I just received a game from Philidor1792, and although it is not a "regular" Jerome Gambit, it is related, or "inspired by" - and it blew my mind. It features a pretty scary gambit, completely "objectively" unsound, but very uncomfortable to play against.
You have been warned.
saprvade - Philidor1792
3 2 blitz, lichess.org 2016
1.e4 e5 2.Bc4 Nf6 3.d4 exd4
So far we have a variation of the Bishop's Opening, Berlin Defense. The position could also arise from the Center Game, von der Lasa Gambit, 1.e4 e5 2.d4 exd4 3.Bc4 Nf6.
Now White can play 4.Nf3 and hope to transpose into the Urusov Gambit, although Black has 4...Nc6, moving the game into a Two Knights Defense or Scotch Gambit.
4. Bxf7+
Say what??
I could find no game examples in The Database, or in the ChessBase Big Database 2016. A Google search online was of no help.
Philidor1792 pointed me toward the database at lichess.org, where I discovered that there were 41 games (including 1 0 bullet games and 3 0 blitz games) where White scored 51%.
4...Kxf7 5.e5 Qe7 6.Qe2 Ne8
Philidor1792 is a long-time Jerome-Gambit-and-related player. He knows what he is doing with Black, but the position already looks quite uncomfortable. Remember, too, this is a 3 2 blitz game.
7.Nf3 d6
Black's move is logical, but is a slip. Best seems the immediate safeguarding of the King with 7...Kg8 e.g. 8.Nxd4 and Stockfish 8 suggests the tortured continuation 8...c5 9.O-O Nc7 (if 9...cxd4 then 10.Qc4+ d5 11.Qxc8 Qd7 12.Qxd7 Nxd7) 10.Nf5 Qe6 11.Ne3 Nc6 12.f4 b6 13.c4 Nd4 and "of course" Black is "winning". Nice work if you can get it, as they used to say.
analysis diagram
8.Ng5+ Kg8
Of course this seems reasonable, but, as Philidor1792 points out, if 8...Kg6 then 9.e6 Bxe6 10.Nxe6 Kf7 11.Nxd4 Nc6 (or 11...Qxe2+ 12.Nxe2 Nc6) 12.Nxc6 Qxe2+ 13.Kxe2 bxc6 with an equal game.
Of course. Easy as pie, in the blink of an eye.
analysis diagram
9.Qc4+ Be6 10.Nxe6 Qf7 11.Qb3 Nd7 12.Nxd4 dxe5 13.Nf3 Bc5 14.Nc3 Qxb3 15.axb3 Nef6
Black has pretty much weathered the storm, although White has an edge due to a slightly better pawn structure and a safer King.
16.O-O Kf7 17.Re1 Rhe8 18.Be3 Bd6 19.Bxa7 b6 20.Nb5 Rec8
White's Queenside adventures do not seem to be amounting to much.
21.Red1 Kg8 22.Nc3 e4 23.Nd4 Nc5 24.Ncb5 Nb7 25.Nc6 Bf8 26.b4 Nd6 27.Nxd6 Bxd6
White is still a bit better, but not by very much. What in the world are his minor pieces doing?
28.c4 Nd7 29.Rxd6 cxd6 30.Ne7+ Kf7 31.Nxc8 Rxc8 32.b3 Ra8
33.Kf1 Ke7 34.Ke2 Kd8 35.c5 bxc5 36.bxc5 dxc5 37.Ra3 Kc7
This looks like a time slip. If the Jerome-ish opening doesn't get you, the clock will...
38.Bxc5 Rxa3 39.Bxa3
Facing a Bishop vs a Knight, and an outside passed pawn, Black can only hope his opponent's flag falls, but it does not do so - although almost 40 more moves are played.
