Jerome Gambit: While I Was Busy on the Computer This Morning...

 

While I was busy on the computer this morning, I decided to put all that work in the background and play a quick game online at FICS (Free Internet Chess Server).

I challenged a relatively weak computer program, Rusalka. Our game reminded me that nowadays the problem is not how to make a chess program smart, but, rather, how to make a smart computer dumber, so that humans have a chance against them. See "Jerome Gambit: Artificial Wha?".

perrypawnpusher - rusalka (c)

2 12 blitz, FICS, 2023

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

The Italian Four Knights Jerome Gambit.

5...Kxf7 6. Nxe5+ Nxe5 7. d4 Bd6

The Database has 83 games with this position. White scores 45%. Looking at my own games, I have scored 50% in 4 tries. This is all in light of the computers' evaluation of the position as being about  4 1/2 pawns better for Black. (What would International Master Erik Kislik think?)

8. dxe5 Bxe5 9.O-O 

This move (new for me) was only Stockfish's second choice (30 ply), as I found out after the game.

Stockfish 15's top choice in post-game analysis was 9.Ne2 which I had played in perrypawnpusher - LttlePrince, Italian Game Classic, Chess.com, 2020 (0-1, 29) and perrypawnpusher - andrewLLL, Italian Game Battlefield, Chess.com, 2020}(1-0, 18). 

The computer's third choice was 9.f4, which I had played in perrypawnpusher - joseluislopez, 10 3 blitz, FICS, 2012 (0-1,55).

The computer's fourth choice was 9.Bg5, which I had played in perrypawnpusher - lixuanxuan, 14 0 blitz, FICS, San Jose, California US 2014 (1-0, 22). 

9...Bxc3 10.bxc3 Nxe4 

Ouch. A painful oversight. Yet, a look at The Database shows 8 previous games with this mistake, with White scoring only 63%.

By comparison, lichess.org has 19 games, with White also scoring 63%

(See this post's final comment, below.)  

11.Qd5+ Kf8 12.Qxe4 d5 13.Ba3+ Kf7 14.Qf4+ Kg8 15.Rad1 

I admit that I was hoping to be able to next play the cheapo 16.Rxd5 Qxd5 17.Qxf8#.

Instead, 15.Rae1 with the plan 16.Re7 was more logical. 

15...c6 

Stockfish 15.1 prefers the defensive 15...Qf6 instead, but its followup for White is nothing that I would have found: 16.Qxc7 h6 17.c4 Qc3 18.Rd3 Qxc4 19.Qe5 Qe4 20.Qa1 Bf5 21.Bb2 Rh7 22.Re1 Qc4 23.Rc3 Qb4 24.Rb3 Qc5 25.Ba3 Qc6 26.Qd4 Rc8 27.Rbe3 Material is even, but White is better.

16.c4 

Still looking for that cheapo. Still 16.Rfe1 with the idea of 17. Re7 was best.

16...Qa5 

Ouch. Best was 16...Qf6 17.Qxf6 gxf6 18.cxd5 cxd5 19.Rfe1!? when any hope that the Bishops-of-opposite-colors ending which is approaching might hold a draw would be too much fantasy.

17.Qf8 checkmate

Some information about rusalka that I gathered from the FICS website

According to east-Slavic folklore, a Rusalka is a water spirit or water nymph. Rusalkas appear as beautiful young women who try to lure men into the water, where they will drown them.

Phalanx XXV, easy level 80, nodes per second limit = 260, CPU ARMv7 4-core 1.5 GHz with 2GB RAM running Lubuntu

easy levels are hardware/cpu load independent

i try to emulate human-like blunders