I suppose that I had expected a series of one-sided crushes, revealing brutal new Jerome Gambit refutations and uncovering scintillating dynamic defenses. It didn't turn out that way - at times the game looks like a positional dance. Let's take a look.
Wall/Stockfish - Crafty
centaur match, 2018
1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+
Okay, we all should be expecting a win for White - Stockfish is a higher-rated computer program than Crafty, and it is partnered with Bill, who is pretty knowledgeable when it comes to the Jerome Gambit.
4...Kxf7 5.Nxe5+ Nxe5 6.Qh5+
Bill played two games with White with Critter as his partner, two with Houdini, two with Rybka, and two wth Stockfish. Each time, he played one 6.Qh5+ game and one 6.d4 game. (6.Qh5+ scored 2 - 4 - 2, while 6.d4 scored 2 - 5 - 1.)
6...Ke6 7.f4 Qf6
Interesting: Black avoids 7...d6, which I have called the annoying or silicon defense, because it annoyingly drains much of the dynamism out of the position, and because it has been a primary choice of computer programs.
The Database has 208 games with 7...d6, with White winning 54% of the time. Instead, it has 133 games with 7...Qf6, with White winning 56% of the time.
8.Rf1 g6
Even computers like to kick the enemy Queen, but 8...d6, working toward development, is probably a little bit better.
9.Qh3+ Ke7 10.fxe5 Qxe5
White has to be careful here (he cannot castle, Black threatens the pawn at e4) and probably should not go into this line without some preparation. That is one suggestion that 6...Qf6 might have been a "book" move inserted by a knowledgeable human.
11.Qf3 Nf6 12.Nc3 c6
The computers tut-tut at this move, which keeps White's Knight off of d5 and prepares ...d7-d5, preferring the more complicated 12...Bd4. They still give Black the advantage after the text.
13.Ne2
Wow. This move has a tactical justification that comes from silicon thinking. Of course, now, 13...Qxe4? would blunder a piece to 14.Qxf6+. It takes a bit more work, however, to see that 13...Nxe4 is met by 14.d4 Bxd4 15.Bf4!?, when 15...Qa4+ 16.c3 d5 17.Nxd4 allows White to recover his sacrificed piece, with a roughly equal game. "Equal", as in mutually complicated; not "drawn".
Instead, a decade ago a couple of computers fought it out after the solid 13.d3 - 13...Bb4 14.Bd2 d6 15.d4 Qe6 16.O-O-O Ng4 17.d5 Qg8 18.Qg3 Ke8 19.dxc6 bxc6 20.Nb5 cxb5 21.Bxb4 Ne5 22.Rxd6 Nc4 23.Qg5 Nxd6 24.Qe5+ Qe6 25.Qxh8+ Kd7 26.Rd1 h5 27.Rxd6+ Qxd6 28.Bxd6 Kxd6 29.Qd8+ Black resigned, Fritz 8 - Fritz 5.32, D1N5TWD1, 2008.
13...Rf8 14.d4
I (sort of) warned you about this move (and the next) in the previous post.
14...Bxd4 15.Bf4 Qc5
It was okay, instead, to take the e-pawn with 15...Qxe4 16.Qxe4+ Nxe4 17.Nxd4, as after 17...g5!? the game would resolve itself into a pawn-plus Queenless middlegame favoring Black with 18.Be3 Rxf1+ 19.Kxf1 d5.
Interestingly enough, 17...d5 (instead of 17...g5!?) would have led to a human game where White ground down his opponent: 18.O-O-O Bg4 19.Rde1 Kd7 20.h3 Bf5 21.Bh6 Rf7 22.Nf3 Re8 23.g4 b6 24.gxf5 Rxf5 25.Nd2 Ng3 26.Rxe8 Kxe8 27.Rxf5 Nxf5 28.Bf4 h5 29.Nf3 Ke7 30.Ne5 Kf6 31.Nxc6 g5 32.Bc7 g4 33.hxg4 hxg4 34.Nxa7 g3 35.Nb5 Kg5 36.Bxb6 Kg4 37.a4 g2 38.Bg1 Kg3 39.a5 Nh4 40.Nd4 Kf4 41.a6 Nf3 42.Nxf3 Black resigned, Vlastous 2344 - Daboa 1799, Chessmaniac.com, 2016.