39...Kc6 40.Ke3 Kd5 41.Bb2 g6 42.b4 Nf8 43.b5 Ne6 44.Ba3 Nd8 45.Be7 Nb7 46.Kf4 Ke6 47.Bf8 Kd5 48.Kg5 Ke6 49.Kh6 Kf7 50.Bb4 Kg8 51.Kg5 Kf7 52.Kf4 h5 53.Kxe4 Ke6 54.Kf4 Kf6 55.b6 g5+ 56.Ke4 Ke6 57.f4 gxf4 58.Kxf4 Kf6 59.Bc3+ Kg6 60.Ke5 Nc5 61.Kd5 Nd7 62.b7 Kf5 63.Kd6 Nb8 64.Be5 Na6 65.Kd5 Nb4+ 66.Kc5 Nd3+ 67.Kd6 Nxe5 68.b8=Q Nc4+ 69.Kc5 Ne5 70.Qh8 Ng6 71.Qxh5+ Kf6 72.Kd6 Nf8 73.h4 Ng6 74.Qg5+ Kf7 75.h5 Nf8 76.Qe7+ Kg8 77.h6 White resigned
The following game has a line in the Jerome Gambit (Delayed?) that I am not very familiar with, so I consulted with The Database and Stockfish 8 for some assistance. It is good to see jfhumphrey move quickly to victory.
jfhumphrey - ckilee
5 0 blitz, FICS, 2017
1.e4 Nc6 2.Nf3 e5 3.Bc4 Bc5 4.O-O
4...Nf6
Transposing to a Two Knights Defense (1.e4 e5 2.Nf3 Nc6 3.Bc4 Nf6) with ...Bc5 added.
The Database has 1,148 games with this position, with White scoring 42%. (For comparison, The Database shows White scoring 45% after 1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+)
5.Bxf7+ Kxf7
Now The Database has 1,909 games with this position (oh, the joys of transposition) with White scoring 40%.
6.Nxe5+ Nxe5 7.d4
The Database has 291 games with this position, with White scoring 45%.
7...Bd6
Now, however, The Database has 35 games with this position, with White scoring 71%. From a practical point of view, Black has made an error that generally gets punished.
Interestingly enough, Stockfish 8 sees the text move as giving White less than a half of a pawn advantage. It prefers 7...d5, and this is backed up by The Database, which has 10 games with that move - all of them wins for Black.
8.dxe5 Bxe5 9.f4 Bd6 10.e5 Bc5+
Stronger than 10...Ne4?! 11.exd6 (11.Qd5+!?) Nxd6 12.Qd5+ Kf8 13.f5 Qf6 14.Nc3 c6 15.Qf3 Nf7 16.Be3 Ne5 17.Qe4 d5 18.Qd4 Kf7 19.Bf4 Re8 20.Bxe5 Qxe5 21.Qf2 Kg8 22.Rae1 Qxe1 23.Rxe1 Rxe1+ 24.Qxe1 Bxf5 25.Qe7 h5 26.h3 Bxc2 27.Qxb7 Rf8 28.Qxc6 Bd3 29.Nxd5 Rf1+ 30.Kh2 Be4 31.Qe6+ Kh7 32.Qxe4+ Black forfeited on time, Teterow - yelloman, FICS, 2011.
11.Kh1 Ne4
Better alternatives are: 11...d6 12.exf6 Qxf6 13.Nc3 c6 (13...Qf5 14.Nd5 c6 15.Nc3 d5 16.Ne2 Re8 17.Ng3 Qf6 18.Bd2 Qxb2 19.Rb1 Qf6 20.Nh5 Qf5 21.Bc3 d4 White forfeited on time, aymmd - JabbaChess, FICS, 2012) 14.Ne4 Qf5 15.Nxd6+ Bxd6 16.Qxd6 Re8 17.Bd2 Be6 18.Bc3 Bd5 19.Qc5 Bxg2+ 20.Kxg2 Qxc5 White resigned, motorfot - theophraste, FICS, 2014;
11...Bb6 12.exf6 Qxf6 13.Qd5+ Qe6 14.Qh5+ Qg6 15.Qf3 Rf8 16.f5 Qc6 17.Qb3+ d5 18.Nd2 Kg8 19.Nf3 Qc4 20.Bg5 Qxb3 21.cxb3 Bxf5 22.Be7 Rfe8 23.Bb4 Be4 24.Ng5 h6 25.Nh3 c5 26.Bc3 Rf8 27.Rfe1 Bf5 28.Nf4 d4 29.Bd2 Rae8 30.Nd5 Black forfeited on time, jfhumphrey - Dimo, FICS, 2012; and
11...d5 12.exf6 gxf6 13.Nc3 Bf5 14.Nxd5 Bxc2 15.Qxc2 Qxd5 16.Bd2 Rag8 17.Bc3 Be3 18.Rad1 Qb5 19.a4 Qc6 20.Rf3 Bc5 21.Qf5 Be7 22.a5 Rd8 23.Re1 Rd5 24.Qe4 Rd7 25.Rh3 White forfeited on time, aymmd - GriffySr, FICS, 2011.
