Jerome Gambit: The Human Move (Part 1)

Thinking about Maia Chess, mentioned in the previous post, I wondered: If its goal is "to play the human move - not necessarily the best move", and if the project wishes to "begin to algorithmically capture what kinds of mistakes players at specific skill levels make" - has it ever attempted the Jerome Gambit (1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+)?

Poring through several million (!) of maia games*, I found two Jeromes by the lesser-skilled engine, maia1, and two by the highest-skilled, maia9.

They were all very, very human.

maia1 - yung-m90

10 3 rapid game, lichess.org, 2022

1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+ 

4...Kxf7 5.Nxe5+ 

Another game continued 5.Ng5+ (You can hardly get more "human" than the "Face Palm Variation"! It appears that maia1's opponent is equally skilled) Ke8 6.Qf3 Nh6 7.d3 Nd4 8.Qh5+ Kf8 9.Nc3 c6 10.Be3 Nxc2+ 11.Kd1 Nxa1 12.Bxc5+ d6 13.Be3 Bg4+ 14.f3 Bxh5 15.Kd2 Qb6 16.Rc1 Nf7 17.Nxf7 Bxf7 18.Bxb6 axb6 19.Rxa1 g6 20.b4 Kg7 21.Rg1 c5 22.bxc5 bxc5 23.Ra1 Be8 24.Nd5 Bf7 25.Nc7 Ra5 26.Ke3 Rc8 27.Nd5 Bxd5 28.exd5 Rca8 29.Ke4 Rxa2 30.Rf1 Ra1 31.Rxa1 Rxa1 32.Ke3 Ra2 33.g3 Rxh2 34.f4 Rh3 35.Kf3 exf4 36.Kxf4 b5 37.Kf3 b4 38.Ke2 b3 39.Kd2 Rh2+ 40.Kc3 b2 41.Kb3 b1=Q+ 42.Kc3 Rc2 checkmate, maia1 - CatrinaM, 7 5 rapid, lichess.org, 2022. 

5...Nxe5 

Or 5...Kf8 6.Nxc6 bxc6 7.d4 Be7 8.Qf3+ Nf6 9.O-O d5 10.Bg5 h6 11.Bxf6 Bxf6 12.e5 Ba6 13.exf6 Bxf1 14.fxg7+ Kg8 15.gxh8=Q+ Kxh8 16.Kxf1 Qe7 17.Nd2 Re8 18.Qe3 Black resigned, maia9 - Jam8sJ0yce, 5 2 blitz, lichess.org, 2023

6.Qh5+ 

Otherwise 6.d4 Bxd4 7.Qxd4 Qf6 8.O-O Nf3+ 9.gxf3 Qxd4 10.Be3 Qxb2 11.Nd2 Qxc2 12.Rac1 Qxa2 13.Rxc7 Ne7 14.Nc4 Ke6 15.Rd1 d5 16.exd5+ Nxd5 17.Re1 Nxc7 18.Bf4+ Kf5 19.Be5 Qxc4 20.Bxg7 Rg8 21.Re5+ Kg6 22.Rg5+ Kxg5 23.Kg2 Rxg7 24.h4+ Kf6+ 25.Kh2 Black resigned, maia9 - soyunbotenajedrez_20, bullet, lichess.org, 2022 

6...g6 7.Qxe5 Nf6 

Readers of this blog might not expect outsiders to know Jerome Gambit theory: best is Whistler's 7...Qe7; historic is Blackburne's 7...d6.

8.Qxc5 Nxe4 9.Qc4+ Kg7 10.Qxe4

This is so "human" that it hurts.

10...Re8 11.Qxe8 Qxe8+ 12.Kd1 d5 13.Nc3 Bg4+ 14.f3 Bh5 15.Re1 Qc6 

Black's move should drop a piece and allow the computer to equalize with 16.g4, winning a piece.

16.d3 g5 17.g4 

Better late than never? Not really.

17...Bg6 18.Bxg5 d4 19.Ne4 Bxe4 20.Rxe4 Kg6 21.Be7 Re8 22.Bh4 Rxe4 23.fxe4 Qe8 

maia1 has a Rook, a Bishop, and 2 pawns for its Queen. Nonetheless, Stockfish 15.1 gives yung-m90 the edge.

In fact, the human now gives a lesson on the dominance of Her Majesty over the enemy pieces.

