Video completed: Winning with the Dynamic Caro-Kann (The Deadly Bronstein-Larsen System)

I recently completed "Winning with the Dynamic Caro-Kann (The Deadly Bronstein-Larsen System)" - Foxy video vol. 162, by IM Andrew Martin. Like others in the Foxy series, it is a collection of recorded lectures, in this case centered on selected games that are narrated all the way through by IM Martin. There is no extra content (game data files, interactive quizzes, etc.) It was published in 2014 and although it contains several classic Bronstein-Larsen games, it focuses more on contemporary master-level examples from international tournaments, typically with players in the 2400+ Elo range.

The video display quality unfortunately is poor, as the demo chessboard is low-res. That said, the strengths of the annotated game format in explaining and demonstrating opening and middlegame ideas outweigh the technical minuses. Full games are presented, so it's a useful product for overall chess training as well, since tactics and strategic ideas are discussed all the way through the endgame. Video presentations by knowledgeable commentators like Martin help bring the material alive, much more than studying lists of variations, and I think the format also aids future recall of specific ideas and maneuvers.

There are 15 separate videos included, with a total running time of 2 hours 7 minutes. The first several look at various alternative move 6 options for White, before moving on to provide examples in the main theoretical 6. c3 line; however, later there are also a mix of options shown (primarily with 6. Nf3). The first 12 videos, containing narrated games, Martin at one point refers to as "introductory", and the last three supposedly contain his specific repertoire recommendations in the different move 6 White lines (6. Nf3, 6. g3 and c3 combination, and 6. Bc4). However, these are really just more example games, although he does present them based on his preference for 6...Bf5 in all cases.

It's worth noting that the "alternate" (to 6. c3) White lines are very important to study for a Black player, since they will likely be the most commonly faced. Especially at club level, 6. c3 - which develops no pieces and only moves the pawn forward one square - may not even occur to your opponent as an option. Other move 6 options, particularly the normal-looking 6. Nf3, are likely to appear on the board from White opponents (of whatever strength) who are not familiar with the Bronstein-Larsen. This probably means the majority of White players, in practical terms.

I found the most useful aspect of going through the narrated example games to be Martin's introduction and explanation of typical ideas and maneuvers, although concrete variations are of course also presented. Key recurring concepts highlighted include:

• Development of Black's light-square bishop to g4 vs. f5
• Deployment of black's rook to g8 along the half-open file, or alternatively using it to support an early h-pawn advance
• The typical development plan of ...e6, ...Nd7, and queenside castling followed by a kingside attack
• Alternative kingside castling for Black and ...Bf5-g6 ideas
• Formation of Bd6/Qc7 battery when possible
• Ideas involving ...Qa5+ and moving along the 5th rank subsequently
• Disruptive ...Bb4 opportunities
• Timing the pawn breaks/advances ...e5 (either in one go, or after a preliminary ...e6) and ...c5
• Black's requirement to play actively with threats and counter-threats in the center and on the kingside, while not being afraid of calculating tactical defense ideas on the queenside.

Various Black responses to White's different move 6 choices are given in the introductory videos. The 6. c3 line responses include 6...Qd5 (recommended by Martin as a good alternative), 6...h5 (more chancy), and the standard 6...Bf5.

In general, these types of video lecture resources help fill in gaps when learning openings, since they do more than just go through book variations and give an evaluation at a certain cut-off point. Where to ideally place your pieces and the trade-offs involved in making these kinds of development decisions are what really underpin opening theory and practice. However, these concepts are too rarely explained in simple, practical terms in most opening theory books. Martin here does a good job at highlighting these ideas for the Bronstein-Larsen, across a number of example games.

Although there is a substantial amount of material covered, with Martin at least looking at the main options in each line, I would still consider this product as complementary in nature to more comprehensive "book" materials (in whatever format) and your personal annotated opening repertoire database. Despite the "Winning with..." title, Martin in my view does a decent job of not over-hyping Black's play, which he summarizes as designed to make White feel uncomfortable. He is also careful in his assessments to highlight practical vs. theoretical considerations. Looking at the Bronstein-Larsen variations in depth will require further research and your own evaluation of them, as is the case with all opening study.

