Alekhine -- Levenfish 1912

Reading Alexander Alekhine's Best Games (1996) this morning, I became caught up studying a miniature. Alekhine -- Levenfish, St. Petersburg 1912 was decided in nineteen moves. Naturally, Levenfish's errors merit attention for anyone who plays the Benoni Defense, and perhaps also for players of the Modern.

After 14...Qxb2

Alekhine,Alexander -- Levenfish,Grigory [A43]
St Petersburg Winter-B St Petersburg, 1912

1.d4 c5

Alekhine criticizes this move, claiming, "White at once obtains a great positional advantage by simply advancing the centre pawns."

2.d5 Nf6 3.Nc3 d6 4.e4 g6 5.f4 Nbd7?!

5...Bg7 is the normal move.

6.Nf3 a6?!

With this move, this game becomes unique in the database.

6...Bg7 7.e5 dxe5 8.fxe5 Ng4 9.e6 fxe6 has been played at least eleven times. Alekhine gives this line to 9.e6, but then has 9...Nde5 10.Bb5+. His line has been played at least twelve times with ten White wins. It seems that 9...fxe6 may be better, although here, too, White has done well.

White to move

7.e5

White already has a clear advantage, according to Branko Tadic, and Goran Arsovic, Encyclopedia of Chess Miniatures (2015), where this game is number 166. Irving Chernev, The 1000 Best Short Games of Chess (1955) has it as well, but the annotations are limited to the last two moves. Tadic and Arsovic mark Black's fifth and sixth moves as dubious. Anyone seeking to play this line as Black would be well to note the urgency of playing Bg7 straight away.

7...dxe5 8.fxe5 Ng4 9.e6

Searching positions with 6...Bg7 in the ChessBase database this morning brought up several games that reached the position via a move order from the Modern Defense. White did well in those games, too, and this e5-e6 thrust was frequently played in those games.

9...Nde5 10.Bf4

Black to move

10...Nxf3+

Black might have tried 10...Bg7. Here Tadic and Arsovic offer 11.Nxe5 Nxe5 12.Qe2 with advantage for White. For his part, Alekhine offers 11.Qe2 Nxf3+ 12.gxf3 Nf6 13.exf7+ Kxf7 14.O-O-O "with an overwhelming advantage for White." John Nunn, who converted Alekhine's games to algebraic and culled from the two volumes of My Best Games to produce Alexander Alekhine's Best Games, suggests an improvement for Black in the line Alekhine gives. Instead of 12...Nf6, Nunn recommends 12...Bxc3+ 13.bxc3 Qxd5 as "more testing". But even here, White gets a strong attack with 14.fxg4! Qxh1 15.O-O-O Qc6 16.exf7+ Kxf7 17.Bg2.

11.gxf3 Nf6 12.Bc4 fxe6

While my coffee was still hot this morning, I spent a little time looking at 12...b5 13.Nxb5 axb5 14.Bxb5+. White ends up a pawn ahead with a strong position.

13.dxe6

Black to move

13...Qb6

Alekhine presents the alternative 13...Qxd1+ 14.Rxd1 Bg7 15.Bc7 O-O 16.Bb6, where, "White wins a pawn, at the same time maintaining all his pressure."

13...Qa5 might be playable, although White still has an edge.

The computer likes Black's move until it sees Alekhine's brilliant fifteenth move.

14.Qe2! Qxb2?

Tadic and Arsovic give 14...Bg7 15.O-O-O with a clear advantage for White. Certainly, Black's last chance was to resist the poisoned pawn.

15.Nb5!!+-

Black to move

Alekhine's double rook sacrifice had to be calculated before playing 14.Qe2. Black, too, needed to see the consequences in order to avoid 13...Qb6

15... Qxa1+

Perhaps Black can survive with 15...Bg7 16.O-O-O O-O 17.Bd6!

Now, we have finish that is reminiscent of Anderssen's Immortal Game.

16.Kf2 Qxh1 17.Nc7+ Kd8 18.Qd2+ Bd7 19.exd7 1–0

Black can delay, but no longer prevent checkmate.

Determination

Reading Chess Informant

Last Thursday, Chess Informant 128 arrived via courier. With each new issue, I resolve to read all the articles or play through all of the games in the traditional section (see "Discovery"). With each new issue, I fall short of my goals.

I am determined to establish a new pattern, reading all of Informant 128.