Black is still for choice, but this is how computer games go - a small slip here, a slight goof there... it all adds up.
16.O-O-O Be5 17.b4
Again: Wow.
Black is close to stabilizing his position with ...d7-d6, so White has to do something.
17...Qb5
This doesn't work. The other (stronger) computers suggest 17...Bxf4+ 18.Qxf4 Qxb4, when 19.Qe5+ Kd8 accents Black's uneasy King and unfinished development. Stockfish 9 then likes 20.Nf4, while Komodo 9 prefers 20.Nd4, both leading to a balanced game, or the slightest edge to Black.
A decade ago, a computer vs computer game saw this continuation: 20.Rxf6 Re8 21.Re6 Rxe6 22.Qxe6 Qf8 23.Qe5 Qe7 24.Qd4 b6 25.e5 Bb7 26.Nc3 c5 27.Qg4 Bc6 28.Qf4 h5 29.Nd5 Bxd5 30.Rxd5 Rc8 31.Qa4 Rc7 32.Qe4 Rc6 33.Qa4 Qe6 34.c4 Rc7 35.Qd1 a6 36.Qe2 Rc6 37.g3 b5 38.h4 bxc4 39.Qxc4 Qf7 40.Qb3 Kc7 41.Qd3 Kc8 42.Rd6 Rxd6 43.Qxd6 Qf1+ 44.Kd2 Qg2+ 45.Kc1 Qc6 46.Qd2 Kc7 47.Qa5+ Kb7 48.Qd2 Qe6 49.Qb2+ Kc6 50.Qg2+ Kb5 51.Qb7+ Ka5 52.Qc7+ Kb4 53.Qb7+ Kc3 54.Qb2+ Kd3 55.Qc2+ Kd4 56.Qb2+ Kd5 57.a3 Qg4 58.Qb7+ Kxe5 59.Qc7+ d6 60.Qe7+ Kd5 61.Qb7+ Ke6 62.Qb3+ Ke5 63.Qb2+ Qd4 64.Qe2+ Kd5 65.Qg2+ Qe4 66.Qd2+ Ke5 67.Qb2+ Kf5 68.Qb8 Qd3 69.Qc8+ Ke5 70.a4 Qc4+ 71.Kd2 Qxa4 72.Qc7 Qd4+ 73.Ke2 Qe4+ 74.Kf2 Qd3 75.Qe7+ Kd5 76.Qe8 Qf5+ 77.Kg1 Kd4 78.Qb8 Ke3 79.Qb3+ Qd3 80.Qf7 Kd2 81.Qa2+ Ke1 82.Qa5+ Qd2 83.Qa1+ Qd1 84.Qc3+ Ke2+ 85.Kh2 Qd4 86.Qa3 Qf2+ White resigned, Fritz 8-Fritz 5.32/D1N5TWD1 2008
18.Nd4 Bxd4
Have to get rid of that annoying Knight, but the exchange opens up dark squares in Black's position. White is now a bit better.
19.Rxd4
19...Kd8
Black's two problems are related: uneasy King and under-development. In the game he works to move the King to a safer place. He should have chosen, instead, 19...d5, unblocking the Bishop that blocks the Rook - a standard defense ailment in the Jerome Gambit. Then, after 20.Bg5 Bg4 21.Bxf6+ Kf7!? 22.Qf4 Kg8, White will probably be able to untangle his pieces, with an edge.
The problem is that White has a powerful response to the text move.
20.Bd6
Remember when the doctor told you "Okay, this is going to hurt a bit"?
20...Qg5+
What else?
21.Kb1 Re8
White now has a forced checkmate in 17 (!), but it is hard to find any acceptible alternative for Black.
Perhaps you are used to seeing computers display their brutal tactical skills, and this game is a good example of how many "positional" games are underlaid by tactical themes.
22.h4 Qxh4 23.e5 Black resigned
Simply brutal. The more you look, the more painful it becomes for Black.
