pawn and a fifth and turns a position with an attack into a position with an extra pawn. What holds it together is : . The h4 pawn goes in that line, which is the price; in return the knight stands on e3, the rook on the h-file, and White almost two pawns better. Objectively is bad, and in practical terms too: nearly two thirds of the players at this strength keep the queen here. What explains it is the clock, 22:29 after the move and fifteen moves to the control. A queen trade takes everything out of a position that could go wrong, the attack included. Before trading queens, count the attackers rather than the material; with two pieces working on the enemy king, trading one of them hands half of it back.
GM Andrei Sokolov (FRA, 2448)
GM Sebastian Bogner (SUI, 2542)
From 22:32 down to 8:43
After the material was level again and the advantage half a pawn. It stood there for five moves, and those five moves are the part of the game that mattered: , and took the clock from 22:32 to 8:43, and the position held through every one of them.
is the expensive one, twelve minutes thirty-three, with 8:56 left after it, and it is the best move in the position. The rook goes back from g6 to g4 so that the second one can come from d3 to g3. What the two of them do together stands on g7: the pawn there is defended by the king on f7 and by nothing else, and when two rooks line up one behind the other, the front one cannot be taken.
In the same stretch Sokolov played , five minutes thirty-nine, and in thirty-four seconds. After he was better on the clock, 14:09 against 8:43, half a pawn worse on the board, and his bishop on c1 was pointing at f4, which is to say at exactly the piece whose exchange pulls the battery apart. A position that stays level for five moves is not quiet, it is loaded: spending the time in it to place your pieces so that one careless move by the opponent is enough is time spent well.
GM Andrei Sokolov (FRA, 2448)
GM Sebastian Bogner (SUI, 2542)
30...a5, and 31.Rxg7+
cost eight minutes twenty-five and left Sokolov on 6:14. It is the move that decided the game, and on the board it does nothing. What had to be played was , and precisely because it gives up Black’s best piece: , material level, White half a pawn better, no battery left. With the bishop on f4 the rook on g3 is attacked, and an attacked rook no longer covers g7.
After , though, is the only move, and the gap to the next one is almost three pawns. The rook lands on a square its own rook on g3 covers, so the king cannot take it; he has to go to e8, and the rook stays on the seventh rank. White is a pawn up, has a rook cutting the black king off from his own queenside, and stands two and a half pawns better than two moves earlier. Two minutes fifty-seven, with 6:16 left: this is the move the twelve minutes on move twenty-nine were for.
Nothing before this is Sokolov’s fault, and is not much of one either: it is a move at the far end of the board in a position where there was one place to watch. That is the rule in it. When two enemy rooks stand one behind the other on the file of the pawn in front of your own king, the question is not which move you would like to make but which one dissolves the battery. Here there was one, and it looked like a concession.
GM Andrei Sokolov (FRA, 2448)
GM Sebastian Bogner (SUI, 2542)
32.Be3, and the choice on e3
after thirty-eight seconds is the move almost nobody plays at this level. Four in ten take with , and after that there is nothing left: , and the position is level, because the check on the back rank forces the knight to f1, where it stands still. offers the same trade in the form where a pawn can do the recapturing.
After one choice was left, and Bogner made it in one second: , and it is worth the whole game. is exactly level, and nearly four in ten players at this strength take with the rook. The difference is who answers the check: taken with the pawn, the king answers it, , and both rooks stay where they are working; taken with the rook, the knight has to answer it, and it stands still on f1 while the pawn on a5 sets off. With two recaptures, what counts is not which piece ends up better placed but which piece is still needed for the defence afterwards.
The rest is technique with two holes worth naming. after four minutes fifty-seven, with 2:21 left before the time control, costs four fifths of a pawn; was the move, and is the one eight in ten players make in that situation. after eleven minutes forty-four costs about a pawn in a position winning by five and a half: collects the rook for a pawn. Neither changed the result. came two moves later anyway, won the knight, and followed by won the rook. In a won position the most accurate move is not the most important one; the most important one is the move that pushes the passed pawn, and was both.
GM Andrei Sokolov (FRA, 2448)
GM Sebastian Bogner (SUI, 2542)
Verdict
A whole point on board two against a grandmaster who has been one since 1984, in the last round of the season. The honest reading is that the advantage was given away twice and taken back once: after twenty-two moves there were two pawns on the credit side, after twenty-six minutes fifty-two cost the first and two moves later cost the second, and after the material was level and the advantage half a pawn. That it still became a win is down to the five moves that followed: , and put two rooks one behind the other on the g-file and made g7 the one square where a careless move is enough, and they took the clock from 22:32 to 8:43 to do it. was that move, the only one that punishes it, almost three pawns clear of everything else, found in two minutes fifty-seven with six minutes left. Credit to Sokolov: up to move thirty he made no move anyone would have to call an error, and is a move you overlook because it hands over your own best piece. It is the same lesson as in round 8 at a different move: there an exchange dissolved an advantage that was not yet due, here dissolved an attack that had only just been finished. For Bogner the point is proof that a game in which two pawns of advantage go missing is still winnable, as long as the pieces keep pointing at the square where it hurts; for Réti it is the point from the board the team had counted on for it.
Moves
1-0
Was this analysis useful?