Since Thursday, I have read all of the narrative text in Ernesto Inarkiev, "Moscow Candidates Tournament 2016"; Aleksandar Colovic, "Moscow Opening Report"; and Sarunas Sulskis, "European Championship 2016". I also solved rather quickly the first ending and was working on the second when I was interrupted by a text message that required action. Using the CD version that comes with the printed book, I have been working through the games and analysis in Inarkiev's article.

Black to move

Sergey Karjakin played 17...cxb4.

Inarkiev's comment brought my reading to a halt. He writes, "This is a tempting decision. Black sacrifices his queen for a rook, minor piece and a pawn, a balance of material that is usually sufficient" (18). My eye did not perceive the danger to the queen until I saw 18.axb4 Bxc3 19.Nc6 on the board, and only then could I begin to calculate the resulting imbalance.

Karjakin's determination at the Candidates shines in Inarkiev's narrative and analysis. The depth of the analysis and the complexity of the games reveals to me that my determination to read all of the issue clear through will require a significant investment of time.

My reading of Informant goes back two decades. The quality of my play has benefited. On this blog, I wrote about the first tangible benefit--a correspondence win in the 1990s--in "Playing by the Book" (March 2011). In a more recent correspondence game, I had this position with White.

White to move

Our game had been following Botsari -- Polgar,J, Eretria 2011, which Polgar won. Botsari played 28.Re7 and Polgar's pieces put an end to White's initiative. Apparently unbeknownst to my opponent, however, was the analysis of this game in Informant 112/51 by Dejan Antic who recommended 28.f7. I played that move and our game followed Antic's line 28...Rg7 29.Re8+ Nf8 30.Qd6 and then instead of the necessary 30...Qh4, my opponent played 30...Qh5+ and I won a few moves later.


Improving

If I continue to read Informant in bits and pieces and utilize my complete collection in electronic format as a reference in correspondence games, I will continue to see benefits. But I want more. I remain determined to improve my performance in USCF tournaments, reverse the slow downward slide that has been occurring since summer 2012, and renew efforts to push to a new peak rating (see "Eleven Consecutive Wins!" [July 2012]). I believe that Informant is among the most useful tools that I have for improving.

In the wake of my tournament success in 2012, I purchased Informant 113 as book and CD combination. Prior to that point, I acquired most of my Informant books after they went on sale because they were a year or more out of date, and acquired even more electronically via the CDs that had a year's worth. Gaps in my collection were filled when I placed fifth in the Reader's Contest for those who purchased Best of the Best 1000 (2008). When I bought Informant 113, I had a complete collection through the current issue. Several months later, when my collection was no longer current, I began systematically working through the games section (see "Training Log: August 2013"). That enterprise lasted several months, but eventually gave way other training priorities and my recurring obsession with blitz.

I bought a few more multi-issue CDs in 2014. Then, with download editions of Informants 119, 120, and 121, I was up-to-date with the latest issues. These were followed with timely downloads of Informants 122 and 123, keeping me current. Beginning with Informant 124, I became a subscriber. Each issue becomes the center of my chess focus for a few weeks after its arrival, but I never get through more than a fraction of the issue. I skim several articles and play through the games that interest me. Several printed issues contain brief notes that I make as I go through the games on the screen of my computer.

I am determined that with Informant 128 I will maintain the focus and self-discipline to work through every article, the games section, and the endings and combinations sections. There will be distractions. Since Thursday, I have played dozens of blitz games online. I also remain committed to study of Greco's and Morphy's games. I use these with my students. Time spent reading Informant must balance with study of classics.

The time and effort spent on Informant will improve the quality of my play.

One Good Game

As a blitz addict, I find innumerable motives for wasting endless hours in futile chess play. Sometimes I can rationalize my binges with a few instructive positions that I may show my students. Indeed, a young student whom I've been coaching the past year started individual lessons yesterday, and among our concerns in the lesson was cultivating his understanding of pawn majorities and king position in the endgame. We looked at this position played in the wee hours of the morning in online blitz.

White to move

This morning, I started played some blitz on a site where I seem to care about rating. Losing the first game to an underrated "cheat"* meant that I could not stop after one game. My second game was a positional and tactical crush of a slightly higher rated player, but I gained less than the first game had taken from me. My opponent was down three pieces for three pawns at move 34 but did not resign until one move from checkmate thirty moves later.

Then, I lost again. Then, another game where my opponent squandered a three pawn advantage in a rook ending to reach a theoretically drawn position that I lost on time. My fifth game was a twelve move win against the same opponent.