Commentary: 2017 Isle of Man International, Round 9 (Tarjan - Kosteniuk)

This next commentary game contains some themes for improving players at several different levels of analysis - meta, strategic, and tactical. "Meta" in this case refers to the overall context - the fact that GM James Tarjan, one of the best US players in the 1970s, was at the time of this game in the third year of his chess career comeback and at age 65 defeated both GM Alexandra Kosteniuk (below) and super-GM Vladimir Kramnik during the 2017 Isle of Man International. This was no fluke, as he had also played for the US in the 2016 chess Olympiad. Seeing these kinds of examples helps combat the "inevitable decline" narrative associated with the aging process, or at least provides fewer excuses for not undertaking effortful study.

While we (or at least I) may not have Tarjan's level of inherent talent, his approach and the example of play here are understandable and instructive. My top observations from the game:

• Tarjan's opening choice is designed to allow White to "play chess" rather than debate opening theory. This strategy used to be frowned upon in general, with purists insisting White always play for a forced if slight advantage. However, Carlsen's repeated use of it over the years has lent it more legitimacy; one game of his is in a similar variation is included in the game notes.
• Master-level choice of candidate moves. I highlight multiple instances where White's move choice probably would not occur to an amateur. I find these to be one of the most important aspects of studying and analyzing master-level games, as they demonstrate how new ideas can be introduced into your own play.
• The interplay of tactical and strategic considerations is evident throughout, especially when Black - probably under time pressure - starts missing key tactics in the latter part of the game. Using tactics to achieve more of a strategic/positional advantage was also possible at several points in the game for both sides.