Incorrect are 11...Nd5 12.Qxd5+ Black resigned, Svrbt - pantyhose, FICS, 2011 and
11...Ng8 12.Qd5+ Ke8 13.Qxc5 Ne7 14.f5 Rf8 15.f6 gxf6 16.Bh6 Rg8 17.exf6 b6 18.Qh5+ Ng6 19.f7+ Ke7 20.Qg5+ Kd6 21.Rd1+ Kc6 22.Qd5 checkmate, chessmoods - MChessplayer, FICS, 2012.
Best is 11...Be7 12. exf6 Bxf6 when White has a slight edge.
12.Qd5+ Ke8
Or 12...Kf8 13.Qxe4 d6 (13...Qe7 14.Be3 d6 15.Bxc5 dxc5 16.Nc3 Kf7 17.Rad1 g6 18.Nd5 Qe6 19.Nxc7 Rb8 20.Nxe6 Black resigned, HauntedKnight - jonisjonis, FICS, 2010) 14.e6 Qe7 15.f5 c6 16.Nc3 d5 17.Qg4 Qf6 18.Bg5 Qd4 19.Rad1 Qxg4 20.Bf4 Ke7 21.h3 Qh5 22.Rf3 g5 23.Be3 Bxe3 24.Rxe3 g4 25.Rf1 gxh3 26.Rxh3 Qg5 27.Ne2 h5 28.c3 Rg8 29.Rg3 Qxg3 30.Nxg3 Rxg3 31.Kh2 Rg8 32.b3 Kd6 33.a4 b6 34.b4 Ba6 35.Rf2 Ke5 36.e7 Rg5 37.f6 Ke6 38.Kh3 Rf5 39.Re2+ Kf7 40.g3 Black forfeited on time, HauntedKnight - heverson, FICS, 2011.
13.Qxe4
White has a pawn advantage. At this point the clock seems to be influencing move choices.
13...Rf8 14.Qxh7 d6 15.Qxg7 dxe5 16.Qxe5+ Be7
White will finish the game off in short order.
17.Nc3 Bf5 18.Be3 b6 19.Rad1 Qc8 20.Nd5 Rf7 21.Nxc7+ Kf8 22.Qh8 checkmate
Bleh...
Abject humiliation.
I just completed a 10 0 online blitz game with the Jerome Gambit.
I won on time.
I won't annoy my opponent by giving his name.
For historical purposes, here is the final position when his flag fell (with Black to move, mind you). I don't need Stockfish's help to figure this one out.
I know that I have shown other Jerome Gambit players escaping similarly, but it's a bit different when it is my game.
What kind of a role model is that?
I have half a mind to take up knitting.
I know that I have said before that I am glad that I read this blog, but my most recent Jerome Gambit game reinforced this habit - on the day the previous post appeared, I played the following game. Using "The Machine Idea", I quickly developed an advantage. Despite my opponent's efforts at counter-attack, and the time ticking off my clock, I was able to construct a checkmate.
perrypawnpusher - Praotorian
5 5 blitz, FICS, 2017
1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+
4...Kxf7 5.Nxe5+ Nxe5 6.Qh5+ Ke6
7.Qf5+
Kd6 8.f4 Bd4
An idea favored by some chess computer programs, looked at in Petasluk - GriffyJr, blitz, FICS, 2017, (0-1, 30).
9.fxe5+ Bxe5
Instead of this capture, Stockfish 8 recommends the following line, which takes the game in a very different direction, but which winds up with a small advantage for Black: 9... Kc6 10.c3 d6 11.e6 Ne7 12.Qf1 Rf8 13.Qc4+
Bc5 14.a4 a6 15.d4 b5 16.Qe2 Ba7 17.axb5+ axb5 18.d5+ Kb7 19.Qxb5+ Bb6
20.Ra3 Rxa3 21.Nxa3 Ka7 22.Qa4+ Ba6 23.Nc2 Qe8 24.Qxe8 Rxe8. White has three pawns for the sacrificed piece, but Black has the two Bishops.
Analysis Diagram
10.d4
I remembered GriffyJr's improvement.
10...Bxd4
A blitz move. Black's best was the counter-attack with 10...Qh4+.