Then, I won another miniature with a classic checkmate sequence.

Black to move

After 17.Bh6??

I made my move and spoke aloud, "take my rook." Doing so, of course, is suicidal. My opponent took the rook and fell to a checkmate in three.

I was able to stop the binge after this game.


*Suspicions of cheating dwarf actual instances of unfair play. In the blitz addict's mind, every untitled player who beats him must therefore be cheating in some manner. Such irrational thinking sometimes renders a game that should be entertaining and even beneficial something only slightly less damaging to healthy existence than substance abuse. Happily, these suspicions are held with a sense of irony. I use the term with full knowledge that it is rooted in paranoid fantasies concerning the extent of my own skill and therefore the extraordinary means that must be taken to defeat me. On the other hand, having analyzed with the computer many thousands of blitz games, I realize that my own pitiful play is the sole cause of most losses and indeed mars even most of my best wins.

Battle in the Benoni

Jim Maki Annotates

FIDE Master Jim Maki sent me this game for posting. His opponent is a rapidly improving high school student who is getting some coaching from Maki. I have played Travis Miller once. I lost. It is one of very few losses in the past ten years to underrated youth players.

Miller,Travis (1819) -- Maki,Jim (2318) [A70]
Spokane Rapid G/20 +3, 29.10.2015

Annotations by Jim Maki

17 year old Travis Miller, a recent arrival from Alaska to the Spokane area, has been making great progress since moving here as evidenced by his multiple 1st place finishes in Open events. This game was 3rd round of the Spokane Chess Club's Game/20 +3 tournament held recently.

1.d4 Nf6 2.c4 e6 3.Nf3 c5 4.d5 exd5 5.cxd5 d6 6.Nc3 g6

White to move

7.Bf4

I had recently shown Travis a line I was a bit worried about in the Benoni so I thought he might have come up with a surprise. Here is the line: 7.Nd2 Bg7 8.Nc4 0–0 9.Bf4 Ne8 10.e3 Bxc3+ 11.bxc3 b5 12.Nd2 g5 13.Bg3 f5 with a very complex and hard to play position from both sides.

7...a6 8.a4 Bg7 9.e4 0–0 10.Bd3 

Usually white plays Bd3 with h3 so I just thought Travis was mixing up systems. I had two chances to play Bg4 to equalize but decided it was too gutless. But sometimes gutless chess is good chess.

10...Qe7 11.0–0 Nh5 

Or the simple: 11...Bg4 12.h3 Bxf3 13.Qxf3 Nh5 14.Be3 Nd7=.

12.Bg5 Bf6 13.Be3

The first real suprise. I was expecting 13.Bh6

13...Nd7

Here I really should have gone for 13...Bg4. Time to get into trouble.

14.Nd2

Black to move

14...Ne5?

Very bad indeed. This comes from playing too much blitz. Better would be 14... Rb8 or: 14...Bd4 15.Bxd4 cxd4 16.Ne2 Nc5 17.Qc2 f5 and if 18.Nxd4 Nf4 19.Bc4 fxe4 and black has good activity.

15.Be2 Ng7 16.f4 Nd7 17.Nc4 Re8 18.Re1

18.Bf3±

18...Rb8 19.Bf3

Yikes! This is known as the "looks like I'm dead lost" variation of the Benoni. That white center is a tidal wave about to come ashore.

Black to move

19...Qf8 20.e5 dxe5 21.Ne4 exf4?

Now I'm just taking stuff hoping he doesn't find the hammer. Hanging by a thread is the move 21... Be7.

White to move

22.Bxc5?

This looks so good at first sight. This is, after all, a 20 minute game and we are both getting into some nasty time trouble. Winning is: 22.Nxf6+ Nxf6 23.Bxf4 Rxe1+ 24.Qxe1

a) 24...Bg4 25.Bxb8 Qxb8 26.Bxg4 Nxg4 27.d6 b5 28.axb5 axb5 29.Qe4+- The 30. Ra8 threat is fatal.

b) 24...Ra8 25.Bd6 Qe8 26.Be7 Nfh5 (26...Nd7 27.Nd6) 27.Nb6 Ra7 28.Bxc5+- Material is even but black's pieces are so bad that random moves win for white.

22...Be7

The only move but good enough. Black has life again.

23.d6 Nxc5 24.Nxc5 Bf6 25.Rxe8 Nxe8 26.Bxb7?