24.Ke2 Qf7 25.Rf1 Qxa2 26.Rf6+ Kg7 27.Rf3 Qxb2 28.e5 Qxc2+ 29.Kf1 Qd1+ 30.Be1 Qxf3+ 31.Bf2 Qxd3+ 32.Kg2 Qe4+ 33.Kg1 d3 34.Bxa7 d2 35.Bb8 d1=Q+ 36.Kf2 Qef3 checkmate

* I had it easy. According to the profile of Maia Chess 1100 at lichess.org, "This version was trained by learning from over 10 million Lichess games between 1100s."

[to be continued]

Jerome Gambit: I'm Really Getting Too Old For This Fast Stuff... (Part 2)

 

[continued from the previous post]

Certainly a blitz game (unknowingly, at the time, against a computer) was a diversion from the events of the day, but it was not the best choice. Tennis, anyone?

perrypawnpusher - guestM

5 1 blitz, FICS, 2023

26.Kh2 Qh6+ 27.Kg1 Rxg3 28.Qe4 Bh3 29.Rff2 

My Jerome Gambit has been a wild ride, but not a successful one.

It is time for my opponent to finish me off.

29...Bxg2 

Forcing, but not best.

30.Rxg2 Rxg2+ 31.Rxg2 d5 

Those annoying pawns. (Page Dante Alighieri.)

32.Qe5 Rg8 

Odd. 

Stockfish 16's analysis starts off with the recommended  32...c4, which seems more to the point.

My opponent seems to have lost its way.

33.Qxd5 

What else?

After the suggested improvement 33.Rg5 Rd8 the pawn is no longer available, although after the game I let Stockfish analyze for a while and it struggled to convert Black's advantage.

33...Qxf4 34.Qxc5 Qc1+ 35.Kh2 Qf4+ 36.Kg1 Qc1+ 

I know that in the past computer programs had difficulty with the endgame - often taking calculating time to assess what humans can understand at a glance (e.g. unopposed pawns advancing) - but that was then and this is now. Besides, this isn't quite an endgame, is it? 

Or is it?

37.Kh2 

Giving Black's Queen choices. Too many choices? Perhaps.

37...Qf4+ 

Again, Stockfish 16 prefers 37...Qh6+, but it still has a hard time bringing the whole point home quickly.

38.Kg1 

At this point the game was drawn by repetition. 

'Tis a puzzlement.

And, for me, an escape. I was happy to collect the half point.

By the way, it turns out that my opponent was Maia Chess, "a human-like neural network chess engine"

Maia’s goal is to play the human move — not necessarily the best move. As a result, Maia has a more human-like style than previous engines, matching moves played by human players in online games over 50% of the time..

Maia is an ongoing research project using chess as a case study for how to design better human-AI interactions. We hope Maia becomes a useful learning tool and is fun to play against. Our research goals include personalizing Maia to individual players, characterizing the kinds of mistakes that are made at each rating level, running Maia on your games and spotting repeated, predictable mistakes, and more...

Maia is particularly good at predicting human mistakes. The move-matching accuracy of any model increases with the quality of the move, since good moves are easier to predict. But even when players make horrific blunders, Maia correctly predicts the exact blunder they make around 25% of the time. This ability to understand how and when people are likely to make mistakes can make Maia a very useful learning tool.

 

My interest in the Jerome Gambit, as I have pointed out in the past, comes from investigating "errors in thinking", so perhaps Maia and I are not so far apart, after all. 

Jerome Gambit: I'm Really Getting Too Old For This Fast Stuff... (Part 1)

After participating as a subject in the latest round of an ongoing study on dementia - you would think that my devotion to the Jerome Gambit would settle the matter, preemptively - I was restless and decided what I really needed to do was play a 5-minute online game of chess. 

Against a computer program.

Although I did not realize it at the time.

[Insert your own joke here.]

All of the comments by Stockfish came after the game was completed.

perrypawnpusher - guestM

5 1 blitz, FICS, 2023

1.e4 e5 2.Nf3 Nc6 3.Bc4 Nf6 4.Nc3 Bc5 5.Bxf7+ 

The Italian Four Knights Jerome Gambit.

I have played it 71 times, scoring 77%.

Not bad, but in the regular Jerome gambit (1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+) - which I have played 436 times - I have scored 81%.

