So, how do you beat the Beth Harman bot at Chess.com?
Well, I have already pointed you to an enjoyable video of the game where Jonathan Schrantz used the Jerome Gambit to defeat "her".
I thought it would be useful to go over the game, and point out a few of the relevant ideas.
Schrantz, Jonathan - Harman (22) bot, Beth
blitz, Chess.com, 2020
Jonathan Schrantz has challenged a number of bots, using unorthodox openings successfully against them. For this game, he noted
...I am going to try to beat Beth Harmon with the worst opening in all of chess. I am going to try to take her down with the Jerome Gambit... Beth is actually going to be the weakest engine that I have played.
1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+
Quite possibly, 6...Kf8 is "book", which would be a very reasonable choice.
7.Qxe5 d6 8.Qg3 Kf7 9.c3
Central to Schrantz's strategy.
9...Nf6 10.d4 Bb6 11.f3 Be6
There's a lot to look at here.
Over the years, on this blog, I have presented a number of human vs human games where White constructs the triangle of pawns d3-e4-f3 and then waits for Black to do something with his Knight-for-2-pawns advantage. While Black is transitioning from defender to attacker, White takes small steps to creep up on him...
But, Schrantz's "shrine" pawn structure, is that new? I thought so, but when I checked with The Database, I did a search and found 856 games with the pawns set at c3-d4-e4-f3. Hmmmm....
So I added the Black Bishop at b6, the Black pawn at d6, and the White Queen at g3, to better refine my search through The Database. I found 30 games. Not a lot, but still a number of predecessors.
So I added the Black King at f7 to the position for my search, and found 6 games. Equally important, White scored 2 - 4 in those games. Schrantz was making an interesting contribution to Jerome Gambit - and he was just getting started.
By the way, a quick search of The Database showed that 8...Kf7 was a novelty. The idea of moving the King up to facilitate castling-by-hand is part of much of defensive strategy against the Jerome Gambit, but I found it interesting that with over 15,700 games with 1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+, Black could still come up with a novelty on move 8.
What next? Well, Schrantz obviously knew something about Beth's quirks
She actually has this tendency against the Jerome Gambit to go kind of wild on the kingside. That's what we are looking forward to.
12.a4 c6 13.O-O h6
As foretold.
15.f4
To open the f-file, where the enemy King is standing.
15...Bc4
Save the Bishop - in case White gets in f4-f5 - and threaten to win the exchange. But the human is ready.
16.Nd2 Bxf1 17.Rxf1
White is down a Rook for a couple of pawns, but he is probably better. At any rate, his position is full ideas.
17...g4
To close off the f-file.
18.e5 Re8
White's attack looks dangerous.
Stockfish 11 suggests that Black complicate things, so that White will eventually sue for peace by repeating the position, i.e. 18...Qd7 19.exf6 Qf5 20.Bf2 Rhg8 21.Nc4 Bc7 22.d5 c5 23.b4 b6 24.Qh4 Qxf6 25.Qh5+ Qg6 26.Qh4 Qf6 etc.
I can't see a human taking Stockfish 11's advice.
19.Nc4 Bc7 20.exf6 Rg8
21.Ne5+
Many of us will play a lot of games without finding a move as enjoyable as this one.
21...dxe5 22.fxe5
Those connected, passed "Jerome pawns" look formidible. Black's next move looks like the kind of thing that engines play when they are "handicapped" to make a less challenging opponent; sometimes the program just cuts down on its "thinking" time, sometimes it randomly distributes mistaken play. But this was the Beth bot playing at her top ability, a rating on the site of 2700.
22...a5
Black defends, but she is without hope.
25.e6+ Kxe6 26.Qe1+ Kd5 27.c4+ Kxc4 28.Qe6+ Kxd4 29.Rd1+ Kc5 30.Be3+ Kb4 31.Rd4+ Kc5 32.Re4 checkmate
And that is why we play the Jerome Gambit.
Thank you, Jonathan Schrantz.