Black to move

Hard to believe but black is now winning.

26...Bxb7

Much better is: 26...Rxb7 27.Nxb7 Bxb7 28.d7 Ng7–+.

27.Nd7 Qg7?

Now white is winning again.

27...Qh6 28.Nxb8 Qg5-/+.

28.Nxb8 Bd4+ 29.Kh1 f3

White to move

I thought I was winning here. Travis had less than 30 seconds on his clock but he finds:

30.Qxf3! Nxd6

Can't take the queen; 31.d7 wins instantly.

31.Qb3? 

31.Nc6+-.

31...Bd5?

Every move a blunder but I have 30 seconds and Travis has 16. Winning is: 31...Ne4 32.Nd7 (32.Qxb7?? Ng3+ 33.hxg3 Qh6#) 32...Qh6 I saw this far but thought here white could play 33. h3 but completely missed: 33.h3

Black to move

Analysis after 33.h3

33...Qf4–+.

32.Re1?

32.Nd7 Ba7 33.Rd1+-.

32...Qf6 33.Nd7 Bxg2+

The natural tendency in extreme time trouble is to play forcing moves like this where I know I can always bail out with a perpetual. But 33...Qd8 is better.

34.Kxg2 Qf2+ 35.Kh3 Qf5+ 36.Kg2 Qf2+ 37.Kh3 ½–½

Black to move

With flags hanging, Travis offers a draw so I take it. So what happens after 37...Qxe1?

38.Nxd6 This is why I took the draw. Black is a piece down, mate threats all over the place, and it seems black will be lucky to get a perpetual. But black is winning. 38...Qf1+ 39.Kh4 (39.Kg3 Bf2+ 40.Kg4 h5+ 41.Kg5 Bh4+ 42.Kxh4 Qf4+ 43.Kh3 Qg4#) 39...Bf2+ 40.Kg5 h6+ 41.Kf4 (41.Kxh6 Qc1+ and mate next move.)

Black to move

Analysis after 41.Kf4

41...Be3+!! 42.Ke5 (42.Kxe3 Qh3+) 42...Qf4+ 43.Kd5 Qf3+ 44.Kc4 Qc6+ 45.Kd3 Qxd6+ 46.Kxe3 Qxd7 and black is better but imagine playing this position on just the 3 second delay.

A Better Day

Taking Care of Business

After losing my round two game and taking my usual round three bye, I started Sunday morning with 1.5/3 in the Eastern Washington Open. My round four opponent is returning to chess after fifty years away and is much stronger than his rating suggests. Even so, he missed a tactical finesse that gave me an opportunity.

Black to move

In round five, I had my third Black in four games. I spent some time with the tournament director looking at the pairings, the logic, and the alternate possibilities. Due to a large number of byes and several upsets by the oldest and youngest players in the event, someone had to forego the usual color equalization. My opponent should not have three Blacks in a row, so he had White.

While I played something resembling the Czech Benoni in round four, I opted for a more normal Modern Benoni in round five. When my opponent played the thematic e4-e5 push, I spent nine minutes working out the best response. Prior to that I had averaged one minute per move.

Black to move

Would you play these two positions the way I did?

I finished the event with 3.5/5 and tied with a bunch of players in third place, sharing the A Class prize with two others. The only Expert went 5-0. All of the A Class players suffered a loss, some more than one. One of my students took out two A Class players on Saturday. The young woman who beat me on Saturday beat another A Class player in round five.

From the Benoni

I had this position a couple of months ago in a correspondence game. The computer says that I squandered an advantage. How should White play?

White to move

Planning, Time, Strategy

Lesson of the Week

Some of my advanced students this week are going through a game and annotations concerned with planning. The game was annotated by the winner and included in a chapter that he added to a book that he translated into English: Paul Keres, and Alexander Kotov, The Art of the Middlegame, trans. H. Golombek (1964). The annotations below are as they appear in the book, except that I have converted descriptive notation to algebraic, and have omitted some passages.

The diagram appears in the book. I start with it, asking students how they would play as White. Then, we go through the game from the beginning. Because I think Golombek's comments are instructive, I add very little.


Golombek,Harry -- Puig Pulido,Pedro [A65]
Varna ol (Men) fin-B Varna (7), 1962
[Golombek,Harry]

This game is an example of "fits-and-starts policy": "start on one plan, switch over to another that seems more attractive, and then, when it is too late, try to return to the original plan."