[Event "Chess.com Isle of Man Open - Masters"] [Site "Douglas (Isle of Man)"] [Date "2017.10.01"] [Round "9.32"] [White "Tarjan, James"] [Black "Kosteniuk, Alexandra"] [Result "1-0"] [ECO "A22"] [WhiteElo "2412"] [BlackElo "2552"] [Annotator "ChessAdmin/Komodo Dragon"] [PlyCount "99"] [EventDate "2017.09.23"] 1. c4 e5 2. g3 Nf6 3. Bg2 d5 4. cxd5 Nxd5 5. Nc3 Nb6 6. b3 {this move takes the game out of reversed Sicilian territory. Nf3 is the most played in the database.} Be7 {Black commits the bishop early. The following top-level game shows an alternate plan of development.} (6... Nc6 7. Bb2 Be6 8. Bxc6+ bxc6 9. Nf3 Bd6 10. d4 exd4 11. Qxd4 f6 12. Qe4 Kf7 13. Qxc6 Qe8 14. Qe4 h5 15. Nd4 Bd7 16. Qxe8+ Rhxe8 17. f3 c5 18. Nc2 Be5 19. Nd1 Bxb2 20. Nxb2 Bb5 21. e4 f5 22. Ne3 fxe4 23. f4 Red8 24. Rd1 Rd4 25. Nf5 Rad8 26. Nxd4 cxd4 27. a4 Ba6 28. a5 Nc8 29. Nc4 Ne7 30. Kf2 Nf5 31. Rhe1 e3+ 32. Kf3 Bb7+ 33. Ke2 Ke6 34. Rg1 Be4 35. Ne5 Rb8 36. Nf3 Bxf3+ 37. Kxf3 Rxb3 38. Ke4 e2 39. Rd3 Rb5 40. Re1 Rxa5 41. Rxe2 g6 42. Kf3+ Kd6 43. h3 Ra4 44. g4 hxg4+ 45. hxg4 Ne7 46. Ke4 Nc6 47. Rh2 Ra1 48. Rh6 Re1+ 49. Kf3 Rf1+ 50. Kg3 Rg1+ 51. Kh4 Rh1+ 52. Kg5 Rxh6 53. Kxh6 Kd5 54. Kxg6 Ke4 55. Rd1 Kxf4 56. g5 Ne5+ 57. Kh5 Ke4 58. Re1+ Kf5 59. Rf1+ Ke4 60. Re1+ Kf5 61. Kh6 Nf7+ 62. Kh5 Ne5 {1/2-1/2 (62) Carlsen,M (2863) -Nepomniachtchi,I (2784) Lichess.org INT 2020}) 7. Bb2 {the bishop must develop here and there is no reason to postpone it. It also gives White a look at Black's next developing move before making any other decisions about piece placement.} Nc6 (7... O-O {seems more consistent with the early Bishop move, also waiting to see where to go with the Nb8.}) 8. Bxc6+ $5 {Tarjan is the only one in the database to play this, sending the game on to a different strategic path. It is a classic trade-off decision, eliminating the strong White bishop but inflicting permanent structural damage on Black and providing White with relatively easy targets to work against. The engines evaluate the position as equal.} (8. Nf3 {is of course the conventional move.}) 8... bxc6 9. Nf3 {developing and attacking e5.} Qd6 (9... f6 {would be the more standard way to protect the pawn in similar Black structures.}) 10. O-O {tucking the king away before Black can play ...Bh3. Interestingly, the engines do not consider this such a problem.} (10. Rc1 $5 {immediately moving to increase pressure on the c-file.} Bh3 11. Qc2 $11 {White's king is in no danger and the rook can go to g1 and still be useful.}) 10... Bf5 $6 {this fights for the e4 square, but not very effectively, considering White's next move.} (10... Bh3) ( 10... O-O) 11. d3 $14 {now White is threatening Ne4, with a discovered attack on the e5 pawn.} Qe6 {avoiding a double attack on the queen after White's next. } 12. Ne4 {Black now has an important strategic decision to make. Her choice to castle queenside is very imbalancing and appears to go against the position's needs. Kosteniuk must have felt that either the king was safe enough there, despite the weak pawns, or that she would be able to attack White first.} O-O-O $6 {Black tactically protects e5, as the Ne4 is now hanging due to the pinned d-pawn. However, White gets out of the pin by moving his queen to a better square, lining up on the c-file, so it is a net minus for Black.} (12... f6 {is the engine recommendation, protecting the pawn.}) ( 12... Bxe4 $6 {looks like a reasonable idea, inflicting some pawn structure damage on White in return. However, Black's c-pawns are still worse than White's e-pawns and Black would no longer have the two bishops.