An Engine-to-Engine Opening

In my previous post, TCEC Season 6 - Chess960, I said I would look at a game between the first and second placed engines, Stockfish and Houdini. I should have said I would look at a position, because it would take many hours to understand the entire game. Engine-to-engine chess games aren't the most transparent battles.

The two engines first met in the sixth round of the 28-round event, Stockfish getting the White pieces. The start position was SP016 BBNQNRKR, shown in the top diagram. The most striking characteristic of this SP is the Bishop pair on the a-/b-files, aimed at the enemy King, which is lodged between the two Rooks. There are three such positions out of the 960 total, with the Queen on one of the c-, d, or e-files, and the Knights taking the last two squares.

The Bishops will be developed by moving the b- & c-Pawns, while the natural square for the e-Knight is f6, leaving d6 for the c-Knight. Because the Queenside (a-side) will be loosened by Pawn moves, castling O-O looks more likely than O-O-O, but the f-Rook must first get out of the way.

The first moves were 1.c4 c5, both sides opening diagonals for the Queen and Bishop, while looking at a subsequent push of the d-Pawn by two squares. This was followed by 2.Nf3 b6, both players conforming to the plan I outlined earlier.

Fast forward to move 15, where White has just castled and Black has played ...Nf8-e6. The four Bishops are active without having moved, the White Knights are looking at the squares b5, d5, and f5, and the Black Knights are positioned both defensively (on d6) and offensively (e6).

Only one of the four Knights is on what I thought would be its natural square, meaning the players have done some maneuvering to bring those pieces to other squares. White's Knights make a better impression than Black's, because the defensively placed Knight on d6 is awkwardly blocking the open d-file.

Black is prepared to castle O-O. In fact, this never happened. The engine eventually played ...h5 instead, moving the Rook down the h-file and leaving the King on g8, where it would have been after O-O.

The Pawn structures are hard to judge. Both players have an isolated Pawn and a semi-open file, where the f-file looks more useful to White than the b-file to Black. Black, however, has the Pawn push ...a7-a5-a4, threatening to weaken White's Pawns on that side of the board.

All in all, I prefer White, but there is still plenty of play in the game. The full game score is copied below.

[Event "TCEC Season 6 - FRC"]
[Site "http://tcec.chessdom.com"]
[Date "2014.07.02"]
[Round "6"]
[White "Stockfish 260614"]
[Black "Houdini 4"]
[Result "1-0"]
[SetUp "1"]
[FEN "bbnqnrkr/pppppppp/8/8/8/8/PPPPPPPP/BBNQNRKR w KQkq - 0 1"]
[PlyCount "102"]

1.c4 c5 2.Nf3 b6 3.d4 Ncd6 4.dxc5 bxc5 5.b3 Ne4 6.Nd2 N4f6 7.Re1 d6 8.f4 e5 9.e4 Nd7 10.fxe5 dxe5 11.Ne2 f6 12.Nf1 Nd6 13.Nc3 Rf7 14.Ne3 Nf8 15.O-O Ne6 16.Qg4 Nd4 17.Ncd5 h5 18.Qh4 a5 19.Bc3 Bc6 20.Kh1 Rh7 21.Bd3 Be8 22.Rd1 Kh8 23.Rc1 Nc6 24.h3 Rb7 25.Rcd1 Nd4 26.Bd2 Ra7 27. Nc2 Ne6 28.Qe1 Bc6 29.a3 Ra8 30.Nce3 Bb7 31.b4 Ra7 32.bxa5 Nd4 33.Rb1 Ba8 34.a4 Qe8 35.Rb2 Rd7 36.Qb1 Ba7 37.Nc3 Rd8 38.Nb5 Qd7 39.a6 Nc8 40.Nd5 Ne6 41.a5 Qf7 42.Be3 Kg8 43.Nbc3 Qf8 44.Rb6 Rd6 45.Rxd6 Qxd6 46.Qe1 Kh8 47.Qf2 Bc6 48.Qf5 Bd7 49.Nb5 Qf8 50.Rb1 Nd4 51.Qxd7 g5 1-0

Black's h-Rook never played a role in the game, which was perhaps the main reason for White's win.