The power of the St. Georges Defence but how did I let the win slip? Advice most welcome.

[MJMcCready]

How did I go wrong here? What should I have done? I thought I had it won. I am black. Cinnamon 1.2c engine is my opponent, rated 1935[Opening "St. George Defence"] [ECO "C00"] 1.e4 e6 2.d4 a6 3.Bd3 b5 4.Nh3 Bb7 5.Bd2 Nf6 6.Qf3 c5 7.dxc5 Bxc5 8.Qf4 O-O 9.a4 b4 10.b3 Nc6 11.Qg3 Kh8 12.Ra2 d6 13.Bf4 e5 14.Bg5 h6 15.Qh4 Nd4 16.Bc4 Ne6 17.Bxe6 fxe6 18.Nd2 d5 19.Ke2 dxe4 20.Nxe4 Bxe4 21.Bxf6 Qxf6 22.Qxe4 Rad8 23.f3 Rd4 24.Qc6 e4 25.f4 Bd6 26.Qb6 e5 27.Rd1 Rxd1 28.Kxd1 exf4 29.a5 Rd8 30.Kc1 f3 31.g3 Re8 32.Rb2 e3 33.Qc6 e2 34.Qxe8+ Kh7 35.Qe4+ g6 36.Nf2 Qc3 37.Nd3 Bc5 38.Qb7+ Kg8 39.Qd5+ Kg7 40.Qd7+ Kg8 41.Qe6+ Kg7 42.Qd7+ Kg8 43.Qd5+ Kg7 44.Qb7+ Kg8 45.Qc8+ Kg7 46.Qc7+ Kg8 47.Qd8+ Kg7 48.Qc7+ Kg8 49.Qc8+ Kg7 50.Qe8 e1=Q+ 51.Nxe1 f2 52.Qd7+ Kg8 53.Qd5+ Kg7 54.Qd7+ Kg8 55.Qe8+ Kg7 56.Qd7+

[LawrenceCooper]

How did I go wrong here? What should I have done? I thought I had it won. I am black. Cinnamon 1.2c engine is my opponent, rated 1935[Opening "St. George Defence"] [ECO "C00"] 1.e4 e6 2.d4 a6 3.Bd3 b5 4.Nh3 Bb7 5.Bd2 Nf6 6.Qf3 c5 7.dxc5 Bxc5 8.Qf4 O-O 9.a4 b4 10.b3 Nc6 11.Qg3 Kh8 12.Ra2 d6 13.Bf4 e5 14.Bg5 h6 15.Qh4 Nd4 16.Bc4 Ne6 17.Bxe6 fxe6 18.Nd2 d5 19.Ke2 dxe4 20.Nxe4 Bxe4 21.Bxf6 Qxf6 22.Qxe4 Rad8 23.f3 Rd4 24.Qc6 e4 25.f4 Bd6 26.Qb6 e5 27.Rd1 Rxd1 28.Kxd1 exf4 29.a5 Rd8 30.Kc1 f3 31.g3 Re8 32.Rb2 e3 33.Qc6 e2 34.Qxe8+ Kh7 35.Qe4+ g6 36.Nf2 Qc3 37.Nd3 Bc5 38.Qb7+ Kg8 39.Qd5+ Kg7 40.Qd7+ Kg8 41.Qe6+ Kg7 42.Qd7+ Kg8 43.Qd5+ Kg7 44.Qb7+ Kg8 45.Qc8+ Kg7 46.Qc7+ Kg8 47.Qd8+ Kg7 48.Qc7+ Kg8 49.Qc8+ Kg7 50.Qe8 e1=Q+ 51.Nxe1 f2 52.Qd7+ Kg8 53.Qd5+ Kg7 54.Qd7+ Kg8 55.Qe8+ Kg7 56.Qd7+

Hanging a rook with check looked unnecessary.

[Paul Cooksey]

2...d5 is an improvement

[Christopher Kreuzer]

Is the final position actually drawn, or can Black evade the checks? Leaving the rook en prise is careless. Would be even more careless if the final position was still winning...

Does anyone know how computers 'limit' themselves to play to a 1935 (or any other) rating? I presume they do it in a different way to how humans would play 'generously' against e.g. a beginner they were trying to encourage to take up the game?

[Christopher Kreuzer]

Apparently:

https://www.chessprogramming.org/Cinnamon

Cinnamon 1.2 applied a kind of classical approach to generate ray-wise sliding piece attacks, captures and quiet moves. Cinnamon 2.0 already used line-wise occupancy lookups. Four pre-calculated attack arrays containing attack bitboards on ranks, files, diagonals and anti-diagonals were indexed by the square of the sliding piece, and the associated occupancy index of that line, determining the blockers affecting the attack set. Rather than to get the occupancy index from incremental updated rotated bitboards for each of the four line kinds, Cinnamon extracts line occupancies into a dense index range using multiplication and shift right, as mentioned in diagonals to rank or flip about the diagonal. The resulting 8-bit occupancy requires 512 KiB for all lookup tables, so considering the inner six bits would quarter the table sizes. In Cinnamon 2.3 the occupancy index may be computed by the PEXT instruction, if the executable is compiled with BMI2 enabled.

There is something hypnotic about that prose...

[Mick Norris]

2...d5 is an improvement

[Matt Mackenzie]

2...d5 is an improvement

Though not much of one

[MJMcCready]

Does it look like I pay enough attention to my opponents moves?

[John Upham]

Does it look like I pay enough attention to my opponents moves?

Yes