11.Qd5+ Ke7 12.Qxd4 Qe8 13.Bg5+ Kf8
Material is even, but White's attack is deadly. Stockfish 8 comlained after the game that 13...Kf7 would have held out longer, but I still would have had a significant advantage.
14.O-O+ Nf6 15.Bxf6 gxf6 16.Rxf6+
Good enough to win, but a post mortem suggested 16.Qxf6+ Kg8 17.Qg5+ Qg6 18.Qe7 Qe8 19.Qxe8+
Kg7 20.Qe5+ Kh6 21.Rf6+ Kg7 22.Qg5 checkmate. This is a good maneuver to remember - even in time pressure.
16...Kg8 17.Qd5+ Kg7 18.Qg5+ Qg6 19.Rxg6+ hxg6
White has a Queen and a couple of pawns for a Rook - if he can manage his clock. Unfortunately, although my opponent's time was moving quickly, mine was falling even faster. My opponent was aware of this, and continued to fight.
20.Qe5+ Kg8 21.Nc3 d6 22.Qf6 Rh7 23.Rf1 Bd7
Heh. In the Jerome Gambit I sacrifice two pieces and challenge my opponent: show me your win. Here Praotorian throws the challenge back at me.
24.Qxg6+ Rg7 25.Qf6 Kh7 26.Rf4 d5 27.Rh4+ Kg8 28.Nxd5 Rf8
Everything is fine, but tick, tick, tick...
29.Qc3 Bb5 30.Ne3 Be2 31.Nf5 Rgf7 32.Qh8 checkmate
Besides being used to annotate Jerome Gambit games of interest, computer programs have served as proponents or opponents in Jerome and Jerome-related games. The program below, GriffyJr, has visited this blog on several occasions - for starters, check out "Bots on Our Side" Part 1, Part 2 and Part 3 as well as "Ionman vs the Bots".
In today's game we examine the most recent example of a computer - generated move. There is still a lot of game to play afterward, however.
Petasluk - GriffyJr
5 0 blitz, FICS, 2017
1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+
4...Kxf7 5.Nxe5+ Nxe5 6.Qh5+ Ke6 7.f4
Bd4
An interesting and somewhat unusual idea. The earliest example of it in The Database was played in 2002 by the computer program BigBook at the online site FICS.
8.Qf5+
Instead, 8.c3 was seen in the oldest and most recent games with the line in The Database: 8...Nd3+ 9.Kd1 (9.Kf1 Nxc1 10.cxd4 a6 11.Qe5+ Kf7 12.Nc3 Nd3 13.Qf5+ Qf6 14.Qxf6+ Nxf6 15.e5 Ne8 16.f5 Nxb2 17.Rb1 Nd3 18.g3
a5 19.a3 b6 20.Ke2 Ba6 21.Ke3 Bc4 22.Nb5 a4 23.Rhc1 Nxc1 24.Rxc1 Bxb5 25.h4 c6 26.g4 d6 27.g5 dxe5 28.dxe5 c5 29.d4 cxd4+ 30.Kxd4 Be2 31.e6+ Ke7
32.Rc6 b5 33.Rb6 Nd6 34.Rc6 Nxf5+ 35.Ke5 Nxh4 36.Rc7+ Kf8 37.e7+ Ke8 38.Ke6 Ng6 39.Rc2 Bf1 40.Rf2 Bd3 White forfeited on time, megagurka - BigBook, FICS, 2002) 9...Nf2+ 10.Ke2 Nf6 11.Qf5+ Kf7 12.cxd4 Nxh1 13.d3 d5 14.e5 Bxf5 White resigned, Petasluk - GriffyJr, FICS, 2017.
And there was also another computer-with-black game: 8.f5+ Kd6 9.d3
g6 10.Qe2 gxf5 11.c3 Qh4+ 12.g3 Bxc3+ 13.bxc3 Qg4 14.d4 Nf3+ 15.Kf1 Qh3+
16.Kf2 Nxh2 17.Bf4+ Ke7 18.exf5+ Kf8 19.Nd2 Nh6 20.Bxh6+ Qxh6 21.Kg2 Qg5
22.Rxh2 Qxf5 23.Rf1 d5 24.Rxf5+ Bxf5 25.Qe5 Kf7 26.Rh5 Bd7 27.Qxd5+ Kg6
28.Ne4 Rae8 29.Qg5+ Kf7 30.Qf6+ Kg8 31.Rg5 checkmate, Moller,M - Mephisto, Denmark 2008.
