Bogerd vs Gosling: Passed Pawns and an Active King
A club game supplied by Marcus Gosling becomes an interactive lesson about queen activity, a timely passed-pawn break, and a balanced bishop ending.
Reproducible analysis run complete
6 saved models on NVIDIA GeForce RTX 5080 · Stockfish 18 · 250,000 nodes per checked line.
Why this game?
The pieces around the king become the trap
Marcus Gosling supplied this complete game score and gave permission to use it. The event and playing date were not included, so this review makes no claim about them. Epsom Chess Club is credited through its junior programme page. The positions show why active pieces and passed-pawn timing matter even when the game finishes peacefully.
- White
- Ethan Bogerd
- Black
- Marcus Gosling
- Event
- Epsom Chess Club contribution
- Result
- 1/2-1/2
Interactive comparison
Choice one: keep the queen active
Black to move on move 25. All six neural evaluators choose 25...Bxc5, while Stockfish prefers the active retreat 25...Qc6. The game continued with 25...Qc2.
3n2k1/1p2b1pp/1P2ppb1/p1NpP3/q2P1P2/P3B3/4B1PP/2Q3K1 b - - 2 25Model candidates
Choose a viewpoint
V6 Active: Bxc5
Measured GPU output. The capture looks attractive, but the fixed-node engine gate places it 147 centipawns behind the best move.
Model top three: Bxc5 · fxe5 · Qxa3
Stockfish 18 safety gate
CompleteGate stopped Bxc5: 147 cp behind 25...Qc6. The -0.22 score is from Black's viewpoint, so White keeps a small edge in this snapshot.
- Engine move
- Qc6
- Depth
- 17
- Node budget
- 250,378
- Evaluation
- -0.22
Verification line
25...Qc6 26.Qb2 fxe5 27.fxe5 Bxc5 28.dxc5 Qa4 29.Bb5 Qd1+ 30.Kf2 Qh5
Position lesson
The queen must stay active without inviting a simple exchange. After 25...Qc2, White can trade queens with 26.Qxc2 and keep the advanced b6-pawn. The retreat to c6 watches the queenside while preserving counterplay.
Interactive comparison
Choice two: push before the blockade
White to move on move 31. Every model chooses 31.exf6+, a safe forcing move. Stockfish finds the stronger passed-pawn break 31.c6, while the game used 31.Bd2.
Model candidates
Choose a viewpoint
V6 Active: exf6+
Measured GPU output. The engine gate accepts the check within 100 cp, although the immediate c-pawn break is more precise.
Model top three: exf6+ · Bc4 · a4
Stockfish 18 safety gate
CompleteThe model choice 31.exf6+ passes the gate at 86 cp behind best. Stockfish gives 31.c6 about +3.10 for White in the fixed-node run.
- Engine move
- c6
- Depth
- 18
- Node budget
- 250,298
- Evaluation
- +3.10
Verification line
31.c6 bxc6 32.Ba6 Kd7 33.b7 Nxb7 34.Bxb7 Bc4 35.Bb6 a4 36.Kf2
Position lesson
The c-pawn can open a route for the protected b6-pawn before Black builds a blockade. After 31.c6 bxc6, Ba6 and b7 force Black to give material back to stop promotion.
Interactive comparison
Choice three: several roads to the draw
Black to move on move 49. All six models choose 49...gxh5. Stockfish slightly prefers 49...g5, and the played 49...Ke4 also holds the balance.
Model candidates
Choose a viewpoint
V6 Active: gxh5
Measured GPU output. Stockfish scores the simplifying capture as equal and keeps it safely inside the 100 cp gate.
Model top three: gxh5 · g5 · Bf1
Stockfish 18 safety gate
CompleteThe engine prefers 49...g5 by only 43 cp. Both the model choice 49...gxh5 and the played 49...Ke4 evaluate as equal in their verified lines.
- Engine move
- g5
- Depth
- 18
- Node budget
- 250,212
- Evaluation
- +0.43
Verification line
49...g5
Position lesson
A small engine difference is not always a practical error. Black can fix the kingside with g5, simplify with gxh5, or activate the king. The supplied game used the king route and ended in a draw.
Published finish
The forcing line
This is the finish recorded in the supplied game score. The draw was agreed after 52.Bh2, with both bishops and several pawns still on the board.
Review method
Source, models, then verification
Start with a permitted club game
Marcus Gosling supplied the full move score and allowed Spider Chess to use it.
Run six saved models on GPU
Every legal move is ranked by each scalar evaluator with the same 30% hybrid scoring formula.
Apply the Stockfish gate
Every distinct model choice and the played move receive an independent fixed-node verification.
What this review can and cannot show
Clear boundaries keep a useful experiment from becoming a bigger claim than the evidence supports.
- The event and playing date were not supplied, so the 2026 label refers to this Epsom Chess Club contribution and review publication.
- The supplied score records a draw by agreement after 52.Bh2; it does not claim that the final board position was forced to be a draw.
- These checkpoints are lightweight scalar position evaluators, not move-policy, search, or explanation models.
- Their inputs encode piece placement only; they do not receive move history, clocks, repetition state, castling rights, or en-passant state.
- The profile labels describe intended training behavior, not proven human-like personalities or internal reasoning.
- Stockfish used a fixed 250,000-node budget per line with two threads; this is a reproducible snapshot, not a claim of perfect play.
- Two-thread engine search can produce small score, depth, node-count, and principal-variation differences between reruns.
- This review measures three positions from one game. It does not establish Elo, general playing strength, or unseen-data accuracy.
- The prose explains observable chess consequences, not hidden neural-model reasoning.
Continue learning
Put the position on your own board
Start from the exact FEN and test whether you can find and convert the idea against a human-style bot.