5...Kxf7 6.Nxe5+ Nxe5 7.d4 Bxd4 8.Qxd4 d6 

9.f4 Nc6 10.Qd3 Re8 

I have also faced 10...Be6 in and perrypawnpusher - hklett, 2 12 blitz, FICS, 2010 (1 - 0, 18) and perrypawnpusher - obmanovichhh, 14 0 blitz, FICS, 2011 (1-0, 34) 

11. O-O Kg8 

Looking over things afterward, it is clear that the computer is doing well, especially since it is unlikely that the "refuted" Jerome Gambit would be deeply covered in its opening "book". 

There is bound to be a reckoning ahead, however.

The issue is not one for the strong contemporary chess programs (e.g. Fritz, Houdini, Komodo, Rybka, Stockfish), but with some of the simpler programs there is the problem that they still play consistently well - so much so that in order to make it "fun" to play against them by the average chess player, they need to be handicapped in some way. For a discussion, see the posts "Artificial Ignorance (Parts 1 and 2)"

Also see Part 2 of this topic, referencing the Maia Chess project.

Meanwhile, instead of the text, I have also seen 11...Nb4 in perrypawnpusher - abczyx, 10 5 blitz, FICS, 2010 (1-0, 39) and 11...a6 in perrypawnpusher - HGBoone, 6 12 blitz FICS, 2010 (1-0, 46). 

12.b3 Nb4 

This is what I mean.

Instead, after 12...d5 13.e5 (if 13.exd5, then 13...Nb4) d4 14. Ne2 Nd5 15. Ng3 Qd7 16. Bd2 Qf7 17.Rae1 Bd7 Black would be better.

13.Qe2 

Sloppy. 13.Qc4+ would lead to the win of the Knight and an even game. 

13...Bg4 

guestM continues to play "kick the Queen", overlooking the consistent and better 13...d5.

14.Qc4+ 

Given a second chance, I grabbed it.

14...Be6 

Again, the move was 14...d5

15.Qxb4 

Thanks to the slipup, I was suddenly better. 

Of course, in a 5-minute game, lots of things happen "suddenly".

15...c5 

On principle, I do not think that grabbing the b-pawn now is a good idea. See "Jerome Gambit: He Who Takes the Queen's Knight Pawn..." and "Jerome: Risking the Gutter".

16.Qa3 

Stockfish 16 disagrees with me, based on concrete analysis: 16.Qxb7 Rb8 17.Qa6 Bc8 18.Qd3 Bb7 19.Bb2 Nxe4 20.Rae1 Nxc3 21.Qxc3 Qd7 and White is better. 

16...a6 17.Be3 b5 18.Qc1 b4 19. Ne2 Nxe4 

I am used to having the annoying pawns that advance and create disturbances - not facing them.

20.Ng3 Nxg3 21.hxg3 Bg4 22.Qd2 Qf6 23.Rad1 

An oversight that should have cost me the game. The Rook was better  placed at e1.

Blitz requires faster brain calculations than I am usually able to produce. 

You could say that both guestM an I were having "cognitive issues".

23...Rad8 

This had to be the kind of glitch seen on move 12.

24.Qd5+ 

Simply moving the Rook on d1 was the right idea. 

Instead, I was in a bit of a panic, rewriting the old reflection to be "panic sees check, panic gives check" ("panic" instead of "patzer").

24...Kh8 

Undeserved good fortune for me. The Queen should have blocked at either e6 or f7.

25.Rd2 

Throwing a piece away.

25...Rxe3 

At this point I was sincerely regretting my commitment to share every Jerome Gambit I played, win, lose or draw...

[to be continued]

Jerome Gambit: Invisible Lines of Play

With an infinite time control, players would probably still not see "everything", but as the time control became shorter and shorter, things would begin to disappear, to drop over the event horizon. 

Blitz games feature a whole lot of invisible lines of play - invisible to one or both of the players; perhaps still visible to those who play over the games.

Scbmine - sevags

5 3 blitz, lichess.org, 2023

1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+ 

4...Kxf7 5.Nxe5+ Nxe5 6.d4 

This is the main alternative to 6.Qh5+.

6...Qh4 

The strongest response, going back to Sorensen - X, Denmark, 1884, (1-0, 27) and which I have referred to as a "pie-in-the-face variation".

7.dxc5 

Frequently played, getting rid of the attacking enemy Bishop. Not as many players choose 7.O-O, "castling into it", Stockfish's preference.

7...Qxe4+ 8.Be3 Ng4 

Black adds another threat, ganging up on the Bishop at e3 while maintaining his threat to the pawn at g2.

White's response is best, even if it still leaves him with the worse game.