Interestingly, Stockfish 8 suggests 8.fxe5 g6 9.Qg4+ Kf7 10.Qf4+ Kg7 11.c3 Bb6 12.a4 a5 13.d4 d6 14.O-O Qe7 15.exd6 cxd6 16.Na3 Be6 17.Kh1 h6 18.d5 Bd7 19.Nc4 Bc5 20.e5 Rf8 21.Qxf8+ Qxf8 22.Rxf8 Kxf8 23.Bf4 Ne7 with an even game because White's pawns are balancing out Black's extra piece.
8...Kd6 9.fxe5+
A clear improvement over 9.c3 Nd3+
10.Ke2 Ne7 11.e5+ Nxe5 12.fxe5+ Bxe5 13.Qd3+ Ke6 14.Qc4+ d5 15.Qg4+ Kd6
16.d4 Bxg4+ White resigned, fehim - blik, FICS, 2006. (Yes, blik is a computer.)
9...Bxe5
If it can't see a reason not to, a computer will grab material. Here GriffyJr shows some nearsightedness.
10.d4
Yes.
Interestingly, two earlier several-days-per move games (all players human) missed the idea:
10.O-O Qh4 11.h3 Qg3 12.Rf2 Nf6 13.d4
Bxd4 14.Bf4+ Kc6 15.Bxg3 Bxf2+ 16.Bxf2 d6 17.Qa5 Nxe4 18.Qa4+ Kd5 19.Nc3+ Nxc3 20.bxc3 Ke6 21.Re1+ Kf6 22.Qf4+ Kg6 23.Re3 Bf5 24.Rg3+ Kf6 25.Bd4+ Ke6 26.Re3+ Kd7 27.Qxf5+ Kd8 28.Bxg7 Rg8 29.Bf6 checkmate, blackburne - eddie43, Jerome Gambit thematic, ChessWorld.net, 2008; and
10.c3 Nh6 11.Qf3
c5 12.Qd3+ Kc7 13.Na3 a6 14.Nc4 d6 15.Rb1 Qh4+ 16.g3 Qh3 17.Nxe5 dxe5 18.Qe3 b6 19.b4 Rf8 20.bxc5 b5 21.Ba3 Ng4 22.Qg5 Rf7 23.Qh4 Qg2 24.Qxg4 Bxg4 White resigned, Gary_Seven - drewbear, JG3 thematic, ChessWorld.net 2008.
10...Qh4+
GriffyJr decides to mix it up.
11.Kf1 Qf6
Yikes. Instead, GriffyJr gets mixed up. I have no idea how or why. Much better was 11...Kc6.
12.dxe5+ Qxe5 13.Qxe5+
Oh, dear. The crusher was, of course, 13.Bf4 winning Black's Queen. These things happen in blitz games.
13...Kxe5
Even game.
14.Nc3 Nf6 15.g3
Nxe4 16.Bf4+ Kf5 17.Nxe4 Kxe4 18.Re1+ Kf5 19.Kg2 b6 20.Rhf1 Bb7+ 21.Kh3
Ba6 22.Rf2 c6
Play continues in a Queenless middlegame, about balanced, despite White's pawn minus (because of his activity).
23.Be5+ Kg6 24.Rd2 d5 25.Rd4 Bc8+ 26.Kg2 Bf5
27.h3 Rhe8
By focusing on development Black has pulled ahead. Still, if White can exchange Rooks, the possibility of a drawn bishops-of-opposite-color endgames is there.
Unfortunately, the computer can move lightning fast, while the human actually has to think - a disadvantage in blitz. White makes a tactical slip.
28.g4 Bxc2 29.h4 Be4+ 30.Kg3 Rxe5 White resigned
Why play the Blackburne Shilling Jerome Gambit, when there are perfectly playable responses for White within the regular lines of the Blackburne Shilling Gambit? Fun and familiarity are two reasons. The chance for a miniature is another.
DVYate - melazzini
FICS, 2017
1.e4 e5 2.Nf3 Nc6 3.Bc4 Nd4
The Blackburne Shilling Gambit.
4. Bxf7+
The Blackburne Shilling Jerome Gambit.
(Certainly White can play 4.0-0, 4.Nxd4 or 4.c3 instead, with advantage.)
4...Kxf7 5. Nxe5+ Ke8
The Database has 2,256 games with this move. White scores 58%.
