Here is the start of my ten game match between the new CGPro2024 and the original CGP. Both machines set to the Classic(London) book and I’m guessing this is the same book on each machine. Have e-mailed Millennium to confirm if this is the case and waiting for their reply.
The CGPro2024 is set to its default ‘Normal’ play style, the older CGP does not have settings for style of play. I’m using a time control of 40 moves / 2 hours then 20 moves/ hour so it will take a little time to complete.
I expect the final result will be close if not a draw, but there may be some interesting games along the way considering they are slightly different programs. The CGPro2024 has extra endgame knowledge but this may make it slower in some positions, so lets find out.
Game 1/10
[Event "CGP2024 V CGP 40m/2h 20m/1h Classic Book"]
[Round "1"]
[White "CGP"]
[Black "CGP2024"]
[Result "1/2-1/2"]
[ECO "E59"]
1.d4 Nf6 2.c4 e6 3.Nc3 Bb4 4.e3 O-O 5.Nf3 d5 6.Bd3 c5 7.O-O Nc6 8.a3 Bxc3 9.bxc3 dxc4 10.Bxc4 Qc7 11.Bd3 e5 12.Qc2 Re8 13.Nxe5 Nxe5 14.dxe5 Qxe5 15.f3 Be6 16.e4 c4 17.Be2 Qc5+ 18.Kh1 Rad8 19.Rb1 b5 20.a4 a6 21.axb5 axb5 22.Qb2 Bd7 23.Rd1 Bc6 24.Rd4 Qe5 25.Be3 Ra8 26.Rbd1 Qc7 27.Qd2 Re6 28.Rd8+ Rxd8 29.Qxd8+ Qxd8 30.Rxd8+ Re8 31.Rxe8+ Nxe8 32.g4 f6 33.Kg2 Kf7 34.h4 h6 35.Kf2 Nc7 36.Kg3 Ne6 37.h5 Ng5 38.Bd2 Ke7 39.Bc1 Bd7 40.Ba3+ Kd8 41.Bf8 Ne6 42.Bd6 Nc7 43.f4 Bc6 44.e5 Ne6 45.Kf2 Ke8 46.Bf3 Bd7 47.f5 Ng5 48.e6 Nxf3 49.Kxf3 Bc6+ 50.Ke3 Bg2 51.Ke2 Be4 52.Ba3 Bd3+ 53.Ke3 Bc2 54.Bd6 Bd1 55.Kf4 Bc2 56.Ba3 1/2-1/2
The game is pretty much balanced until move 44 when the CGP2024 playing black makes a bad move. This is easy for me to say after having SF analyse the game while a drank my coffee 😊. It’s quite an interesting position though and worth looking at to appreciate that neither of the CGPros would be able to calculate far enough ahead in the time available to make the best move. Here is the position after CGP makes the move 44.e5:
White has a slight advantage and his pawn on e5 is causing Black some concern. For example, he would not want to move 44…PxP? White would take back with his Bishop and Black’s Queen-side Pawns are in trouble.
Instead, the CGP2024 moved 44… Ne6 maybe in an attempt to stop the e5 pawn advancing and prevent white’s bishop moving to f8. The problem with this move (after seeing the PC engine analysis and thinking about the position) is that this knight will eventually be attacked by White’s f pawn and will have to move away opening up a route for White’s Bishop to get to the f8 square. A possible continuation is: 45.exf6 gxf6 46.Bf3 Be8 47.f5 Kd7 48.Bb4
Now we can see Black has to move the knight or lose it, thus opening up the diagonal for White’s Bishop on b4 and Black will eventually lose his Pawn on h6. That’s 12 ply until the capture and I noticed that neither machine was able to fully explore to this depth in the time available.
Alas, White also made a bad move a blunder in fact. After Black had moved 48…Nxf3 we have this position:
The CGP moved 49.KxN which results in a likely draw, but no chance of these machines seeing that. I think we are able to look at the position and judge it’s a draw though. That being said, after move 56 the CGP2024 offered a draw so I accepted on behalf of the CGP!
So I think all considered this was good game.