9.O-O Nxe3 10.fxe3+ Nf6 11.Nc3 Qxe3+ 12.Kh1 

The excitement has died down, and Black has an extra piece. After he captures at c5, he will have an extra pawn as well.

With his next move, however, he begins to lose the thread of the game.

White is nowhere near willing to surrender.

12...c6

What is wrong with this?

It helps White jump start his development and counter-attack.

13.Qh5+ Kf8 

A bit safer was 13...Kg8.

Now White is better.

14.Rae1 

The Rooks look scary on the e- and f-files, but this move is an oversight. (Temporizing with 14.Qh4 was better.)

As happens in a blitz game, however, it goes unnoticed by either player.

14...Qd4

A doubly painful move.

First, it allows 15.Re8# or 15.Qe8#.

Secondly, it overlooks that the White Rook can be taken: 14...Qxe1 15.Rxe1 Nxh5 and he would be up a Rook and a Bishop.

15.Ne4 

Still winning.

15...d6 16.Nxf6 g6 17.Nxh7+ Black resigned

Black is powerless in the face of his opponent's firepower - something that he did not forsee 10, or even 5, moves ago.

Jerome Gambit: Baffling

From the What's Up in Chess? Chess News and Chess Gems website comes the entry "What to do against BAFFLING Chess openings" which teases

Facing off against a player who plays fifty 30+0 games a day is a tall task. What kind of person plays fifty classical games everyday?

It turns out, the answer is, in part, Someone who plays the Jerome Gambit, of course.

Here is the game that is presented, with my addition of comments and diagrams. It is an entertaining battle to the bitter end.

Trage2000 - mikethedoge

30 0 classical, lichess.org, 2022

1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+ 

4...Kxf7 5.Nxe5+ Kf8 

This position is as old as analysis presented by Jerome in the April 1874 issue of the Dubuque Ches Journal.

6.Qh5 

The Banks Variation.

6...Qf6 

Stockfish prefers this Queen to be on e7, at which point it recommends that the other Queen retreat to f3.

7.Nxd7+ Bxd7 8.Qxc5+ Nge7 9.d3 Qe5 

White is a bit better, with 3 pawns for the piece (and a safer King).

10.Qxe5 Nxe5 11.Bf4 N7g6 12.Bg3 c6 

13.Na3 Ke7 14.h4 h5 15.Bxe5 Nxe5 16.O-O-O 

Likely 16.d4 was better.

16...Rhf8 17.f3 b5 18.Rde1 Ng6 19.d4 Rf4 20.d5 cxd5 21.exd5+ Kd6 

22.Re4 

A tactical slip both players overlook. Now Black can simply exchange Rooks ad then capture the h-pawn.

22...Rf5 23.Rd4 

Better 23.Nxb5+.

23...Rxd5 24.Rxd5+ Kxd5 

25.g3 Kc6 26.Rd1 Re8 27.Nxb5 Kxb5 28.Rxd7 Re7 29.Rd6 Ne5 30.f4 Nc4 31.Rd5+ Kc6 32.Rd1 Re3 

Black's pieces have out-duelled the pawns.

33.b3 Nd6 34.Rg1 Ne4 35.g4 hxg4 36.Rxg4 Rg3 37.Rxg3 Nxg3 38.Kd2 Kc5 39.c3 Kd5 40.Ke3 Nf5+ 

White can afford to let the h-pawn go. He would do better with his King closer to his Queenside pawns.

As the game goes, the Kingside is liquidated, giving Black all the play on the Queenside. Still, things remain even.

41.Kf3 Nxh4+ 42.Kg4 Ng2 43.Kg5 Ne3 44.Kg6 Ke4 45.Kxg7 Kxf4 46.Kf6 Ke4

47.c4 Nxc4 

One way to split the point.

48.bxc4 Kd4 49.Ke6 Kxc4 

The fact that the remaning pawns are Rook pawns helps make the position so drawish. Yet, Black persists.

50.Kd6 a5 51.Kc6 a4 

Do you know your endgame? What move must White play now?

52.Kb6 

White needed to chase Black's King in front of his own pawn, and trap him there e.g. 52.Kd6 Kc3 53.Kd5 a3 54.Ke4 Kb2 55.Kd3 Kxa2 56.Kc2.

52...a3 53.Ka5 Kc3 54.Ka4 Kb2 55.Kb4 Kxa2 

The difference from the above analysis is that Black's King can now get out of the way of his pawn.

56.Kc3 Kb1 57.Kb3 a2 White resigned

Wow.

