Burrows vs Turner: A Rook Sacrifice That Would Not End
Three critical choices from a fierce final-round draw: create practical danger, stop a passed pawn, and step out of a mating net.
Reproducible analysis run complete
6 saved models on NVIDIA GeForce RTX 5080 · Stockfish 18 · 250,000 nodes per checked line.
Why this game?
A public thank-you to Wigston Chess Club
Martin Burrows welcomed Spider Chess using this published game, and Lewis Turner offered another rook-sacrifice example too. This review thanks Wigston Chess Club by naming both players, linking the exact club page, and turning their dramatic Hastings draw into an interactive lesson about practical danger and king safety.
- White
- Martin Burrows
- Black
- Lewis Turner
- Event
- Hastings Masters, round 9
- Result
- 1/2-1/2
Interactive comparison
Choice one: create danger, not just damage
Black to move after 20.f4. Lewis Turner played 20...Nxa3, giving up a knight to tear open White's king, while every model chose an immediate queen capture.
r5k1/5p1p/2p3pb/p2pPb2/1qnQ1P2/PNN5/1PP3PP/2KR3R b - - 0 20Model candidates
Choose a viewpoint
V6 Active: Qxc3
Measured GPU choice. Trading queens removes Black's attack and leaves White with a large advantage.
Model top three: Qxc3 · Qxb3 · Nxb2
Stockfish 18 safety gate
CompleteStockfish gives Black about +1.09 after 20...Qb8. The played 20...Nxa3 is 308 cp behind that snapshot, but it creates the dangerous practical battle seen in the game.
- Engine move
- Qb8
- Depth
- 15
- Node budget
- 250,137
- Evaluation
- +1.09
Verification line
20...Qb8 21.a4 Nxe5 22.Rhf1 Nc4 23.Rde1 Qb7 24.h3 Rb8 25.Qf2 Bf8
Position lesson
The practical sacrifice creates checks and forces White to solve difficult problems, but it is not the engine's safest choice. Stockfish prefers the patient retreat 20...Qb8. Both model captures lose the initiative even faster.
Interactive comparison
Choice two: stop the passed pawn
White to move after 27...d3. Black's pawn is one step from c2 and the queen is already beside White's king. The natural captures chosen by the models do not solve that threat.
Model candidates
Choose a viewpoint
V6 Active: Qxc5
Measured GPU choice. The pawn capture looks useful, but Bxf4+ and dxc2 keep White under heavy pressure.
Model top three: Qxc5 · Nxh7 · Ng4
Stockfish 18 safety gate
CompleteStockfish prefers 28.Rb6 and keeps Black's edge to about 0.93. 28.Qxc5 falls 420 cp behind the best line, while 28.Nxh7 falls 638 cp behind.
- Engine move
- Rb6
- Depth
- 14
- Node budget
- 250,401
- Evaluation
- -0.93
Verification line
28.Rb6 dxc2
Position lesson
28.Rb6 is a defensive move with several jobs: it controls the sixth rank, keeps the rook active, and prepares to meet the advancing pawn. Taking on c5 or h7 gives Black time to promote with checks.
Interactive comparison
Choice three: step out of the mating net
White to move after 31...Rd1+. All six models capture the checking rook, but the c-pawn then promotes with check and Black forces mate. Martin Burrows found 32.Kf2 over the board.
Model candidates
Choose a viewpoint
V6 Active: Rxd1
Measured GPU choice. After cxd1=Q+, Black forces mate in four.
Model top three: Rxd1 · Ke2 · Kf2
Stockfish 18 safety gate
CompleteStockfish keeps 32.Kf2 near equality at about 0.47 for Black. The unanimous model choice 32.Rxd1 allows cxd1=Q+ and mate in four.
- Engine move
- Kf2
- Depth
- 18
- Node budget
- 250,337
- Evaluation
- -0.47
Verification line
32.Kf2 Qxc1 33.Qxc1 Rxc1 34.Rxc1 Bxf4 35.Rxc2 Bxe5 36.Ne8+ Kf8 37.Ra2 Kxe8 38.Rxa5
Position lesson
When several pieces attack the king, the obvious capture can open a new line. Moving to f2 keeps the queen on e3, the rook on c1, and enough defenders around the promotion square to survive.
Published finish
The forcing line
The rook sacrifice begins a long king hunt. White later misses the strongest defence on move 28, then finds 32.Kf2 to avoid a forced mate and keep the game alive.
Review method
Source, models, then verification
Start with the Wigston game
Every position comes from the complete Burrows versus Turner score linked by Wigston Chess Club.
Measure six saved evaluators
Each saved checkpoint ranks every legal move on the NVIDIA GPU with the same 30% hybrid positional weight.
Apply the Stockfish safety gate
Stockfish checks the engine best move, every distinct model choice, and the move played in the game at a fixed node budget.
What this review can and cannot show
Clear boundaries keep a useful experiment from becoming a bigger claim than the evidence supports.
- This review uses original teaching prose. It does not reproduce the club member's annotations.
- The six neural checkpoints are scalar evaluators, not move-policy, search, or explanation models.
- Their inputs encode piece placement only; move history, repetition, clocks, castling rights, and en-passant state are not model inputs.
- Profile labels describe intended training behavior, not proven human-like personalities or internal reasoning.
- Stockfish used 250,000 nodes per checked line, two threads, and 256 MB hash. This is a reproducible snapshot, not a claim of perfect play.
- Two-thread search can produce small score, depth, node-count, and principal-variation differences on another run.
- The prose describes visible board features and measured move consequences. It does not claim to reveal a model's internal 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.