}) 13. Qc2 $16 f6 14. Rfc1 {putting a rook on c1 is clearly a good idea, creating a battery with additional pressure. An argument could be made for moving the other rook, since the a-pawn will not need its support to advance, but White evidently wanted to put both heavy pieces on the queenside.} Kb7 {a reasonable-looking defensive move, but Black may be better off looking for some counterplay.} ( 14... Bg4 $5 {is the engines' preference, threatening to exchange on f3 and undermine White's d-pawn.}) 15. b4 $5 {a move that no amateur would be likely to consider, as it just appears to lose a pawn.} (15. a4 {is the more obvious way of launching operations on the queenside.}) 15... Bxb4 16. Rab1 {now we see the idea of the pawn sacrifice, to attack down the b-file. It appears to be slower than alternatives, though, as Black looks to equalize with ...Rd5 or exchange off an attacking piece with ...Bxe4. In practical terms, it is still difficult to meet, however.} (16. a4 Rd5 17. a5 Nd7 18. Bxe5 $5 fxe5) 16... Be7 {withdraws the bishop from its current vulnerable square, eliminating tactics involving a discovered attack after the Bb2 moves (for example Bd4 or Bxe5). This does little to impede White's attacking ideas, however.} (16... Rd5 { this defends along the 5th rank, although White still has pressure and tactical ideas in the center against the rook and bishop. For example} 17. Bd4 Ba3 18. Nc5+ Bxc5 19. Bxc5 e4 20. Nd4 Rxd4 {the defensive exchange sacrifice is necessary} (20... Qd7 $2 21. dxe4) 21. Bxd4 {at this point the engine shows complete equality, despite the material difference, meaning that with accurate defensive play Black will be all right. If Kosteniuk calculated this far, however, it would hardly look appetizing.}) (16... Bxe4 $5 {may be the best practical choice here, eliminating an attacking piece and reducing complexity.} 17. dxe4 a5 {physically impeding White's future a-pawn advance} 18. a3 Be7 $14) 17. a4 {with the idea of pushing away the knight. Bringing another piece into the attack might boost its effectiveness.} (17. Nfd2 {appears to be a useful preparatory move, as the Nf3 now is not doing much otherwise.}) (17. Bc3) 17... Ka8 {getting off the open file.} 18. a5 Nd7 {although Black has been pushed around, White does not have anything decisive and now looks to recover the pawn. However, there is no rush to do so and he ends up in a slightly worse position with the move played, so alternatives are worth looking at.} 19. Qa4 ( 19. Ra1 $5 {with the idea of Ba3 and trading off White's worse bishop, helping fight for the dark squares onthe queenside.}) (19. Qxc6+ Qxc6 20. Rxc6 { regains the pawn more forthrightly and seems to give Black less leeway than the game continuation.}) 19... Rb8 {challenging on the open file and giving back the pawn for equality.} (19... Nb8 $6 {is a move amateurs might play, defending the c6 pawn at all cost. This would renew White's attacking chances, however. For example} 20. Nfd2 Rd5 21. Ba3 Bxa3 22. Qxa3 Na6 23. Rc4 {followed by Rbc1.}) 20. Rxc6 {this forces Black's next move, otherwise the c-pawn is lost.} Qb3 21. Qxb3 Rxb3 22. Nfd2 {finally the knight gets into the action.} ( 22. Rxc7 $2 {greed is punished by} Rhb8 23. Rc2 Ba3 $19) 22... Rb7 {now Black is doing fine and the open b-file is more of a benefit for her than it is for her opponent.} 23. Rc2 {this overprotects the Bb2, freeing the Rb1 to move, but seems a bit passive.} (23. a6 $5 Rb6 24. Rxc7 Rhb8 25. Rc2 Rxa6 {with a slight advantage to Black, with the passed a-pawn.}) 23... Rhb8 24. Nc4 { overprotecting the Bb2 again and getting the knight further into play.