Jerome Gambit: White Queen, Black Queen

Here is another game by ryuugu-rena. That is not surprising.  He has 682 games in The Database, scoring 56%.

Here, his Queen sets the tempo. The defender's Queen can not keep up.

ryuugu-rena - Arusha

10 0 rapid, lichess.org, 2023

1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Bxf7+ 

4...Kxf7 5.Nxe5+ Nxe5 6.Qh5+ Ke6 

Black does not want to give up the Knight.

7.f4 g6 

Arusha has faced the Jerome before (9 games in all), and it has not been kind to him.

In this position, it is so tempting to kick the Queen.

8.Qxe5+ Kf7 9.Qxh8 

Capturing the Rook at h8 is thematic against a number of defenses, including Blackburne's and Whistler's. See the post "Jerome Gambit: What About the Rook?"

In this position, White's Queen can escape, and Black's Queen's counter-attack (see below) is not sufficient.

White can also settle for a smaller material gain: 9.Qxc5 Qh4+ (9...d6 10.Qc3 Bg4 11.Qxh8 Qe7 12.Qd4 Nf6 13.O-O h5 14.e5 dxe5 15.fxe5 Black resigned, ryuugu-rena -Omr1984m, 3 0 blitz lichess.org, 2023) 10.g3 Qg4 11.Qd5+ Ke8 12.Qe5+ Qe6 13.Qxh8 Qxe4+ 14.Kf2 Qxh1 15.Qe5+? (15.Qxg8+) 15...Ne7?! (15...Kf8) 16.Nc3 (16.Qh8+) 16...Qxh2+ 17.Kf3 Qh1+ 18.Ke3 d6 19.Qh8+ Kd7 20.Qg7 Qg1+ 21.Kd3 Qxg3+ 22.Kc4 Qxf4+ 23.Kb3 Ke8?! 24.d3? (24.Qxe7+ Kxe7 25.Nd5+) 24...Qf5? (24...Qf7+) 25.Qh8+ Qf8 26.Qxh7 Be6+ 27.Ka3 c6?! 28.Bg5 Bf5? 29.Re1 b5 30.Rxe7+ Kd8 31.Rf7+ Ke8 32.Rxf8+ Kxf8 33.Qh8+ Kf7 34.Qxa8 c5 35.Qxa7+ Ke6 36.Nxb5 (36.Qe7#) 36...d5 37.Qb6+ (37.Qe7#) 37...Ke5 38.Qxc5 (38.Qd6#) 38...Be6 39.Nc7 (39.Qd6+ Kf5 40.Qf4#) 39...Bf5 40.Qxd5 checkmate, ryuugu-rena - YaSanya047, 3 0 blitz, lichess.org, 2023) 

9...Kf8 

Or 9...Qh4+ 10.g3 Qh5 11.c3? (11.Qe5) 11...b6? (11...Qf3) 12.d4? (12.Qe5) 12...Ba6 13.Kf2 Bd6?! (13...Qe2+ 14.Kg1 Qf1#) 14.Re1 Qxh2+ 15.Kf3 Qh5+ 16.Ke3 Bb7 17.g4 Qh3+ 18.Kd2 Qh2+ 19.Re2 Bxf4+ 20.Kd1 Qh1+ 21.Re1 Qf3+ 22.Re2 Qf1+ 23.Re1 Qf3+ 24.Re2 Bxc1 25.Qxh7+ Kf6?! (25...Kf8) 26.e5+?! (26.Nd2) 26...Kg5 27.Qxd7? (27.Nd2) 27...Qf1+ 28.Re1 Bf3+ 29.Kc2 Qxe1 30.Kb3 Bd1+ 31.Ka3 Bxb2+ 32.Kxb2 Qe2+ 33.Ka3 Qc2 34.Qb5 a6 35.e6+ White won on time, ryuugu-rena - daniel_jason, 3 0 blitz, lichess.org, 2023 

10.d4 Be7 

White's "Jerome pawns" are impressive. With the material advantage of the exchange and two pawns - plus a safer King - he is clearly better.

11.Qxh7 d6 12.Qxg6 Be6 13.O-O Bf7 

The reinforcements arrive too late.

14.Qg3 Nf6 15.e5 dxe5 16.fxe5 Nh5 17.Qf3 Qxd4+ 18.Be3 Black resigned

Black's Queen can not hold things together, for example 18...Qc4 19.Nd2 Qe6 20.Qxh5.