"Best" seems to be 5...Ke6.
6.Qh5+ g6
Likewise, The Database has 1,356 games with this move. White still scores 58%.
7. Nxg6 Bg7
This Bishop moves seems, at first glance, to be a practical response - arranging to capture White's Knight after it takes the Rook at h8.
There are three things wrong with this move, however: the game continuation, the recommendation in the notes, and the straight-forward capture 7...hxg6 (best).
8.Nxh8+
At the very least White is now going to wind up ahead a Rook and 4 pawns. That is more than enough.
However, in all fairness, he also had the King hunt 8.Ne5+ Ke7 9.Qf7+ Kd6 10.Nc4+ Kc5 11.Qd5+ Kb4 12.a3+ Ka4 13.Nc3 checkmate.
8...Kf8
Holding out longer is 8...Ke7, but why bother?
9.Qf7 checkmate
I recently ran across an unusual line in the Blackburne Shilling Jerome Gambit that led to some interesting play - no doubt "enhanced" by the blitz time limit. White's creativity is to be applauded.
sahistonline - sriwijaya
5 5 blitz, FICS 2016
1.e4 e5 2.Bc4 Nc6 3.Nf3 Nd4
Blackburne Shilling Gambit.
4.Bxf7+
Blackburne Shilling Jerome Gambit.
4...Kxf7 5.Nxe5+ Ke6 6.c3 Kxe5
7.f4+
A rare, but interesting idea. Most recommended is 7.cxd4+ leading to an equal game.
7...Kxe4
Not 7...Kxf4 which gave White an immediate payoff after 8.Rf1+ (8.Qh5!?) Ke5 9.cxd4+ Kxd4?! 10.Qa4+ (10.Rf5!?) Ke5 11.d4+ Ke6 12.Qb3+ (12.d5!?) Kd6?! 13.Qd5+ Ke7 Black resigned, drakorg - borodin, FICS, 2002.
Best, as always in this line of the Blackburne Shilling Jerome Gambit, was to retreat the King, e.g. 7...Ke6.
8.d3+
White should not seek to immediately recover his sacrificed material, i.e. 8.cxd4 Kxd4 (8...Kxf4 leads to mate 9.Rf1+ Kg5 10.d3+ Kg6 11.Qg4+ Black resigned, jorgemlfranco - FoldZero, FICS, 2012; 8...d5!?) 9.O-O Bc5 10.Kh1 Qh4 11.Qa4+ Kd5 12.Nc3+ Kd6 13.d4 Bb6 14.Bd2 Nf6 15.Qb4+ Ke6 16.Rae1+ Kf7 17.Qc4+ Kg6 18.f5+ Kh5 19.Ne2 d5 20.Qd3 Ne4 21.Nf4+ Kg4 22.Qf3+ Kxf5 23.Ng6+ Kxg6 24.Qf7 checkmate, richiehill - steadylooking, FICS, 2011.
Best is to lock the enemy King in the center with 8.Qh5.
8...Kd5 9.Qh5+
The King is free. White might as well play 9.cxd4 and accept that he is worse.
But, remember, this is a blitz game!
9...Kc6 10.cxd4 Qe7+ 11.Kf2 Nf6 12.Qf3+ Nd5
In the BSJG, as well as in the Jerome Gambit, Black should always be concerned about blocking in his light-squared Bishop, which in turn blocks his Rook. Best, therefore, was 12...d5!?.
13.Nc3 Qd6 14.Be3 b6 15.Rac1 Kb7 16.Nxd5 c6 17.Nc3
Black has castled-by-hand on the Queenside.
White's advantage isn't his extra, doubled pawn, it is his great lead in development.
17...g5 18.Rhf1 Be7
Black would have done better to stick with his original idea 18...gxf4 19.Qxf4 Qxf4+ 20.Bxf4 d5 and White has only an edge.
19.Nb5
More direct was opening the position with 19.d5!?
19...Qd5
How ironic: Black erred earlier by putting his Knight on this square, now he is mistaken putting his Queen there. Then as now, ...d5!? was the move with near equality. The locked up pieces will never become part of the game.
20.Qxd5 cxd5 21.fxg5 Rf8+ 22.Kg1 Rxf1+ 23.Kxf1 Bd8
24.Nd6+ Ka6 25.Nxc8 Rxc8 26.Rxc8 Black resigned