} a6 { physically blocking the further advance of the White a-pawn.} 25. Rbc1 { moving the rook out of the pin and doubling pressure on the c-file. White has managed to rearrange his pieces to be less awkward and can look to exchange off his worse dark-square bishop with Ba3.} Rb3 {this temporarily stops the bishop exchange idea, but White could still insist on it.} 26. Bc3 (26. Ned2 R3b5 27. Ba3 Bxa3 28. Nxa3 Rxa5 29. Nac4 Rc5 30. Ra2 Rc6 31. Rca1 $11 {with play against the a-pawn.}) (26. f4 $5 {could also be played immediately.}) 26... Be6 {Black again passes up the chance to exchange on e4. This seems to help validate White's previous move, though, as the bishop is now more useful on the a5-e1 diagonal protecting the a-pawn.} (26... Bxe4 27. dxe4 Nc5) 27. f4 $6 {one of the common cases where playing the right idea at the wrong time - one tempo later, in this case - could result in a setback.} exf4 {Black makes the obvious move, to avoid losing the e-pawn (attacked three times, defended twice).} (27... Bxc4 {this possibility is the difference.} 28. dxc4 exf4 29. gxf4 Nc5 30. Ng3 Bd6 31. e3 $17 {now White has three weak pawns (a5, c4, e3) to defend and Black's pieces are much better placed.}) 28. gxf4 {this allows Black to play ...Bxc4 again, but she overlooks this.} (28. Ncd2 {is the engines' recommendation, sacrificing the pawn for vigorous piece play.}) 28... R3b5 29. Bd4 $11 {now White is completely equal.} Bb4 {attempting to pick up the a-pawn. Komodo recommends to simply ignore it, as White's rooks could then make threats on the c-file.} (29... Bxc4 30. Rxc4 $11) 30. Bc3 {an equal defensive move. Given the repetition of moves that occurs here, it appears both players were likely low on time.} (30. Ne3 $5 {this unleashes the rooks.} Bxa5 31. Rc6 Bb6 32. Bxb6 R8xb6 33. Rxc7 $16) 30... Be7 31. Bd4 R8b7 32. Kg2 { White appears to be just marking time here. Perhaps this succeeded in provoking Black's next, which is a non-obvious blunder.} Kb8 $2 {this appears to be a reasonable move, getting the king out of the corner and reinforcing the pawn on c7. It has a tactical problem, though, which Tarjan finds.} (32... Bd5 {or moving to f7 or g8 would avoid the problem, which is created by the king making itself vulnerable to a back-rank check, physically blocking the rook from coming back to b8.}) 33. Ncd6 $1 {an example of a reloader tactic on d6, made possible by the threat of Rc8+} Bxd6 (33... Rb1 {the engines suggest leaving the bishop on the board rather than exchange it, but it's still a win for White, who is up the exchange with no compensation for Black.} 34. Nxb7 Rxb7 (34... Rxc1 35. Rxc1 Kxb7 36. Nc5+ $18) 35. Kf2 $18) 34. Nxd6 {and now Black loses material.} Rd5 35. Nxb7 Rxd4 36. Nd8 Bd5+ 37. e4 Kc8 38. Nc6 Bxc6 39. Rxc6 Rxd3 40. Rxc7+ $18 {at the end of the sequence, White is simple an exchange up with a dominating rook pair.} Kd8 41. Ra7 {both getting behind the a-pawn and leaving the c7 square potentially available for the other rook.} Rd6 42. Kf3 g5 {attempting to get any counterplay possible, by breaking up the pawn shield in front of White's king, or getting a kingside pawn majority.} 43. Rg1 h6 {this looks like it is reinforcing the g-pawn, but ends up giving another pawn to White.} 44. h4 {another interesting master idea, although an amateur might have more of a chance of spotting this tactic. The h-pawn cannot be taken due to the mate, and Black's g-pawn cannot be further reinforced, so it is lost.} Ke7 45. hxg5 hxg5 46. fxg5 fxg5 47. Rxg5 Kf6 48. Kf4 Ke6 49. Rg6+ Nf6 {allowing one final tactic.} 50. Rxf6+ {and now the pawn forks on e5 after the rook is recaptured.} 1-0