1.d4 Nf6 2.c4 g6 3.Nc3 Bg7 4.e4 0–0 5.f3 c5

Already the nature of Black's plan has defined itself. He is prepared to allow White a certain pawn preponderance in the center provided that in return he is able to concentrate on a counter-attack on the queen's side: an excellent plan that has proved its worth in many a modern game and one that would work here too--always provided Black sticks to his plan.

6.d5 d6 7.Bd3 e6 8.Nge2 exd5 9.cxd5

Now White's plan is clear. He intends to use his pawn majority in the center to make a thrust there; hence he recaptures with the c-pawn rather than the e-pawn so as to have in reserve the eventual advance of e4-e5.

9...Na6 10.0–0 Nb4

The first change in plan. Black, lured away by the possibility of attacking White's king bishop, neglects to follow the logical line of counter-attack on the queen's side. He should have played 10...Nc7 11.-- with the idea  11...a6 12.-- Rb8 13.-- b5.

11.Bc4 Re8 12.a3 Na6 13.Bg5

White has not only mere development in mind with this move. If Black drives away the bishop by h6 and g5 then he will have weakened his King's side and driven the bishop to a post from which it can aid the central pawn thrust of e5. There is also a more insidious notion in the move--it is designed to induce Black to change his plan yet once more.

13...Qb6

And Black does exactly this; he forms a fresh plan with the idea of unpinning himself and then manuevering the queen knight to d7 so as to hold back the advance of White's e-pawn. That this plan utterly fails is due to the waste of time caused by Black's constant change of plan.

14.Qd2 Nb8 15.Ng3 Nbd7 16.f4 h6 17.Bh4 Ng4 18.Rae1

All part of the plan of central advance.

18...Ngf6 19.Kh1 Nh7

White to move

With this move Black deems that he has adequately guarded e5 and prevented White's central advance. But now comes the logical follow-up of White's plan.

20.e5! g5 

Desperation; but what else can Black do?

20...dxe5 21.d6 -- 22.Nd5.

21.fxg5 hxg5 22.e6! 

A good illustration of White's theme--the central pawn advance. The plan has won through and it only remains to gather the fruits.

22...f6 23.Bb5 gxh4 24.Nf5 Bf8 25.exd7 Rxe1 26.Rxe1 Bxd7 27.Bxd7 Rd8 28.Be6+ Kh8 29.Nxh4 1-0

Black resigns. A case of too many plans spoiling the broth.

Lesson of the Week

In this position, I have misplayed the Benoni Defense and offered my opponent the opportunity to gain a decisive advantage. This game was played on US Chess Live in 2001. I found it looking through my databases for examples of certain common pins.

Opponent -- Stripes
USCL 13.01.2001

1.d4 Nf6 2.Nf3 c5 3.d5 e6 4.c4 exd5 5.cxd5 d6 6.Nc3 g6 7.e4 Bg7 8.Bg5 Bg4 9.Bb5+ Nbd7 10.h3 Bxf3 11.Qxf3 O-O

White to move

11.O-O

White missed the win, and after 12...Ne5 Black began to solve his self-imposed problems and went on to win.

Find the move that White should have played. As a bonus for readers of this blog, identify Black's critical error leading to the diagram position.

Busting the Benoni

The second strongest opponent that I have defeated on Chess.com is rated more than 200 Elo above me.* We met for two games in an ambitiously titled tournament created by a member who is no longer on the site. It's called the chess.com championships, but has no official status as such. In our first game, he outmaneuvered me in the middlegame of a French Tarrasch when I was playing for a draw. In order to remain in the tournament and advance to the next round, I needed a win with White.

Stripes,J (2121) -- Internet Opponent (2325) [A67]
chess.com championships - Round 3 Chess.com, 28.04.2013

1.d4 Nf6 2.c4 c5 3.d5 e6 4.Nc3 exd5 5.cxd5 d6 6.e4 g6 7.f4 Bg7 8.Bb5+

Black to move

When I was playing the Benoni as often as possible several years ago, the Flick-Knife Attack gave me the most trouble.

8...Nfd7

Some of the trouble that I faced was due to playing 8...Nbd7 here. In consequence of playing the wrong knight, White's thematic e4-e5 comes immediately.

9.Nf3!?

I opted for the third most popular move here, in part to test the level of my opponent's preparation. In The Modern Benoni (1994), David Norwood alleges that this move is purposeless as it permits Black to go through with the usual queenside expansion. Norwood asserts that 9.a4 is the correct move.