"How I won more games by improving my chess thought process" - article

NM Dan Heisman has posted an article by one of his students on his Chess.com blog - "How I won more games by improving my chess thought process" - that is one of the better practical descriptions I've seen of the central importance of your thought process to gaining chess strength. The main issues described, constantly calculating variations while also missing good candidate moves, are common ones and unlikely to be overcome unless the improving player is self-aware enough. Of course, once you are aware of the problem, it is then useful to have some concrete guidance on how to address it, which the article offers.

In my own chess improvement process, through analyzing my own games, I early on identified the lack of a consistent thought process as a significant handicap, resulting in the Simplified Thought Process (That Works). More recently, I've had some success in better identifying my opponent's resources (i.e. their candidate moves which result in threats) using tactics training to strengthen the thought process.

Training quote of the day #36: Jacob Aagaard

Grandmasters know that a positional advantage is better than a material advantage

In my experience amateurs are reluctant to take extra material when they should, but have a tendency to do so when they should not. I think it is all about confidence. When you are under pressure from a grandmaster, it can be psychologically difficult to accept a pawn or a piece on offer. Often amateurs choose to take a worse position with material equality, and subsequently get hammered. At the same time amateurs have a tendency to cash in on a positional advantage far too soon, instead of maintaining the pressure. Again it is a matter of confidence. If you don't believe in your own abilities, you are less likely to trust your evaluation of the position and consequently seek some sort of security outside yourself, such as an extra pawn.

From Grandmaster Versus Amateur, Chapter 1

Commentary: 2017 U.S. Championship, Round 10 (Zherebukh - Nakamura)

Here we continue with the theme of Caro-Kann Classical wins by Black at the top level, with this game featuring dynamic opposite-side attacking play by GM Hikaru Nakamura from the 2017 U.S. Championship. Unlike the 2020 commentary game in which he lost as White to Awonder Liang, here Nakamura plays actively, even aggressively, but in a way that is in tune with the position's characteristics rather than attempting to force an attack. His opponent, GM Yaroslav Zherebukh, had been having a great tournament up to that point, but apparently was caught out by Nakamura's opening choice and subsequent play.

White's problems stemmed from his choice on move 14 to prematurely exchange minor pieces. This resulted in the h-file being opened for Black's rook, and he was also able to start pushing around White in the center, seizing the initiative and conducting a punishing attack. Nakamura's execution of it is particularly instructive. The variations show the interplay of various tactical ideas, revolving around fixing White's vulnerabilities around the king position - particularly the h1 and f2 squares - and taking advantage of loose/awkward piece placement. A devastating win in only 21 moves, showing that the Caro-Kann Classical can be a fighting choice as well as a solid one for Black.

[Event "U.S. Championships Men 2017"] [Site "Saint Louis"] [Date "2017.04.08"] [Round "10"] [White "Zherebukh, Yaroslav"] [Black "Nakamura, Hikaru"] [Result "0-1"] [ECO "B18"] [WhiteElo "2605"] [BlackElo "2793"] [Annotator "ChessAdmin/Komodo Dragon"] [PlyCount "42"] [EventDate "2017.??.??"] 1. e4 c6 2. d4 d5 3. Nc3 dxe4 4. Nxe4 Bf5 5. Ng3 Bg6 6. Nh3 {this has the same idea as the "book" N1e2, namely deploying the knight to f4. It does not seem to have independent significance and is not mentioned in theory books, although it is interesting to see that it is roughly equivalent in the database - actually with a few more games - to N1e2.} Nf6 (6... e6 {is the other main response.}) (6... h6 {is the third most popular, but scores very well for White. There is no need to prevent the knight from going to g5 and preparing to retreat the Bg6 loses time for Black.}) 7. Nf4 Nbd7 (7... e5 { has been the traditional antidote to this variation since the Tal-Botvinnik matches in the 1960s, but the line can be drawish, so likely was avoided by Nakamura for this reason. Here's a high level example:} 8. Nxg6 hxg6 9. dxe5 Qa5+ 10. Bd2 Qxe5+ 11. Qe2 Qxe2+ 12. Bxe2 Nbd7 13. O-O O-O-O 14. Rad1 Nb6 15. Bc1 Bd6 16. Rfe1 Kc7 17. h3 Rhe8 18. Bd3 {1/2-1/2 (18) Radjabov,T (2735)-Anand, V (2799) Morelia/Linares 2008}) 8. Bc4 Qc7 9. O-O e6 10. c3 {while structurally sound, this is a rather passive move.} (10. Re1 {is a more typical White idea in this position, pressuring the e-file and threatening some tactical ideas if Black gets careless with development.} Bd6 {and now White can sac on e6, but it only maintains equality if Black defends correctly. However, it could certainly be a good practical try, Ivanchuk lost a blitz game against Duda in this line.} 11. Rxe6+ fxe6 12. Nxe6 Qb6 13. Qe2 Ke7 { and now the defensive idea is to play ...Rae8, perhaps after a4 and ...a5 are inserted. That way the king can flee via d8 and not lock the rook out of the game.}) 10... Bd6 11. Qf3 {Defending the Nf4. Also, by placing his queen on the h1-a8 diagonal, he influences e4 and d5 and eyes b7 in the event Black goes for the typical ...c5 break, which Nakamura does play immediately.} c5 { after an exchange on c5, recapturing will help activate Black's minor pieces. White could also push the d-pawn in response, but Black does not have any problems there.} 12. dxc5 (12. d5 {there are not in fact any database games with this line. White players likely do not want to give up the center so easily.} e5 $5 {may be the easiest way to neutralize White's chances.}) 12... Bxc5 (12... Nxc5 {no one plays this way, since the knight will just have to go back.} 13. Bb5+ Ncd7) 13. Bb3 {White takes care of his hanging bishop, otherwise ...Bxf2+ is threatened.} O-O-O {by castling opposite sides, Black signals further that he is looking for winning chances. It is also the engines' top choice. While Black's king position looks more open, he has effective local superiority on the queenside, as White is much better positioned for action on the kingside. The half-open c-file and queenside pawn majority are also not immediate problems, as it will be tough for White to mobilize anything there. Finally, White's dark-square bishop is currently out of the game, something which is corrected shortly, but still leads to problems. } 14. Nxg6 $2 {positionally, this allows for the bishop to come to a more active square, but ignores the even more significant extra activity that Black gains as a result, most notably through the opening of the h-file. Tactically, White needed to prepare the piece exchange more, as now he gets pushed around and gives Black the initiative.} (14. Be3) (14. Qe2) 14... hxg6 $17 {this is exactly what Black wanted with queenside castling, now the Rh8 is ideally placed to exert pressure on White's king position.} 15. Bf4 e5 {now Black is firmly in control.} 16. Be3 e4 {I think it likely that White underestimated or missed this move, attacking the queen with tempo. The Ng3 is pinned against the mate threat on h2, is the problem.} 17. Qe2 {abandoning the kingside to its fate.} (17. Qf4 $5 {would put up more resistance on the kingside.} Bd6 18. Qg5 Ne5 $19) 17... Bxe3 (17... Rxh2 {is also possible immediately, but trading bishops simplifies the calculations.}) 18. Qxe3 (18. fxe3 {would avoid the threat of a K+Q fork on h2 and e3, but has its own problems, namely that the Ng3 will no longer be protected after the sacrifice on h2.} Rxh2 {and the rook still cannot be taken, due to} 19. Kxh2 Rh8+ 20. Kg1 Qxg3 {with a winning attack, thanks to the Nf6. Eliminating it is not enough, either.} 21. Rxf6 Nxf6 22. Bxf7 Kb8 {and Black wins material after ...Qh2+, thanks to the hanging Ra1. Trying to defend with Qf2 loses to the deflection tactic ...Rh1+}) 18... Rxh2 { this thematic sacrificial breakthrough idea should always be looked at when the other rook is available to immediately come to the h-file.} 19. Rfd1 { nothing better than to give White's king a path to run.} (19. Kxh2 $2 Ng4+) 19... Rdh8 {Black is now simply up material with a strong attack and dominating position. He does not have an immediate combinational win, but White is going to have to drop material or worsen his king position considerably.} 20. Qxa7 {re-establishing material equality at least, in the hopes that Black cannot carry through with his attack.} (20. Rxd7 Nxd7 21. Rd1 Kb8 22. Qxe4 {and now Black can win easily by entering a material-up endgame with} (22. Bxf7 {and White looks like he might have some compensation, but a nice sacrificial winning line for Black is} Ne5 23. Be6 Nf3+ $1 24. gxf3 Qxg3+ 25. fxg3 Rh1+ 26. Kf2 R8h2#) 22... f5 23. Qe7 Nc5 24. Qxc7+ Kxc7 $19) (20. Kf1 Kb8 {and now White has no good defense, as Black's knights can mobilize further to c5, e5 and g4 and White's king is too open. For example} 21. Qd4 Ne5 22. Bc2 e3 {similar to the game} 23. Qxe3 Nfg4 $19) 20... e3 $1 {now the threat is ...Qxg3, and the queen cannot be taken because of the mate on h1; the e3 pawn covers the f2 flight square.} 21. Qxe3 (21. Qa8+ Nb8 22. Rd4 R2h4 23. Rad1 (23. Rxh4 exf2+ 24. Kxf2 Rxh4 $19) 23... e2 $19) (21. Rd4 Qxg3 22. Qa8+ Qb8 23. Rc4+ Kd8 24. Qxb8+ Nxb8 25. Rd1+ Nbd7 $19) 21... Ng4 {because the knight replaces the black pawn in covering the f2 flight square, the ... Qxg3 threat is renewed, along with mate threats on h1, and they can no longer be met without White losing his queen.} 0-1