9...a6 10.Bd3 b5 11.0–0 0–0 12.Kh1 Re8

White to move

This position appears a mere 72 times in ChessBase Online, and yet it should seem to be the most likely position after White's uncommon 9.Nf3. Indeed, this position is found in line 4 of A67 in the Encyclopedia of Chess Openings. The principal move given there is 13.Qe1. Reference games are presented also for 13.Be3, which is not recommended, and 13.Bc2.

13.f5

Before I played my ninth move, I had looked forward in the databases and had gone through several games that reached this position. These games were in the spirit of an all out attack on the kingside before Black could finish untangling his pieces on the queenside--the drawback to Black's necessary eighth move.

13.f5 appears in Chess Informant 107 for the first time in three games played by Argentine GM Fernando Peralta. I looked at these games while playing. This research aspect is one of the pleasures of correspondence chess.

13...Nf6

I remember preparing for 13...c4, but I lost the notes to this preparation when my hard drive crashed. Perhaps my plans included 14.Bc2 Nc5 15.Ng5! Ra7 16.Qf3 1–0 L'Ami,A (2362) -- Valenti,G (2204) Reykjavik 2013

14.fxg6

At this point in the game, I was spending a lot of time going through a handful of games in the database. I concluded from this study that this move was more dangerous than the more common 14.Bg5, which had been Peralta's choice. 14.Bg5 leaves Black choices for how to meet White's attack.

Reference game: 14.Bg5 Nbd7 15.Qd2 b4 16.Ne2 c4 17.Bc2 Nc5 18.Ng3 Qc7 19.Rae1 Nfd7 20.e5 dxe5 21.d6 Qc6 22.Be7 b3 23.axb3 cxb3 24.Bb1 Bb7 25.Ne4 Rac8 26.f6 Bh8 27.Qh6 Ne6 28.Rc1 Qd5 29.Rcd1 Qc6 30.Rc1 Qd5 31.Rcd1 Qc6 32.Rf2 Qb6 33.Nfg5 Nef8 34.Nxf7 Bxe4 35.Qxf8+ 1–0 Peralta,F (2557) -- Almeida Quintana,O (2542) Barcelona 2009.

14...hxg6 15.e5!

Black to move

At this point we are following a single reference game from 1999. This move gives up a pawn to maintain the attack.

15...dxe5 16.Ng5 c4

16...Ra7 was played in my reference game 17.Qf3 c4 18.Bc2 b4 19.Qf2 Rc7 20.Nce4 Nxe4 21.Nxe4 f5 22.d6 Rf7 23.Bg5 Qd7 24.Be7 and here Black opted to exchange rook for bishop and knight, but still went on to lose (Narciso Dublan,M [2459] -- Kovacevic,S [2442], La Pobla de Lillet 1999).

17.Bc2 Ra7

White to move

18.Be3

In addition to harassing the rook as the reference game above at 19.Qf2, this move creates the possibility of locating the bishop on the a3-f8 diagonal.

18...Rd7

18...Rc7 is no good due to 19.Bb6.

19.Qf3 Bb7 20.Qh3

Black to move

At this point in the game, my opponent took a long vacation. When he finally moved again, he had exhausted his vacation time and was under twelve hours on the clock. Not having looked at the game in over a month, I had forgotten that I had a nice position with good compensation for the pawn. I was beginning to hope for a time-out victory.

20...b4??

White maintains an advantage after 20...Bxd5, but must continue to find strong moves, or the attack will dissipate. Then, White's loss of two pawns may become decisive.

I saw that my opponent had moved while I was preparing some BBQ chicken for Saturday dinner. I quickly saw that my knight was safe for at least another two moves. The exchange sacrifice appeared to promise a strong attack. After five minutes of looking at Black's choices, I played my move.

21.Rxf6! 1-0

Black resigned a few hours later.

The game might have continued 21...Qxf6, which struck me as Black's only reply. Then, 22.Qh7+ Kf8 23.Nce4 (I planned 23.Bc5+, which Stockfish 3 considers an inaccurate move order. 23...Rde7 24.Nce4 Qf5 reaching the same position as the main computer line) 23...Qf5 24.Bc5+ Rde7.


*The strongest committed an elementary error in the King's Indian Defense and was later banned for engine use. I suspect that he turned his engine on too late in our game.